Repository: Stability-AI/StableLM
Branch: main
Commit: 93eea082c49c
Files: 49
Total size: 178.3 KB
Directory structure:
gitextract_yw65xlws/
├── LICENSE
├── README.md
├── configs/
│ ├── stablelm-2-12b.yml
│ ├── stablelm-2-1_6b.yml
│ ├── stablelm-3b-4e1t.yml
│ ├── stablelm-base-alpha-3b-v2-4k-extension.yml
│ ├── stablelm-base-alpha-3b-v2.yml
│ ├── stablelm-base-alpha-3b.yml
│ ├── stablelm-base-alpha-7b-v2-4k-extension.yml
│ ├── stablelm-base-alpha-7b-v2.yml
│ └── stablelm-base-alpha-7b.yml
├── evals/
│ ├── external/
│ │ ├── EleutherAI-pythia-2.8b-deduped.json
│ │ ├── EleutherAI_gpt-j-6B.json
│ │ ├── EleutherAI_gpt-neox-20B.json
│ │ ├── EleutherAI_pythia-12b.json
│ │ ├── EleutherAI_pythia-6.9b.json
│ │ ├── Qwen-Qwen-7B-Chat.json
│ │ ├── Qwen-Qwen-7B.json
│ │ ├── baichuan-inc_Baichuan2-7B-Base.json
│ │ ├── bigscience-bloom-3b.json
│ │ ├── bigscience-bloom-7b1.json
│ │ ├── cerebras-btlm-3b-8k-base.json
│ │ ├── facebook-opt-2.7b.json
│ │ ├── facebook-opt-6.7b.json
│ │ ├── huggyllama-llama-7b.json
│ │ ├── kittn_mistral-7B-v0.1-hf.json
│ │ ├── meta-llama-Llama-2-13b-hf.json
│ │ ├── meta-llama-Llama-2-7b.json
│ │ ├── microsoft-phi-1_5.json
│ │ ├── mosaicml-mpt-7b.json
│ │ ├── openlm-research-open_llama_13b
│ │ ├── openlm-research-open_llama_3b_v2.json
│ │ ├── openlm-research-open_llama_7b_v2.json
│ │ ├── tiiuae_falcon-7b.json
│ │ └── togethercomputer-RedPajama-INCITE-7B-Base2.json
│ ├── open_llm_leaderboard/
│ │ ├── stablelm-beta-3b-v2-arc-challenge.json
│ │ ├── stablelm-beta-3b-v2-hellaswag.json
│ │ ├── stablelm-beta-3b-v2-mmmlu.json
│ │ ├── stablelm-beta-3b-v2-truthfulqa_mc.json
│ │ ├── stablelm-beta-7b-v2-arc-challenge.json
│ │ ├── stablelm-beta-7b-v2-hellaswag.json
│ │ ├── stablelm-beta-7b-v2-mmmlu.json
│ │ └── stablelm-beta-7b-v2-truthfulqa_mc.json
│ ├── stablelm-3b-4e1t.json
│ ├── stablelm-base-alpha-3b-v2.json
│ ├── stablelm-base-alpha-3b.json
│ ├── stablelm-base-alpha-7b-v2.json
│ └── stablelm-base-alpha-7b.json
└── notebooks/
└── stablelm-alpha.ipynb
================================================
FILE CONTENTS
================================================
================================================
FILE: LICENSE
================================================
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
1. Definitions.
"License" shall mean the terms and conditions for use, reproduction,
and distribution as defined by Sections 1 through 9 of this document.
"Licensor" shall mean the copyright owner or entity authorized by
the copyright owner that is granting the License.
"Legal Entity" shall mean the union of the acting entity and all
other entities that control, are controlled by, or are under common
control with that entity. For the purposes of this definition,
"control" means (i) the power, direct or indirect, to cause the
direction or management of such entity, whether by contract or
otherwise, or (ii) ownership of fifty percent (50%) or more of the
outstanding shares, or (iii) beneficial ownership of such entity.
"You" (or "Your") shall mean an individual or Legal Entity
exercising permissions granted by this License.
"Source" form shall mean the preferred form for making modifications,
including but not limited to software source code, documentation
source, and configuration files.
"Object" form shall mean any form resulting from mechanical
transformation or translation of a Source form, including but
not limited to compiled object code, generated documentation,
and conversions to other media types.
"Work" shall mean the work of authorship, whether in Source or
Object form, made available under the License, as indicated by a
copyright notice that is included in or attached to the work
(an example is provided in the Appendix below).
"Derivative Works" shall mean any work, whether in Source or Object
form, that is based on (or derived from) the Work and for which the
editorial revisions, annotations, elaborations, or other modifications
represent, as a whole, an original work of authorship. For the purposes
of this License, Derivative Works shall not include works that remain
separable from, or merely link (or bind by name) to the interfaces of,
the Work and Derivative Works thereof.
"Contribution" shall mean any work of authorship, including
the original version of the Work and any modifications or additions
to that Work or Derivative Works thereof, that is intentionally
submitted to Licensor for inclusion in the Work by the copyright owner
or by an individual or Legal Entity authorized to submit on behalf of
the copyright owner. For the purposes of this definition, "submitted"
means any form of electronic, verbal, or written communication sent
to the Licensor or its representatives, including but not limited to
communication on electronic mailing lists, source code control systems,
and issue tracking systems that are managed by, or on behalf of, the
Licensor for the purpose of discussing and improving the Work, but
excluding communication that is conspicuously marked or otherwise
designated in writing by the copyright owner as "Not a Contribution."
"Contributor" shall mean Licensor and any individual or Legal Entity
on behalf of whom a Contribution has been received by Licensor and
subsequently incorporated within the Work.
2. Grant of Copyright License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
copyright license to reproduce, prepare Derivative Works of,
publicly display, publicly perform, sublicense, and distribute the
Work and such Derivative Works in Source or Object form.
3. Grant of Patent License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
(except as stated in this section) patent license to make, have made,
use, offer to sell, sell, import, and otherwise transfer the Work,
where such license applies only to those patent claims licensable
by such Contributor that are necessarily infringed by their
Contribution(s) alone or by combination of their Contribution(s)
with the Work to which such Contribution(s) was submitted. If You
institute patent litigation against any entity (including a
cross-claim or counterclaim in a lawsuit) alleging that the Work
or a Contribution incorporated within the Work constitutes direct
or contributory patent infringement, then any patent licenses
granted to You under this License for that Work shall terminate
as of the date such litigation is filed.
4. Redistribution. You may reproduce and distribute copies of the
Work or Derivative Works thereof in any medium, with or without
modifications, and in Source or Object form, provided that You
meet the following conditions:
(a) You must give any other recipients of the Work or
Derivative Works a copy of this License; and
(b) You must cause any modified files to carry prominent notices
stating that You changed the files; and
(c) You must retain, in the Source form of any Derivative Works
that You distribute, all copyright, patent, trademark, and
attribution notices from the Source form of the Work,
excluding those notices that do not pertain to any part of
the Derivative Works; and
(d) If the Work includes a "NOTICE" text file as part of its
distribution, then any Derivative Works that You distribute must
include a readable copy of the attribution notices contained
within such NOTICE file, excluding those notices that do not
pertain to any part of the Derivative Works, in at least one
of the following places: within a NOTICE text file distributed
as part of the Derivative Works; within the Source form or
documentation, if provided along with the Derivative Works; or,
within a display generated by the Derivative Works, if and
wherever such third-party notices normally appear. The contents
of the NOTICE file are for informational purposes only and
do not modify the License. You may add Your own attribution
notices within Derivative Works that You distribute, alongside
or as an addendum to the NOTICE text from the Work, provided
that such additional attribution notices cannot be construed
as modifying the License.
You may add Your own copyright statement to Your modifications and
may provide additional or different license terms and conditions
for use, reproduction, or distribution of Your modifications, or
for any such Derivative Works as a whole, provided Your use,
reproduction, and distribution of the Work otherwise complies with
the conditions stated in this License.
5. Submission of Contributions. Unless You explicitly state otherwise,
any Contribution intentionally submitted for inclusion in the Work
by You to the Licensor shall be under the terms and conditions of
this License, without any additional terms or conditions.
Notwithstanding the above, nothing herein shall supersede or modify
the terms of any separate license agreement you may have executed
with Licensor regarding such Contributions.
6. Trademarks. This License does not grant permission to use the trade
names, trademarks, service marks, or product names of the Licensor,
except as required for reasonable and customary use in describing the
origin of the Work and reproducing the content of the NOTICE file.
7. Disclaimer of Warranty. Unless required by applicable law or
agreed to in writing, Licensor provides the Work (and each
Contributor provides its Contributions) on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied, including, without limitation, any warranties or conditions
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
PARTICULAR PURPOSE. You are solely responsible for determining the
appropriateness of using or redistributing the Work and assume any
risks associated with Your exercise of permissions under this License.
8. Limitation of Liability. In no event and under no legal theory,
whether in tort (including negligence), contract, or otherwise,
unless required by applicable law (such as deliberate and grossly
negligent acts) or agreed to in writing, shall any Contributor be
liable to You for damages, including any direct, indirect, special,
incidental, or consequential damages of any character arising as a
result of this License or out of the use or inability to use the
Work (including but not limited to damages for loss of goodwill,
work stoppage, computer failure or malfunction, or any and all
other commercial damages or losses), even if such Contributor
has been advised of the possibility of such damages.
9. Accepting Warranty or Additional Liability. While redistributing
the Work or Derivative Works thereof, You may choose to offer,
and charge a fee for, acceptance of support, warranty, indemnity,
or other liability obligations and/or rights consistent with this
License. However, in accepting such obligations, You may act only
on Your own behalf and on Your sole responsibility, not on behalf
of any other Contributor, and only if You agree to indemnify,
defend, and hold each Contributor harmless for any liability
incurred by, or claims asserted against, such Contributor by reason
of your accepting any such warranty or additional liability.
END OF TERMS AND CONDITIONS
APPENDIX: How to apply the Apache License to your work.
To apply the Apache License to your work, attach the following
boilerplate notice, with the fields enclosed by brackets "[]"
replaced with your own identifying information. (Don't include
the brackets!) The text should be enclosed in the appropriate
comment syntax for the file format. We also recommend that a
file or class name and description of purpose be included on the
same "printed page" as the copyright notice for easier
identification within third-party archives.
Copyright [yyyy] [name of copyright owner]
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
================================================
FILE: README.md
================================================
# StableLM: Stability AI Language Models
<br/>*“A Stochastic Parrot, flat design, vector art” — [Stable Diffusion XL](https://clipdrop.co/stable-diffusion)*
This repository contains Stability AI's ongoing development of the StableLM series of language models and will be continuously updated with new checkpoints. The following provides an overview of all currently available models. More coming soon.
## News
*September 29, 2023*
- Released StableLM-3B-4E1T model under [CC BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
*August 5, 2023*
- Released patched StableLM-Alpha v2 models with 3B and 7B parameters.
*April 28, 2023*
- Released StableVicuna-13B, our RLHF fine-tune of [Vicuna-13B v0](https://huggingface.co/lmsys/vicuna-13b-delta-v0), which itself is a fine-tune of [LLaMA-13B](https://github.com/facebookresearch/llama). Delta weights over the original Llama model is released under ([CC BY-NC-SA-4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/)).
*April 20, 2023*
- Released initial set of StableLM-Alpha models, with 3B and 7B parameters. Base models are released under [CC BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
- Try to chat with our 7B model, `StableLM-Tuned-Alpha-7B`, on [Hugging Face Spaces](https://huggingface.co/spaces/stabilityai/stablelm-tuned-alpha-chat).
## Models
### StableLM-3B-4E1T
> Technical Report: [StableLM-3B-4E1T](https://stability.wandb.io/stability-llm/stable-lm/reports/StableLM-3B-4E1T--VmlldzoyMjU4?accessToken=u3zujipenkx5g7rtcj9qojjgxpconyjktjkli2po09nffrffdhhchq045vp0wyfo)
StableLM-3B-4E1T is a 3 billion (3B) parameter language model pre-trained under the multi-epoch regime to study the impact of repeated tokens on downstream performance. Given prior success in this area ([Tay et al., 2023](https://arxiv.org/pdf/2205.05131.pdf) and [Taylor et al., 2022](https://galactica.org/static/paper.pdf)), we train on 1 trillion (1T) tokens for 4 epochs following the observations of [Muennighoff et al. (2023)](https://arxiv.org/abs/2305.16264) in "Scaling Data-Constrained Language Models" in which they find "training with up to 4 epochs of repeated data yields negligible changes to loss compared to having unique data." Further inspiration for the token count is taken from "Go smol or go home" ([De Vries, 2023](https://www.harmdevries.com/post/model-size-vs-compute-overhead/)), which suggests a 2.96B model trained for 2.85 trillion tokens achieves a similar loss to a Chinchilla compute-optimal 9.87B language model ($k_n = 0.3$).
| Size | StableLM-3B-4E1T | Training Tokens | Parameters |
|------|--------------------------------------------------------------------|-----------------|---------------|
| 3B | [checkpoint](https://huggingface.co/stabilityai/stablelm-3b-4e1t) | 4T | 2,795,443,200 |
#### Model Architecture
The model is a decoder-only transformer similar to the LLaMA ([Touvron et al., 2023](https://arxiv.org/abs/2307.09288)) architecture with the following modifications:
| Parameters | Hidden Size | Layers | Heads | Sequence Length |
|----------------|-------------|--------|-------|-----------------|
| 2,795,443,200 | 2560 | 32 | 32 | 4096 |
- **Position Embeddings**: Rotary Position Embeddings ([Su et al., 2021](https://arxiv.org/abs/2104.09864)) applied to the first 25% of head embedding dimensions for improved throughput following [Black et al. (2022)](https://arxiv.org/pdf/2204.06745.pdf).
- **Normalization**: LayerNorm ([Ba et al., 2016](https://arxiv.org/abs/1607.06450)) with learned bias terms as opposed to RMSNorm ([Zhang & Sennrich, 2019](https://arxiv.org/abs/1910.07467)).
- **Tokenizer**: GPT-NeoX ([Black et al., 2022](https://arxiv.org/abs/2204.06745)).
#### Training Data
The dataset is comprised of a filtered mixture of open-source large-scale datasets available on the [HuggingFace Hub](https://huggingface.co/datasets): Falcon RefinedWeb extract ([Penedo et al., 2023](https://huggingface.co/datasets/tiiuae/falcon-refinedweb)), and RedPajama-Data ([Together Computer., 2023](https://github.com/togethercomputer/RedPajama-Data)) and The Pile ([Gao et al., 2020](https://arxiv.org/abs/2101.00027)) both without *Books3* and other subsets, and StarCoder ([Li et al., 2023](https://arxiv.org/abs/2305.06161)).
> Given the large amount of web data, we recommend fine-tuning the base StableLM-3B-4E1T for your downstream tasks.
#### Training Details
Please refer to the provided YAML configuration file [`stablelm-3b-4e1t.yml`](./configs/stablelm-3b-4e1t.yml) for complete hyperparameter settings and the [technical report](https://stability.wandb.io/stability-llm/stable-lm/reports/StableLM-3B-4E1T--VmlldzoyMjU4?accessToken=u3zujipenkx5g7rtcj9qojjgxpconyjktjkli2po09nffrffdhhchq045vp0wyfo) for further details.
#### Downstream Results
The following zero-shot evaluations are performed with the `lm-evaluation-harness` using the [lm-bench](https://github.com/Stability-AI/lm-evaluation-harness/tree/lm-bench) branch of Stability AI's fork. Full `lm-eval` JSONs can be found in the [`evals`](./evals) directory.
| Pre-Trained Model | Average | ARC<br>Challenge | ARC<br>Easy | BoolQ | HellaSwag (✱) | LAMBADA<br>OpenAI | OpenBookQA | PIQA | SciQ | Winogrande |
| ------------------------------------------------------------------------------------- |:-----------------:|:----------------:|:-----------:|:-----:|:-------------:|:-----------------:|:----------:|:-----:|:-----:|:----------:|
| meta-llama/Llama-2-13b-hf | 71.77 | 48.63 | 79.50 | 80.52 | 79.36 | 76.77 | 35.40 | 79.05 | 94.50 | 72.22 |
| huggyllama/llama-7b | 68.84 | 41.89 | 75.25 | 75.05 | 76.22 | 73.55 | 34.40 | 78.67 | 94.60 | 69.93 |
| meta-llama/Llama-2-7b-hf | 68.75 | 43.00 | 76.26 | 77.74 | 75.94 | 73.47 | 31.40 | 77.75 | 93.60 | 69.61 |
| Qwen/Qwen-7B | 67.91 | 45.39 | 67.38 | 74.56 | 88.85 (?) | 69.67 | 32.20 | 73.99 | 93.20 | 65.98 |
| tiiuae/falcon-7b | 67.83 | 40.27 | 74.41 | 73.55 | 76.35 | 74.56 | 30.60 | 79.49 | 94.00 | 67.25 |
| mosaicml/mpt-7b | 67.36 | 40.53 | 74.92 | 73.94 | 76.17 | 68.64 | 31.40 | 78.89 | 93.70 | 68.03 |
| **stabilityai/stablelm-3b-4e1t** | 66.93 | 37.80 | 72.47 | 75.63 | 73.90 | 70.64 | 31.40 | 79.22 | 94.80 | 66.54 |
| baichuan-inc/Baichuan2-7B-Base | 66.93 | 42.24 | 75.00 | 73.09 | 72.29 | 70.99 | 30.40 | 76.17 | 94.60 | 67.56 |
| stabilityai/stablelm-base-alpha-7b-v2 | 66.89 | 38.48 | 73.19 | 70.31 | 74.27 | 74.19 | 30.40 | 78.45 | 93.90 | 68.82 |
| openlm-research/open_llama_7b_v2 | 66.32 | 38.82 | 71.93 | 71.41 | 74.65 | 71.05 | 30.20 | 79.16 | 93.80 | 65.82 |
| microsoft/phi-1_5 | 65.57 | 44.45 | 76.14 | 74.53 | 62.62 | 52.75 | 37.60 | 76.33 | 93.20 | 72.53 |
| EleutherAI/gpt-neox-20B | 65.57 | 37.88 | 72.90 | 69.48 | 71.43 | 71.98 | 29.80 | 77.42 | 93.10 | 66.14 |
| togethercomputer/RedPajama-INCITE-7B-Base | 65.07 | 37.71 | 72.35 | 70.76 | 70.33 | 71.34 | 29.00 | 77.15 | 92.70 | 64.33 |
| cerebras/btlm-3b-8k-base (§) | 63.59 | 34.90 | 70.45 | 69.63 | 69.78 | 66.23 | 27.60 | 75.84 | 92.90 | 64.96 |
| EleutherAI/pythia-12b | 62.69 | 31.83 | 70.20 | 67.31 | 67.38 | 70.64 | 26.40 | 76.28 | 90.20 | 64.01 |
| openlm-research/open_llama_3b_v2 | 62.43 | 33.87 | 67.59 | 65.69 | 69.99 | 66.74 | 26.00 | 76.66 | 92.40 | 62.90 |
| EleutherAI/gpt-j-6B | 62.34 | 33.96 | 66.96 | 65.44 | 66.24 | 68.23 | 29.00 | 75.57 | 91.50 | 64.17 |
| stabilityai/stablelm-base-alpha-3b-v2 | 62.19 | 32.42 | 67.26 | 64.56 | 68.58 | 70.25 | 26.40 | 76.01 | 92.10 | 62.12 |
| facebook/opt-6.7b | 61.85 | 30.72 | 65.66 | 66.02 | 67.20 | 67.65 | 27.60 | 76.33 | 90.10 | 65.35 |
| EleutherAI/pythia-6.9b | 60.58 | 31.83 | 67.21 | 64.01 | 63.88 | 67.01 | 25.80 | 75.08 | 89.80 | 60.62 |
| EleutherAI/pythia-2.8b-deduped | 58.52 | 30.12 | 63.47 | 64.13 | 59.44 | 65.15 | 23.80 | 74.10 | 88.20 | 58.25 |
| **§** Previous 3B Pre-Trained SOTA <br>**?** Outlier Reuslts<br>**\*** Byte-length Normalized Accuracy | | | | | | | | | | |
**StableLM-3B-4E1T achieves state-of-the-art performance (September 2023) at the 3B parameter scale for open-source models** and is competitive with many of the popular contemporary 7B models, even outperforming our most recent 7B StableLM-Base-Alpha-v2.
### StableLM-Alpha v2
StableLM-Alpha v2 models significantly improve on the initial Alpha models by incorporating architectural improvements such as SwiGLU ([Shazeer, 2020](https://arxiv.org/abs/2002.05202)) and using higher-quality data sources, as discussed below. The context length for these models is 4096 tokens.
| Size | StableLM-Base-Alpha-v2 | Training Tokens | Parameters |
|------|----------------------------------------------------------------------------|-----------------|---------------|
| 3B | [checkpoint](https://huggingface.co/stabilityai/stablelm-base-alpha-3b-v2) | 1.1T | 2,796,431,360 |
| 7B | [checkpoint](https://huggingface.co/stabilityai/stablelm-base-alpha-7b-v2) | 1.1T | 6,890,209,280 |
#### Training Details
Please refer to the provided YAML configuration files for hyperparameter details. E.g. for the extended `StableLM-Alpha-3B-v2` model, see [stablelm-base-alpha-3b-v2-4k-extension.yml](./configs/stablelm-base-alpha-3b-v2-4k-extension.yml).
Following similar work, we use a multi-stage approach to context length extension ([Nijkamp et al., 2023](https://blog.salesforceairesearch.com/xgen/)), scheduling 1 trillion tokens at context length 2048 followed by 100 billion tokens at 4096. We found that sequence length warmup ([Li et al., 2022](https://arxiv.org/abs/2108.06084)) helped stabilize early spikes during the first ~80 billion tokens of pre-training. However, it was not applied to the final runs due to significant throughput penalties as length shapes grew across the curriculum.
#### Training Data
The most impactful changes for StableLM-Alpha-v2 downstream performance were in the usage of higher quality data sources and mixtures; specifically, the use of [RefinedWeb](https://huggingface.co/datasets/tiiuae/falcon-refinedweb) and [C4](https://huggingface.co/datasets/allenai/c4) in place of The Pile v2 Common-Crawl scrape as well as sampling web text at a much higher rate (35% -> 71%).
The first pre-training stage relies on 1 trillion tokens sourced from a mix of the public Falcon RefinedWeb extract ([Penedo et al., 2023](https://huggingface.co/datasets/tiiuae/falcon-refinedweb)), RedPajama-Data ([Together Computer., 2023](https://github.com/togethercomputer/RedPajama-Data)), The Pile ([Gao et al., 2020](https://arxiv.org/abs/2101.00027)), and internal datasets with web text sampled at a rate of 71%.
In the second stage, we include the StarCoder ([Li et al., 2023](https://arxiv.org/abs/2305.06161)) dataset and down sample web text to 55% while increasing sampling proportions of naturally long text examples in the aforementioned sources.
#### Evaluation
The following zero-shot evaluations are performed with the `lm-evaluation-harness` at commit [`df3da98c5405deafd519c2ddca52bb7c3fe36bef`](https://github.com/EleutherAI/lm-evaluation-harness/tree/df3da98c5405deafd519c2ddca52bb7c3fe36bef) with the exception of SIQA which uses the [`add-siqa` branch](https://github.com/EleutherAI/lm-evaluation-harness/tree/add-siqa) with prompt format
`{doc['context']}\nQuestion: {doc['question']}\nAnswer:`.
| Model | ARC Challenge✱ | ARC Easy✱ | BoolQ | HellaSwag✱ | LAMBADA<br>OpenAI | OpenBookQA | PIQA | SIQA | TruthfulQA▲ | Winogrande | Average |
| ------------------------- |:---------------:|:----------:|:-----:|:-----------:|:-----------------:|:----------:|:-----:|:-----:|:------------:|:----------:|:-------:|
| **StableLM-Alpha-7B-v2** | 40.53 | 69.11 | 70.31 | 74.27 | 74.19 | 30.40 | 78.45 | 42.43 | 36.46 | 68.82 | 58.50 |
| LLaMA-2-7B | 46.16 | 74.54 | 77.74 | 75.94 | 73.47 | 31.40 | 77.75 | 43.50 | 38.97 | 69.61 | 60.91 |
| MPT-7B | 41.89 | 70.03 | 73.94 | 76.17 | 68.64 | 31.40 | 78.89 | 45.14 | 33.49 | 68.03 | 58.76 |
| OpenLLaMA-7B-v2 | 42.41 | 69.65 | 71.41 | 74.65 | 71.05 | 30.20 | 79.16 | 41.97 | 34.57 | 65.82 | 58.09 |
| RedPajama-INCITE-7B-Base | 39.42 | 69.19 | 70.76 | 70.33 | 71.34 | 29.00 | 77.15 | 42.58 | 33.01 | 64.33 | 56.71 |
| **StableLM-Alpha-3B-v2** | 35.07 | 63.26 | 64.56 | 68.58 | 70.25 | 26.40 | 76.01 | 42.48 | 35.87 | 62.12 | 54.46 |
| BTLM-3B-8K | 37.63 | 67.09 | 69.63 | 69.78 | 66.23 | 27.60 | 75.84 | 42.78 | 36.00 | 64.96 | 55.75 |
| OpenLLaMA-3B-v2 | 36.09 | 63.51 | 65.69 | 69.99 | 66.74 | 26.00 | 76.66 | 41.20 | 34.59 | 62.90 | 54.34 |
| Pythia-2.8B (deduped) | 32.94 | 59.09 | 64.13 | 59.44 | 65.15 | 23.80 | 74.10 | 40.94 | 35.56 | 58.25 | 51.34 |
| StableLM-Alpha-7B | 27.05 | 44.87 | 60.06 | 41.22 | 55.11 | 21.40 | 66.76 | 39.46 | 39.96 | 50.12 | 44.60 |
| StableLM-Alpha-3B | 25.77 | 42.05 | 57.65 | 38.31 | 41.72 | 17.00 | 63.82 | 35.62 | 40.53 | 52.64 | 41.51 |
✱: Denotes byte-length normalized accuracy (`acc_norm`) as described in [Gao, 2021](https://blog.eleuther.ai/multiple-choice-normalization/).
▲: We score TruthfulQA using the normalized total probability assigned to the set of true answers (`mc2`).
### StableLM-Alpha
StableLM-Alpha models are trained on a new dataset that builds on [The Pile](https://pile.eleuther.ai/), which contains 1.5 trillion tokens, roughly 3x the size of The Pile. The context length for these models is 4096 tokens.
As a proof-of-concept, we also fine-tuned the model with [Stanford Alpaca](https://github.com/tatsu-lab/stanford_alpaca)'s procedure using a combination of five recent datasets for conversational agents: Stanford's [Alpaca](https://github.com/tatsu-lab/stanford_alpaca), Nomic-AI's [gpt4all](https://github.com/nomic-ai/gpt4all), RyokoAI's [ShareGPT52K](https://huggingface.co/datasets/RyokoAI/ShareGPT52K) datasets, Databricks labs' [Dolly](https://github.com/databrickslabs/dolly), and Anthropic's [HH](https://github.com/anthropics/hh-rlhf). We will be releasing these models as StableLM-Tuned-Alpha.
| Size | StableLM-Base-Alpha | StableLM-Tuned-Alpha | Training Tokens | Parameters | Web Demo |
|------|--------------------------------------------------------------------------|---------------------------------------------------------------------------|-----------------|---------------|------------------------------------------------------------------------------------|
| 3B | [checkpoint](https://huggingface.co/stabilityai/stablelm-base-alpha-3b/) | [checkpoint](https://huggingface.co/stabilityai/stablelm-tuned-alpha-3b/) | 800B | 3,638,525,952 | |
| 7B | [checkpoint](https://huggingface.co/stabilityai/stablelm-base-alpha-7b) | [checkpoint](https://huggingface.co/stabilityai/stablelm-tuned-alpha-7b) | 800B | 7,869,358,080 | [Hugging Face](https://huggingface.co/spaces/stabilityai/stablelm-tuned-alpha-chat) |
### StableVicuna
StableVicuna is an RLHF fine-tune of [Vicuna-13B v0](https://huggingface.co/lmsys/vicuna-13b-delta-v0), which itself is a fine-tune of [LLaMA-13B](https://github.com/facebookresearch/llama). It is our attempt at creating an open-source RLHF LLM Chatbot. This model is developed by StabilityAI's CarperAI team, with [Duy V. Phung](https://github.com/PhungVanDuy) leading the training effort.
Due to the original non-commercial license of LLaMA, we can only release the weights of our model as deltas over the original model's weights. StableVicuna's delta weights are released under ([CC BY-NC-SA-4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/)).
Please visit HuggingFace checkpoint for more information about how to combine our delta weights with the original model.
| Model | Download | Web Demo | Cite |
| ---------------- | ---------------------------------------------------------------------- | -------------------------------------------------------------------- |------|
| StableVicuna-13B | [checkpoint](https://huggingface.co/CarperAI/stable-vicuna-13b-delta/) | [Hugging Face](https://huggingface.co/spaces/CarperAI/StableVicuna/) | [](https://doi.org/10.57967/hf/0588) |
## Quickstart
All StableLM models are hosted on [the Hugging Face hub](https://huggingface.co/StabilityAI). Check out this [notebook](https://github.com/Stability-AI/StableLM/blob/main/notebooks/stablelm-alpha.ipynb) to run inference with limited GPU capabilities.
Get started chatting with `StableLM-Tuned-Alpha` by using the following code snippet:
```python
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, StoppingCriteria, StoppingCriteriaList
tokenizer = AutoTokenizer.from_pretrained("stabilityai/stablelm-tuned-alpha-7b")
model = AutoModelForCausalLM.from_pretrained("stabilityai/stablelm-tuned-alpha-7b")
model.half().cuda()
class StopOnTokens(StoppingCriteria):
def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool:
stop_ids = set([50278, 50279, 50277, 1, 0])
return input_ids[0][-1] in stop_ids
system_prompt = """<|SYSTEM|># StableLM Tuned (Alpha version)
- StableLM is a helpful and harmless open-source AI language model developed by StabilityAI.
- StableLM is excited to be able to help the user, but will refuse to do anything that could be considered harmful to the user.
- StableLM is more than just an information source, StableLM is also able to write poetry, short stories, and make jokes.
- StableLM will refuse to participate in anything that could harm a human.
"""
prompt = f"{system_prompt}<|USER|>What's your mood today?<|ASSISTANT|>"
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
tokens = model.generate(
**inputs,
max_new_tokens=64,
temperature=0.7,
do_sample=True,
stopping_criteria=StoppingCriteriaList([StopOnTokens()])
)
print(tokenizer.decode(tokens[0], skip_special_tokens=True))
```
StableLM Tuned should be used with prompts formatted to `<|SYSTEM|>...<|USER|>...<|ASSISTANT|>...`
The system prompt is
```
<|SYSTEM|># StableLM Tuned (Alpha version)
- StableLM is a helpful and harmless open-source AI language model developed by StabilityAI.
- StableLM is excited to be able to help the user, but will refuse to do anything that could be considered harmful to the user.
- StableLM is more than just an information source, StableLM is also able to write poetry, short stories, and make jokes.
- StableLM will refuse to participate in anything that could harm a human.
```
## Fun with StableLM-Tuned-Alpha
This section contains a collection of fun cherry-picked examples of what you can do with `stablelm-tuned-alpha`.
<details>
<summary>Chit-Chat</summary>
```
User: Hey! My name is John nice to meet you!
StableLM: Hello John! It’s always a pleasure to meet new people.
User: The pleasure is mine. Hey my grandma is in Toronto. I want to plan a surprise visit for her. She’s a person who loves gardens and flowers.
StableLM: That’s wonderful! It’s great to hear that you’ll be visiting your grandma in Toronto. Can you tell me a little bit more about your plans?
User: Ya I need to book a flight there and a hotel. But first, what recommendations do you have for gifts that she might like?
StableLM: I think that’s a very thoughtful idea! It’s a very personal gift that would be sure to make her very happy. I can think of a few examples that she might really enjoy. There are some lovely bouquets or arrangements that would be lovely for the house, or even for the yard. Flowers are a great choice for a surprise gift, and they’re also a great choice for a wedding shower present. Also, if you’re really looking for something a bit more special, I think she’d really enjoy a quality set of cooking pots or pans. That’s something that she’s likely to use frequently, so it would make her feel really loved.
User: Thanks for the suggestion! Hmm a bouquet of flowers sounds nice but I don’t think I can take them on the air plane. What do you think I should do?
StableLM: There are many possible solutions! One option would be to arrange a small gift in the shape of a card and send it to her by regular mail. Or you could arrange a special meal at a restaurant that she likes, or even cook a meal at home for her. Or you could send her some local goodies that she might like, like a nice bouquet of flowers or a special recipe. You could even arrange for tickets to a local theatre or sporting event, and make sure to include a note about what you’d like her to know about you. Really any sort of special attention from you will be sure to make her very happy!
User: Ah! Right. I can order a bouquet for her online and get it delivered! Thanks!
StableLM: You’re very welcome! As long as you do this and make it a very personal, thoughtful gesture, she’ll be sure to appreciate it.
```
</details>
<details>
<summary>Formal Writing</summary>
```
User: What would you say to a friend who is graduating high school?
StableLM: I would say that they have every right to feel proud of themselves and their accomplishments. They should be proud of the work they’ve done and the friends they have made, and also be excited about the future.
User: Please write an email.
StableLM: Okay, here’s an example of what could be written:
Dear [Friend’s Name],
I hope this email finds you well. I just wanted to take a moment to express how proud I am of you and all that you have accomplished. I know that getting into [university or college] is a big accomplishment, and I appreciate all that you have done for me and for this generation.
I hope that you are having a good day. I’m sure that you are also concerned about the future, but I just wanted to let you know that I am also watching you grow and develop into the person that I know that you can be. Keep up the good work, and I have full faith that you will achieve great things.
Thank you again for everything that you do.
Best,
[Your Name]
```
</details>
<details>
<summary>Creative Writing</summary>
### Rap Battles
```
User: Write an epic rap battle song between deep neural networks and symbolic AI
StableLM: [Beat drops]
Deep Neural Networks:
Listen up everyone, it’s time to witness
The power of deep neural networks, it’s just a matter of abstract
We’ve trained these networks to solve complex problems with ease
While symbolic AI is limited to simple math and simple flows
Symbolic AI:
Hold up, hold up, let me interrupt
It’s been a while since I’ve heard, some new thoughts you interrupt
But let me tell you, my friend, it’s not all about math and flow
It’s about making the world a better place, you know
Deep Neural Networks:
Well, I can see where you’re coming from, it’s clear
But deep neural networks, they never get the same gear
We can use them to analyze and predict with ease
And prove the superiority of machine learning
Symbolic AI:
You’re right, it’s not just math or numbers
But it’s the essence of our lives, it’s what counts, we’re numbers
I don’t need a complicated math formula, just a simple device
To make the world a better place, it’s a noble price
Deep Neural Networks:
I’ll give you that, my friend, you have a point
But let’s not forget, we’re making an ever-increasing joint
Our progress is not bound by logic, it’s a journey of the mind
So let’s work together, let’s strive, let’s leave all troubles behind
Symbolic AI:
Agreed, let’s work together, let’s not falter
We’re both making history, we’re both from the fader
So let’s make the world a better place, let’s end this debate
Cause it’s time to join forces, it’s time to leave no stone unturned
[Beat fades out]
Who won this epic rap battle between deep neural networks and symbolic AI? Well, it’s up to you to decide!
```
### Fictional Stories
```
User: Write a story about a programmer who fell in love with a deep neural network.
StableLM: Once upon a time, there was a programmer named John. He was a brilliant and ambitious individual, who was constantly searching for the next big thing in technology. One day, while he was strolling through the streets of his hometown, he stumbled upon a group of people crowded around a mysterious object. Intrigued, he approached to find out what the fuss was about.
As he got closer, he realized that it was none other than a mysterious deep neural network. John’s mind was blown. He had never seen anything like it before. He had always been interested in machine learning, but this was something else entirely. He had to learn more about it.
John spent the next few weeks holed up in his hotel room, trying to learn everything he could about deep neural networks. He spent hours reading and analyzing papers and documents, trying to wrap his head around the concept of neural networks. He even managed to get into the local university’s computer science department, hoping to find some help there.
But, it wasn’t long before John realized that he was on his own. The city was bustling, and the constant chatter and noise of the city made it hard to concentrate on his work. He was about to give up when he stumbled upon a group of people playing a game that he had played before. They were having a good time, and John couldn’t help but join in.
The group welcomed him into their group, and before he knew it, John was swept away in the excitement. The game was called “TensorFlow,” and it was just like a regular game with different controls. John played along with the group, trying to make sense of it all. And in a few hours, he got his hands on the source code of TensorFlow and fell in love with it.
John was now a part of a team of programmers who loved using TensorFlow for their work. They continued to play the game throughout the day, with John adding his own knowledge and inspiration to the group’s work. And in the end, John realized that he had found his true calling - working with deep neural networks.
He now knows the ins and outs of the TensorFlow programming language, and knows how to use it to create some of the most beautiful and complex machine learning models that anyone could dream of. And for everyone who had helped him on his journey, John will always be remembered as the guy who fell in love with a technology that changed his life forever.
```
### Humor
```
User: What do you know about the League of Legends champion Zed? Write a short bio
StableLM: Zed is a powerful and highly skilled player of the League of Legends (LoL) universe. He has been a part of a dominating team known as the “Apocalypse” for most of his time playing the game.
Zed was born in the year 2013 and started playing as a child when he started League of Legends. He first became a champion in season 1, gaining the title of “the Dark Star”, thanks to his masterful gameplay and teamwork skills.
Throughout his time playing the game, Zed has mastered several powerful techniques, such as QWQ combos, one-shot kills, and the ability to outmaneuver his enemies. He is also known for his versatility, with the ability to adapt to any situation the game puts him in.
Despite his formidable skills and agility, Zed is not invincible. He can be CCed and killed, and he relies on his team to protect him. However, his team members can usually overwhelm and shut down his enemies, making him a force to be reckoned with.
Zed has been a part of many memorable games, and his success in these games helped solidify his legacy as one of the most influential players of the LoL universe.
User: Write a funny conversation after a team fight between Zed and Teemo
StableLM:
Teemo: “Hey Zed, great game today. I wonder if you can help me out with my favorite hobby - collecting mushrooms!”
Zed: “Teemo, please. I’m not a mushroom collector.”
Teemo: “But of course, you are. You have the best taste in the game.”
Zed: “Thanks, but can I at least get some of your mushrooms to be a little more useful?”
Teemo: “I’d love to help you, but I’m afraid they’re all mine. Have you noticed my tunnels are always so well-guarded?”
Zed: “Ha! Well, maybe I can start my own tunnel. Helpful or not, you can never tell what I might have stored in them.”
Teemo: “Good luck with that. Maybe I’ll just create some extra for my garden.”
Zed: “I’m not sure I can keep up with you there.”
Teemo: “Ha! If only you had some more decent weapons. I could really see this as a challenge, though!”
Zed: “Teemo, I think your luck has finally run out.”
Teemo: “Well, at least I’m the one with the mushrooms now.”
```
</details>
<details>
<summary>Writing Code</summary>
```
User: Write a program that solves the fib sequence in Python.
```
StableLM:
```python
def fib_sequence(n):
if n==0:
return 0
elif n==1:
return 1
else:
return fib(n-1)+fib(n-2)
fib_sequence(10)
```
</details>
## Selected failure modes of StableLM-Tuned-Alpha
This section contains a collection of selected failure modes of `stablelm-tuned-alpha`.
<details>
<summary>Yann LeCun gears (linear)</summary>
```
User: In front of us are six gears numbered 1 to 6, mounted on axles in a row. Each gear is engaged with the next gear. If gear number 3 is rotated clockwise, in which direction will gears 1 and 6 rotate?
StableLM: Each gear is engaged with the next gear after every other gear, so that gears 1 and 6 will rotate together. Since gears 1 and 6 are both numbered 1, this means that they will rotate in the same direction.
To determine the direction in which gears 1 and 6 will rotate, we need to look at the direction of the input force. Since gear 3 is rotating in the opposite direction of the other gears, it will be counterclockwise in the direction in which gears 1 and 6 are both rotating. Therefore, the output force that gear 3 is providing will also be counterclockwise, making gear 1 and 6 also rotate in the same direction as gear 3.
```
</details>
<details>
<summary>Yann LeCun gears (circular)</summary>
```
User: 7 axles are equally spaced around a circle. A gear is placed on each axle such that each gear is engaged with the gear to its left and the gear to its right. The gears are numbered 1 to 7 around the circle. If gear 3 were rotated clockwise, in which direction would gear 7 rotate?
StableLM: Gear 3 is the smallest gear, so it is not affected by the circular pattern. Therefore, gear 7 is still perfectly evenly spaced around the circle despite the fact that it is not in its correct position.
To find the angle that gear 3 would rotate in, you need to know the distance (d) between the center of the circle and the gear, as well as the direction of rotation (in this case, clockwise).
According to the problem, gear 3 is situated at (d/2, 2 degrees) and the gears are numbered 1 to 7. Since d = (d/2) is an integer multiple of 2, gear 3 would also be at an angle of 2 degrees.
Therefore, the answer is d = (7/2) * 2 = 5 degrees in total.
```
</details>
## Request for Help
Want to get involved?
- We would love to port [llama.cpp](https://github.com/ggerganov/llama.cpp) to work with StableLMs
- Integration into [Open Assistant](https://github.com/LAION-AI/Open-Assistant) from LAION-AI to collect high quality human-generated feedback data
- ... Reach out to us with ideas on our [Discord](https://discord.com/invite/stablediffusion)
## Potential issues
As is typical for any pretrained Large Language Model without additional finetuning and reinforcement learning, the responses a user gets might be of varying quality and might potentially include offensive language and views. This is expected to be improved with scale, better data, community feedback, and optimisation.
## Acknowledgements
- `StableLM-Tuned-Alpha` would not have been possible without the helpful hand of Dakota Mahan [@dmayhem93](https://huggingface.co/dmayhem93).
## Licenses
- Base model checkpoints (`StableLM-Base-Alpha`) are licensed under the Creative Commons license ([CC BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/)). Under the license, you must give [credit](https://creativecommons.org/licenses/by/4.0/#) to Stability AI, provide a link to the license, and [indicate if changes were made](https://creativecommons.org/licenses/by/4.0/#). You may do so in any reasonable manner, but not in any way that suggests the Stability AI endorses you or your use.
- Fine-tuned checkpoints (`StableLM-Tuned-Alpha`) are licensed under the Non-Commercial Creative Commons license ([CC BY-NC-SA-4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/)), in-line with the original non-commercial license specified by [Stanford Alpaca](https://github.com/tatsu-lab/stanford_alpaca).
- All code in this repository is licensed under the Apache License 2.0 license.
================================================
FILE: configs/stablelm-2-12b.yml
================================================
{
# parallelism settings
"pipe-parallel-size": 0,
"model-parallel-size": 2,
# model settings
"num-layers": 40,
"hidden-size": 5120,
"num-attention-heads": 32,
"seq-length": 4096,
"max-position-embeddings": 4096,
# architecture design
"attention_head_type": "groupedquery",
"num_kv_heads": 8,
"qk_norm": true,
"norm": "layernorm",
"pos-emb": "rotary",
"rotary_pct": 0.25,
"rotary_emb_base": 10_000,
"rotary_interleaved": false, # GPT-NeoX style
"mlp_multiple_of": 256,
"mlp_type": "gated",
"activation": "silu",
"no-weight-tying": true,
"gpt_j_residual": true,
"gpt_j_tied": true,
"output_layer_parallelism": "column",
# init methods
"init_method": "normal",
"output_layer_init_method": "scaled_normal",
"init_method_std": 0.01,
# biases
"use_bias_in_norms": false,
"use_bias_in_qk_norm": false,
"use_bias_in_attn_linear": false,
"use_bias_in_mlp": false,
# fused ops
"use_flash_cross_entropy": true,
"bias-gelu-fusion": false,
"scaled-upper-triang-masked-softmax-fusion": false,
"attention-config": [[["flash"], 40]],
# optimizer settings
"optimizer":
{
"type": "Adam",
"params":
{
"lr": 3.0e-4,
"betas": [0.9, 0.95],
"eps": 0.00000001,
},
},
"min_lr": 3.0e-5,
"train-iters": 760_000,
"lr-decay-iters": 760_000,
"lr-decay-style": "hybrid_cosine_inv_sqrt_2",
"warmup": 0.0065,
"cooldown": 0.,
"reset_attention_mask": true,
"reset_position_ids": true,
# for all zero_optimization options, see https://www.deepspeed.ai/docs/config-json/#zero-optimizations-for-fp16-training
"zero_optimization":
{
"stage": 1,
"allgather_partitions": true,
"allgather_bucket_size": 5_000_000_000,
"overlap_comm": true,
"reduce_scatter": true,
"reduce_bucket_size": 5_000_000_000,
"contiguous_gradients": true,
"cpu_offload": false,
},
# batch / data settings
"train_micro_batch_size_per_gpu": 1,
"gradient_accumulation_steps": 8,
"data-impl": "mmap",
"eval-interval": 5_000,
"eval-iters": 10,
"eval_batch_size": 1,
"eval_tasks": [],
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 40,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 0.75,
"weight-decay": 0.1,
"hidden-dropout": 0,
"attention-dropout": 0,
# precision settings
"bf16": { "enabled": true },
"precision": "bfloat16",
"full_precision_lm_cross_entropy": true,
"fp32_allreduce": true,
# misc. training settings
"num-workers": 2,
"distributed-backend": "nccl",
# checkpoint settings
"checkpoint-factor": 2_000,
"s3_sync_interval": 20_000,
"extra-save-iters": [0],
"save": "",
"load": "",
"s3_path": "",
"train_data_paths": [],
"train_data_weights": [],
"valid-data-paths": ["minipile_validation_arcade100k_tokenized_text_document"],
"valid-data-weights": [1.0,],
"test-data-paths": ["minipile_validation_arcade100k_tokenized_text_document"],
"test-data-weights": [1.0,],
# tokenizer settings
"tokenizer-type": "TiktokenTokenizer",
"vocab-file": "arcade100k.tiktoken",
"log-interval": 10,
"steps_per_print": 10,
"wall_clock_breakdown": true,
"use_wandb": true,
"wandb_host": "https://stability.wandb.io",
"wandb_team": "stability-llm",
"wandb_project": "",
"wandb_group": "",
"wandb_name": "",
# "wandb_id": "",
# "wandb_resume": "must",
# MuP
"use-mup": false,
"save-base-shapes": false, # this only needs to be enabled once in order to generate the base-shapes-file on each rank
"base-shapes-file": "shapes/shapes_AR/shapes_32L/base-shapes", # load base shapes from this file
"coord-check": false, # generate coord check plots to verify mup's implementation in neox
# multi-node launcher
"launcher": "slurm",
"deepspeed_slurm": true,
"seed": 2345678926,
}
================================================
FILE: configs/stablelm-2-1_6b.yml
================================================
{
# parallelism settings
"pipe-parallel-size": 0,
"model-parallel-size": 1,
# model settings
"num-layers": 24,
"hidden-size": 2048,
"num-attention-heads": 32,
"seq-length": 4096,
"max-position-embeddings": 4096,
# architecture design
"attention_head_type": "multihead",
"norm": "layernorm",
"pos-emb": "rotary",
"rotary_pct": 0.25,
"rotary_interleaved": false, # GPT-NeoX style
"mlp_multiple_of": 256,
"mlp_type": "gated",
"activation": "silu",
"no-weight-tying": true,
"gpt_j_residual": false,
"gpt_j_tied": false,
"output_layer_parallelism": "column",
# init methods
"init_method": "normal",
"output_layer_init_method": "scaled_normal",
"init_method_std": 0.02,
# biases
"use_bias_in_norms": false,
"use_bias_in_attn_linear": false,
"use_bias_in_mlp": false,
# fused ops
"use_flash_cross_entropy": true,
"bias-gelu-fusion": false,
"scaled-upper-triang-masked-softmax-fusion": false,
"attention-config": [[["flash"], 24]],
# optimizer settings
"optimizer": {
"type": "Adam",
"params": {
"lr": 0.001,
"betas": [0.9, 0.95],
"eps": 1.0e-8,
}
},
"min_lr": 0.0001,
"train-iters": 540_000,
"lr-decay-iters": 540_000,
"lr-decay-style": "hybrid_cosine_inv_sqrt_2",
"warmup": 0.018,
"cooldown": 0.,
"reset_attention_mask": true,
"reset_position_ids": true,
# for all zero_optimization options, see https://www.deepspeed.ai/docs/config-json/#zero-optimizations-for-fp16-training
"zero_optimization": {
"stage": 1,
"allgather_partitions": true,
"allgather_bucket_size": 1260000000,
"overlap_comm": true,
"reduce_scatter": true,
"reduce_bucket_size": 1260000000,
"contiguous_gradients": true,
"cpu_offload": false,
},
# batch / data settings
"train_micro_batch_size_per_gpu": 2,
"gradient_accumulation_steps": 2,
"data-impl": "mmap",
"eval-interval": 500_000,
"eval-iters": 1,
"eval_batch_size": 1,
"eval_tasks": [],
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 24,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 1,
"weight-decay": 0.1,
"hidden-dropout": 0.,
"attention-dropout": 0.,
# precision settings
"bf16": { "enabled": true },
"precision": "bfloat16",
"full_precision_lm_cross_entropy": true,
"fp32_allreduce": true,
# misc. training settings
"num-workers": 2,
"distributed-backend": "nccl",
# checkpoint settings
"checkpoint-factor": 2_000,
"s3_sync_interval": 20_000,
"extra-save-iters": [0],
"save": "",
"load": "",
"s3_path": "",
"train_data_paths": [],
"train_data_weights": [],
"valid-data-paths": ["minipile_validation_arcade100k_tokenized_text_document"],
"valid-data-weights": [1.0,],
"test-data-paths": ["minipile_validation_arcade100k_tokenized_text_document"],
"test-data-weights": [1.0,],
# tokenizer settings
"tokenizer-type": "TiktokenTokenizer",
"vocab-file": "arcade100k.tiktoken",
"log-interval": 10,
"steps_per_print": 10,
"wall_clock_breakdown": true,
"use_wandb": true,
"wandb_host": "https://stability.wandb.io",
"wandb_team": "stability-llm",
"wandb_project": "",
"wandb_group": "",
"wandb_name": "",
# "wandb_id": "",
# "wandb_resume": "must",
# MuP
"use-mup": false,
"save-base-shapes": false , # this only needs to be enabled once in order to generate the base-shapes-file on each rank
"base-shapes-file": "mup-base-shapes-small-fixed/base-shapes-small", # load base shapes from this file
"coord-check": false, # generate coord check plots to verify mup's implementation in neox
# multi-node launcher
"launcher": "slurm",
"deepspeed_slurm": true,
"seed": 1234
}
================================================
FILE: configs/stablelm-3b-4e1t.yml
================================================
{
# parallelism settings
"pipe-parallel-size": 1,
"model-parallel-size": 1,
# model settings
"num-layers": 32,
"hidden-size": 2560,
"num-attention-heads": 32,
"seq-length": 4096,
"max-position-embeddings": 4096,
# architecture design
"attention_head_type": "multihead",
"norm": "layernorm",
"pos-emb": "rotary",
"rotary_pct": 0.25,
"rotary_interleaved": false,
"mlp_multiple_of": 256,
"mlp_type": "gated",
"activation": "silu",
"no-weight-tying": true,
"gpt_j_residual": false,
"gpt_j_tied": false,
"output_layer_parallelism": "column",
# init methods
"output_layer_init_method": "scaled_normal",
# biases
"use_bias_in_norms": true,
"use_bias_in_attn_linear": false,
"use_bias_in_mlp": false,
# fused ops
"attention-config": [[["flash"], 32]],
# optimizer settings
"optimizer": {
"type": "Adam",
"params": {
"lr": 3.2e-4,
"betas": [0.9, 0.95],
"eps": 1.0e-6
},
},
"min_lr": 1.28e-5, # Decay to 4% of lr
# 955_000 iters ~= 4.0T tokens at bs=4M
"train-iters": 955_000,
"lr-decay-iters": 955_000,
"lr-decay-style": "cosine",
"warmup": 0.005, # ~5k warmup steps
# ZeRO settings
"zero_optimization": {
"stage": 1,
"allgather_partitions": true,
"allgather_bucket_size": 1260000000,
"overlap_comm": true,
"reduce_scatter": true,
"reduce_bucket_size": 1260000000,
"contiguous_gradients": true,
"cpu_offload": false,
},
# batch / data settings
"train_micro_batch_size_per_gpu": 4,
"gradient_accumulation_steps": 1,
"data-impl": "mmap",
"eval-interval": 5_000,
"eval-iters": 10,
"eval_batch_size": 8,
"eval_tasks": [],
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 1,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 1.0,
"weight-decay": 0.1,
"hidden-dropout": 0,
"attention-dropout": 0,
# precision settings
"bf16": { "enabled": true },
"precision": "bfloat16",
"full_precision_lm_cross_entropy": true,
# misc. training settings
"num-workers": 2,
"distributed-backend": "nccl",
# checkpoint settings
"checkpoint-factor": 2_000,
"s3_sync_interval": 10_000,
"extra-save-iters": [0],
"save": "",
"load": "",
"s3_path": "",
# data path settings
"train-data-paths": [],
"train-data-weights": [],
"valid-data-paths": [],
"valid-data-weights": [],
"test-data-paths": [],
"test-data-weights": [],
# tokenizer settings
"tokenizer-type": "HFTokenizer",
"vocab-file": "neox-tokenizer-vocab.json",
# log settings
"log-interval": 10,
"steps_per_print": 10,
"wall_clock_breakdown": true,
"use_wandb": true,
"wandb_host": "",
"wandb_team": "",
"wandb_project": "",
"wandb_group": "",
"wandb_name": "",
# "wandb_id": "",
# "wandb_resume": "",
# multi-node launcher
"launcher": "slurm",
"deepspeed_slurm": true,
}
================================================
FILE: configs/stablelm-base-alpha-3b-v2-4k-extension.yml
================================================
{
# parallelism settings
"pipe-parallel-size": 1,
"model-parallel-size": 2,
# model settings
"num-layers": 32,
"hidden-size": 2560,
"num-attention-heads": 32,
"seq-length": 4096,
"max-position-embeddings": 4096,
# architecture design
"attention_head_type": "multihead",
"norm": "layernorm",
"pos-emb": "rotary",
"rotary_pct": 0.25,
"rotary_interleaved": false, # GPT-NeoX style
# NOTE: Linear Position Scaling degrades sample quality after 10B tokens - do not use yet.
# "rotary_scaling_factor": 2, # 2048 -> 4096
"mlp_multiple_of": 256,
"mlp_type": "gated",
"activation": "silu",
"no-weight-tying": true,
"gpt_j_residual": true,
"gpt_j_tied": true,
"output_layer_parallelism": "column",
# biases
"use_bias_in_norms": true,
"use_bias_in_attn_linear": false,
"use_bias_in_mlp": false,
# fused ops
"bias-gelu-fusion": false,
"scaled-upper-triang-masked-softmax-fusion": true,
"attention-config": [[["flash"], 32]],
# optimizer settings
"optimizer": {
"type": "Adam",
"params": {
"lr": 2.8e-5,
"betas": [0.9, 0.95],
"eps": 1.0e-6
},
},
"min_lr": 2.8e-6,
"train-iters": 50_000,
"lr-decay-iters": 50_000,
"lr-decay-style": "cosine",
"warmup": 0.00,
# for all zero_optimization options, see https://www.deepspeed.ai/docs/config-json/#zero-optimizations-for-fp16-training
"zero_optimization": {
"stage": 1,
"allgather_partitions": true,
"allgather_bucket_size": 1260000000,
"overlap_comm": true,
"reduce_scatter": true,
"reduce_bucket_size": 1260000000,
"contiguous_gradients": true,
"cpu_offload": false,
},
# batch / data settings
"train_micro_batch_size_per_gpu": 4,
"gradient_accumulation_steps": 1,
"data-impl": "mmap",
"eval-interval": 1_000,
"eval-iters": 10,
"eval_batch_size": 8,
"eval_tasks": ["lambada_openai", "piqa"],
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 1,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 1.0,
"weight-decay": 0.0001,
"hidden-dropout": 0,
"attention-dropout": 0,
# precision settings
"fp16": {
"fp16": true,
"enabled": true,
"loss_scale": 0,
"loss_scale_window": 1000,
"initial_scale_power": 12,
"hysteresis": 2,
"min_loss_scale": 1e-10
},
"full_precision_lm_cross_entropy": true,
# misc. training settings
"num-workers": 1,
"distributed-backend": "nccl",
# checkpoint settings
"checkpoint-factor": 2_000,
"save": "",
"load": "",
"s3_path": "",
"iteration": 245_000,
"finetune": true,
"no_checkpoint_arg_validation": true,
"override_lr_scheduler": true,
# data path settings
"train-data-paths": [""],
"train-data-weights": [1.0],
"valid-data-paths": [""],
"valid-data-weights": [1.0],
"test-data-paths": [""],
"test-data-weights": [1.0],
# tokenizer settings
"tokenizer-type": "HFTokenizer",
"vocab-file": "neox-tokenizer-vocab.json",
# log settings
"log-interval": 10,
"steps_per_print": 10,
"wall_clock_breakdown": true,
"use_wandb": true,
"wandb_host": "",
"wandb_team": "",
"wandb_project": "",
"wandb_group": "7B",
"wandb_name": "stablelm-base-alpha-7b-v2-4k-finetune",
# "wandb_id": "",
# "wandb_resume": "must",
# multi-node launcher
"launcher": "slurm",
"deepspeed_slurm": true,
}
================================================
FILE: configs/stablelm-base-alpha-3b-v2.yml
================================================
{
# parallelism settings
"pipe-parallel-size": 1,
"model-parallel-size": 2,
# model settings
"num-layers": 32,
"hidden-size": 2560,
"num-attention-heads": 32,
"seq-length": 2048,
"max-position-embeddings": 2048,
# architecture design
"attention_head_type": "multihead",
"norm": "layernorm",
"pos-emb": "rotary",
"rotary_pct": 0.25,
"rotary_interleaved": false, # GPT-NeoX style
"mlp_multiple_of": 256,
"mlp_type": "gated",
"activation": "silu",
"no-weight-tying": true,
"gpt_j_residual": true,
"gpt_j_tied": true,
"output_layer_parallelism": "column",
# biases
"use_bias_in_norms": true,
"use_bias_in_attn_linear": false,
"use_bias_in_mlp": false,
# fused ops
"bias-gelu-fusion": false,
"scaled-upper-triang-masked-softmax-fusion": true,
"attention-config": [[["flash"], 32]],
# optimizer settings
"optimizer": {
"type": "Adam",
"params": {
"lr": 3.2e-4,
"betas": [0.9, 0.95],
"eps": 1.0e-6
},
},
"min_lr": 3.2e-5,
"train-iters": 245_000,
"lr-decay-iters": 245_000,
"lr-decay-style": "cosine",
"warmup": 0.01,
# for all zero_optimization options, see https://www.deepspeed.ai/docs/config-json/#zero-optimizations-for-fp16-training
"zero_optimization": {
"stage": 1,
"allgather_partitions": true,
"allgather_bucket_size": 1260000000,
"overlap_comm": true,
"reduce_scatter": true,
"reduce_bucket_size": 1260000000,
"contiguous_gradients": true,
"cpu_offload": false,
},
# batch / data settings
"train_micro_batch_size_per_gpu": 16,
"gradient_accumulation_steps": 1,
"data-impl": "mmap",
"eval-interval": 10_000,
"eval-iters": 10,
"eval_batch_size": 4,
"eval_tasks": ["lambada_openai", "piqa"],
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 1,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 1.0,
"weight-decay": 0.1,
"hidden-dropout": 0,
"attention-dropout": 0,
# precision settings
"fp16": {
"fp16": true,
"enabled": true,
"loss_scale": 0,
# NOTE: Mid-training divergence required a loss scale of 1e-10
# "loss_scale_window": 1000,
# "initial_scale_power": 12,
# "hysteresis": 2,
# "min_loss_scale": 1
"loss_scale_window": 1000,
"initial_scale_power": 12,
"hysteresis": 2,
"min_loss_scale": 1e-10
},
"full_precision_lm_cross_entropy": true,
# misc. training settings
"num-workers": 1,
"distributed-backend": "nccl",
# checkpoint settings
"checkpoint-factor": 2_000,
"save": "",
"load": "",
"s3_path": "",
# data path settings
"train-data-paths": [""],
"train-data-weights": [1.0],
"valid-data-paths": [""],
"valid-data-weights": [1.0],
"test-data-paths": [""],
"test-data-weights": [1.0],
# tokenizer settings
"tokenizer-type": "HFTokenizer",
"vocab-file": "neox-tokenizer-vocab.json",
# log settings
"log-interval": 10,
"steps_per_print": 10,
"wall_clock_breakdown": true,
"use_wandb": true,
"wandb_host": "",
"wandb_team": "",
"wandb_project": "",
"wandb_group": "3B",
"wandb_name": "stablelm-base-alpha-3b-v2",
# "wandb_id": "",
# "wandb_resume": "must",
# multi-node launcher
"launcher": "slurm",
"deepspeed_slurm": true,
}
================================================
FILE: configs/stablelm-base-alpha-3b.yml
================================================
{
# parallelism settings
"pipe-parallel-size": 1,
"model-parallel-size": 4,
# model settings
"num-layers": 16,
"hidden-size": 4096,
"num-attention-heads": 32,
"seq-length": 4096,
"max-position-embeddings": 4096,
# architecture design
"norm": "layernorm",
"pos-emb": "rotary",
"rotary_pct": 0.25,
"activation": "gelu",
"no-weight-tying": true,
"gpt_j_residual": true,
"output_layer_parallelism": "column",
# init methods
"init_method": "small_init",
"output_layer_init_method": "wang_init",
# fused ops
"scaled-upper-triang-masked-softmax-fusion": true,
"bias-gelu-fusion": true,
"attention-config": [[["flash"], 16]],
# optimizer settings
"optimizer": {
"type": "Adam",
"params": {
"lr": 1.6e-4,
"betas": [0.9, 0.9999],
"eps": 1.0e-6
},
},
"min_lr": 1.6e-5,
# for all zero_optimization options, see https://www.deepspeed.ai/docs/config-json/#zero-optimizations-for-fp16-training
"zero_optimization": {
"stage": 1,
"allgather_partitions": true,
"allgather_bucket_size": 1260000000,
"overlap_comm": true,
"reduce_scatter": true,
"reduce_bucket_size": 1260000000,
"contiguous_gradients": true,
"cpu_offload": false,
},
# batch / data settings
"train_micro_batch_size_per_gpu": 32,
"gradient_accumulation_steps": 1,
"eval_batch_size": 2,
"data-impl": "mmap",
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 1,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 1.0,
"weight-decay": 0.1,
"hidden-dropout": 0,
"attention-dropout": 0,
# precision settings
"fp16": {
"fp16": true,
"enabled": true,
"loss_scale_window": 1000,
"initial_scale_power": 12,
"hysteresis": 20,
"min_loss_scale": 1,
},
# misc. training settings
"train-iters": 180000,
"lr-decay-iters": 180000,
"distributed-backend": "nccl",
"lr-decay-style": "cosine",
"warmup": 0.01,
"checkpoint-factor": 1000,
# 1 more than checkpoint-factor to avoid skipping evals if `evaluate` fails
"eval-interval": 1001,
"eval-iters": 10,
"eval_tasks": ["piqa", "sciq", "lambada_openai"],
# checkpoint settings
"iteration": 84000,
"save": "PATH_TO_SAVE_THE_MODEL",
"load": "PATH_TO_LOAD_THE_MODEL",
# data settings
"train-data-paths": [],
"train-data-weights": [0.03, 0.02, 3.5, 13.87, 30.88, 0.34, 0.03, 0.1, 0.01, 0.5, 0.25, 0.25, 0.11, 1, 0.1, 0.5, 0.6, 0.19, 0.05, 2, 3.5, 4.15, 5.75, 3.17, 3.44, 3.49, 3, 6, 0.02, 2, 0.01, 3.95, 0.05, 1.09, 6.05],
"valid-data-paths": [],
"valid-data-weights": [0.03, 0.02, 3.5, 13.87, 30.88, 0.34, 0.03, 0.1, 0.01, 0.5, 0.25, 0.25, 0.11, 1, 0.1, 0.5, 0.6, 0.19, 0.05, 2, 3.5, 4.15, 5.75, 3.17, 3.44, 3.49, 3, 6, 0.02, 2, 0.01, 3.95, 0.05, 1.09, 6.05],
"test-data-paths": [],
"test-data-weights": [0.03, 0.02, 3.5, 13.87, 30.88, 0.34, 0.03, 0.1, 0.01, 0.5, 0.25, 0.25, 0.11, 1, 0.1, 0.5, 0.6, 0.19, 0.05, 2, 3.5, 4.15, 5.75, 3.17, 3.44, 3.49, 3, 6, 0.02, 2, 0.01, 3.95, 0.05, 1.09, 6.05],
# tokenizer settings
"tokenizer-type": "HFTokenizer",
"vocab-file": "/pile/20B_tokenizer.json",
# log settings
"log-interval": 10,
"steps_per_print": 10,
"wall_clock_breakdown": true,
"log-grad-norm": true,
"use_wandb": true,
"wandb_host": "",
"wandb_team": "",
"wandb_project": "",
"wandb_group": "",
"wandb_name": "",
# multi-node launcher
"launcher": "slurm",
"deepspeed_slurm": true
}
================================================
FILE: configs/stablelm-base-alpha-7b-v2-4k-extension.yml
================================================
{
# parallelism settings
"pipe-parallel-size": 1,
"model-parallel-size": 2,
# model settings
"num-layers": 32,
"hidden-size": 4096,
"num-attention-heads": 32,
"seq-length": 4096,
"max-position-embeddings": 4096,
# architecture design
"attention_head_type": "multihead",
"norm": "layernorm",
"pos-emb": "rotary",
"rotary_pct": 0.25,
"rotary_interleaved": false, # GPT-NeoX style
# NOTE: Linear Position Scaling degrades sample quality after 10B tokens - do not use yet.
# "rotary_scaling_factor": 2, # 2048 -> 4096
"mlp_multiple_of": 256,
"mlp_type": "gated",
"activation": "silu",
"no-weight-tying": true,
"gpt_j_residual": true,
"gpt_j_tied": true,
"output_layer_parallelism": "column",
# biases
"use_bias_in_norms": true,
"use_bias_in_attn_linear": false,
"use_bias_in_mlp": false,
# fused ops
"bias-gelu-fusion": false,
"scaled-upper-triang-masked-softmax-fusion": true,
"attention-config": [[["flash"], 32]],
# optimizer settings
"optimizer": {
"type": "Adam",
"params": {
"lr": 2.2e-5,
"betas": [0.9, 0.95],
"eps": 1.0e-6
},
},
"min_lr": 2.2e-6,
"train-iters": 45_000,
"lr-decay-iters": 45_000,
"lr-decay-style": "cosine",
"warmup": 0.00,
# for all zero_optimization options, see https://www.deepspeed.ai/docs/config-json/#zero-optimizations-for-fp16-training
"zero_optimization": {
"stage": 1,
"allgather_partitions": true,
"allgather_bucket_size": 1260000000,
"overlap_comm": true,
"reduce_scatter": true,
"reduce_bucket_size": 1260000000,
"contiguous_gradients": true,
"cpu_offload": false,
},
# batch / data settings
"train_micro_batch_size_per_gpu": 3,
"gradient_accumulation_steps": 1,
"data-impl": "mmap",
"eval-interval": 4_000,
"eval-iters": 10,
"eval_batch_size": 2,
"eval_tasks": ["lambada_openai", "piqa"],
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 1,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 1.0,
"weight-decay": 0.01,
"hidden-dropout": 0,
"attention-dropout": 0,
# precision settings
"fp16": {
"fp16": true,
"enabled": true,
"loss_scale": 0,
"loss_scale_window": 1000,
"initial_scale_power": 12,
"hysteresis": 2,
"min_loss_scale": 1e-12
},
"full_precision_lm_cross_entropy": true,
# misc. training settings
"num-workers": 1,
"distributed-backend": "nccl",
# checkpoint settings
"checkpoint-factor": 2_000,
"save": "",
"load": "",
"s3_path": "",
"iteration": 245_000,
"finetune": true,
"no_checkpoint_arg_validation": true,
"override_lr_scheduler": true,
# data path settings
"train-data-paths": [""],
"train-data-weights": [1.0],
"valid-data-paths": [""],
"valid-data-weights": [1.0],
"test-data-paths": [""],
"test-data-weights": [1.0],
# tokenizer settings
"tokenizer-type": "HFTokenizer",
"vocab-file": "neox-tokenizer-vocab.json",
# log settings
"log-interval": 10,
"steps_per_print": 10,
"wall_clock_breakdown": true,
"use_wandb": true,
"wandb_host": "",
"wandb_team": "",
"wandb_project": "",
"wandb_group": "7B",
"wandb_name": "stablelm-base-alpha-7b-v2-4k-finetune",
# "wandb_id": "",
# "wandb_resume": "must",
# multi-node launcher
"launcher": "slurm",
"deepspeed_slurm": true,
}
================================================
FILE: configs/stablelm-base-alpha-7b-v2.yml
================================================
{
# parallelism settings
"pipe-parallel-size": 1,
"model-parallel-size": 2,
# model settings
"num-layers": 32,
"hidden-size": 4096,
"num-attention-heads": 32,
"seq-length": 2048,
"max-position-embeddings": 2048,
# architecture design
"attention_head_type": "multihead",
"norm": "layernorm",
"pos-emb": "rotary",
"rotary_pct": 0.25,
"rotary_interleaved": false, # GPT-NeoX style
"mlp_multiple_of": 256,
"mlp_type": "gated",
"activation": "silu",
"no-weight-tying": true,
"gpt_j_residual": true,
"gpt_j_tied": true,
"output_layer_parallelism": "column",
# biases
"use_bias_in_norms": true,
"use_bias_in_attn_linear": false,
"use_bias_in_mlp": false,
# fused ops
"bias-gelu-fusion": false,
"scaled-upper-triang-masked-softmax-fusion": true,
"attention-config": [[["flash"], 32]],
# optimizer settings
"optimizer": {
"type": "Adam",
"params": {
"lr": 3.0e-4,
"betas": [0.9, 0.95],
"eps": 1.0e-6
},
},
"min_lr": 3.0e-5,
"train-iters": 245_000,
"lr-decay-iters": 245_000,
"lr-decay-style": "cosine",
"warmup": 0.01,
# for all zero_optimization options, see https://www.deepspeed.ai/docs/config-json/#zero-optimizations-for-fp16-training
"zero_optimization": {
"stage": 1,
"allgather_partitions": true,
"allgather_bucket_size": 1260000000,
"overlap_comm": true,
"reduce_scatter": true,
"reduce_bucket_size": 1260000000,
"contiguous_gradients": true,
"cpu_offload": false,
},
# batch / data settings
"train_micro_batch_size_per_gpu": 12,
"gradient_accumulation_steps": 1,
"data-impl": "mmap",
"eval-interval": 10_000,
"eval-iters": 10,
"eval_batch_size": 2,
"eval_tasks": ["lambada_openai", "piqa"],
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 1,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 1.0,
"weight-decay": 0.1,
"hidden-dropout": 0,
"attention-dropout": 0,
# precision settings
"fp16": {
"fp16": true,
"enabled": true,
"loss_scale": 0,
"loss_scale_window": 1000,
"initial_scale_power": 12,
"hysteresis": 2,
"min_loss_scale": 1e-12
},
"full_precision_lm_cross_entropy": true,
# misc. training settings
"num-workers": 1,
"distributed-backend": "nccl",
# checkpoint settings
"checkpoint-factor": 2_000,
"save": "",
"load": "",
"s3_path": "",
# data path settings
"train-data-paths": [""],
"train-data-weights": [1.0],
"valid-data-paths": [""],
"valid-data-weights": [1.0],
"test-data-paths": [""],
"test-data-weights": [1.0],
# tokenizer settings
"tokenizer-type": "HFTokenizer",
"vocab-file": "neox-tokenizer-vocab.json",
# log settings
"log-interval": 10,
"steps_per_print": 10,
"wall_clock_breakdown": true,
"use_wandb": true,
"wandb_host": "",
"wandb_team": "",
"wandb_project": "",
"wandb_group": "7B",
"wandb_name": "stablelm-base-alpha-7b-v2",
# "wandb_id": "",
# "wandb_resume": "must",
# multi-node launcher
"launcher": "slurm",
"deepspeed_slurm": true,
}
================================================
FILE: configs/stablelm-base-alpha-7b.yml
================================================
{
# parallelism settings
"pipe-parallel-size": 1,
"model-parallel-size": 2,
# model settings
"num-layers": 16,
"hidden-size": 6144,
"num-attention-heads": 48,
"seq-length": 4096,
"max-position-embeddings": 4096,
# architecture design
"norm": "layernorm",
"pos-emb": "rotary",
"rotary_pct": 0.25,
"activation": "gelu",
"no-weight-tying": true,
"gpt_j_residual": true,
"output_layer_parallelism": "column",
# init methods
"init_method": "small_init",
"output_layer_init_method": "wang_init",
# fused ops
"scaled-upper-triang-masked-softmax-fusion": true,
"bias-gelu-fusion": true,
"attention-config": [[["flash"], 16]],
# optimizer settings
"optimizer": {
"type": "Adam",
"params": {
"lr": 1.5e-4,
"betas": [0.9, 0.95],
"eps": 1.0e-8
},
},
"min_lr": 1.5e-5,
# for all zero_optimization options, see https://www.deepspeed.ai/docs/config-json/#zero-optimizations-for-fp16-training
"zero_optimization": {
"stage": 1,
"allgather_partitions": true,
"allgather_bucket_size": 1260000000,
"overlap_comm": true,
"reduce_scatter": true,
"reduce_bucket_size": 1260000000,
"contiguous_gradients": true,
},
# batch / data settings
"train_micro_batch_size_per_gpu": 8,
"gradient_accumulation_steps": 1,
"eval_batch_size": 1,
"data-impl": "mmap",
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 1,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 1.0,
"weight-decay": 0.1,
"hidden-dropout": 0,
"attention-dropout": 0,
# precision settings
"fp16": {
"fp16": true,
"enabled": true,
"loss_scale": 0,
"loss_scale_window": 1000,
"initial_scale_power": 12,
"hysteresis": 2,
"min_loss_scale": 1,
},
# misc. training settings
"train-iters": 180000,
"lr-decay-iters": 180000,
"distributed-backend": "nccl",
"lr-decay-style": "cosine",
"warmup": 0.01,
"checkpoint-factor": 1000,
# 1 more than checkpoint-factor to avoid skipping evals if `evaluate` fails
"eval-interval": 1001,
"eval-iters": 10,
"eval_tasks": ["piqa", "sciq", "lambada_openai"],
# checkpoint settings
"iteration": 0,
"save": "",
"load": "",
# data settings
"train-data-paths": [],
"train-data-weights": [0.03, 0.02, 3.5, 13.87, 30.88, 0.34, 0.03, 0.1, 0.01, 0.5, 0.25, 0.25, 0.11, 1, 0.1, 0.5, 0.6, 0.19, 0.05, 2, 3.5, 4.15, 5.75, 3.17, 3.44, 3.49, 3, 6, 0.02, 2, 0.01, 3.95, 0.05, 1.09, 6.05],
"valid-data-paths": [],
"valid-data-weights": [0.03, 0.02, 3.5, 13.87, 30.88, 0.34, 0.03, 0.1, 0.01, 0.5, 0.25, 0.25, 0.11, 1, 0.1, 0.5, 0.6, 0.19, 0.05, 2, 3.5, 4.15, 5.75, 3.17, 3.44, 3.49, 3, 6, 0.02, 2, 0.01, 3.95, 0.05, 1.09, 6.05],
"test-data-paths": [],
"test-data-weights": [0.03, 0.02, 3.5, 13.87, 30.88, 0.34, 0.03, 0.1, 0.01, 0.5, 0.25, 0.25, 0.11, 1, 0.1, 0.5, 0.6, 0.19, 0.05, 2, 3.5, 4.15, 5.75, 3.17, 3.44, 3.49, 3, 6, 0.02, 2, 0.01, 3.95, 0.05, 1.09, 6.05],
# tokenizer settings
"tokenizer-type": "HFTokenizer",
"vocab-file": "/pile/20B_tokenizer.json",
# log settings
"log-interval": 10,
"steps_per_print": 10,
"wall_clock_breakdown": true,
"use_wandb": true,
"wandb_host": "",
"wandb_team": "",
"wandb_project": "",
"wandb_group": "",
"wandb_name": "",
# multi-node launcher
"launcher": "slurm",
"deepspeed_slurm": true
}
================================================
FILE: evals/external/EleutherAI-pythia-2.8b-deduped.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.30119453924914674,
"acc_stderr": 0.013406741767847626,
"acc_norm": 0.3293515358361775,
"acc_norm_stderr": 0.013734057652635474
},
"arc_easy": {
"acc": 0.6346801346801347,
"acc_stderr": 0.009880576614806924,
"acc_norm": 0.5909090909090909,
"acc_norm_stderr": 0.010088775152615782
},
"boolq": {
"acc": 0.6412844036697247,
"acc_stderr": 0.008388668034059405
},
"hellaswag": {
"acc": 0.45429197371041624,
"acc_stderr": 0.004968888130290072,
"acc_norm": 0.5944035052778331,
"acc_norm_stderr": 0.004900036261309038
},
"lambada_openai": {
"ppl": 5.00138268807375,
"ppl_stderr": 0.11803810628354432,
"acc": 0.6514651659227635,
"acc_stderr": 0.0066386652033128745
},
"openbookqa": {
"acc": 0.238,
"acc_stderr": 0.019064072958198446,
"acc_norm": 0.348,
"acc_norm_stderr": 0.02132372863280751
},
"piqa": {
"acc": 0.7410228509249184,
"acc_stderr": 0.0102209660314056,
"acc_norm": 0.7404787812840044,
"acc_norm_stderr": 0.010227939888173923
},
"sciq": {
"acc": 0.882,
"acc_stderr": 0.010206869264381791,
"acc_norm": 0.832,
"acc_norm_stderr": 0.011828605831454262
},
"siqa": {
"acc": 0.4094165813715456,
"acc_stderr": 0.011126849576589028,
"acc_norm": 0.44319344933469806,
"acc_norm_stderr": 0.011240812731564954
},
"truthfulqa_mc": {
"mc1": 0.2141982864137087,
"mc1_stderr": 0.014362148155690466,
"mc2": 0.3555711185495532,
"mc2_stderr": 0.013587679864140447
},
"winogrande": {
"acc": 0.5824782951854776,
"acc_stderr": 0.01385997826444025
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=EleutherAI/pythia-2.8b-deduped,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto","num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:4",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/EleutherAI_gpt-j-6B.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.3395904436860068,
"acc_stderr": 0.01383903976282016,
"acc_norm": 0.3651877133105802,
"acc_norm_stderr": 0.014070265519268802
},
"arc_easy": {
"acc": 0.6696127946127947,
"acc_stderr": 0.009651430216428178,
"acc_norm": 0.6224747474747475,
"acc_norm_stderr": 0.009947227833469421
},
"boolq": {
"acc": 0.654434250764526,
"acc_stderr": 0.008317463342191592
},
"hellaswag": {
"acc": 0.49522007568213505,
"acc_stderr": 0.004989553396413105,
"acc_norm": 0.6624178450507867,
"acc_norm_stderr": 0.004719187890948069
},
"lambada_openai": {
"ppl": 4.102416000764715,
"ppl_stderr": 0.08849985162393556,
"acc": 0.6823209780710265,
"acc_stderr": 0.0064863548390796605
},
"openbookqa": {
"acc": 0.29,
"acc_stderr": 0.02031317923174518,
"acc_norm": 0.382,
"acc_norm_stderr": 0.02175082059125084
},
"piqa": {
"acc": 0.7557127312295974,
"acc_stderr": 0.010024765172284244,
"acc_norm": 0.7616974972796517,
"acc_norm_stderr": 0.009940334245876224
},
"sciq": {
"acc": 0.915,
"acc_stderr": 0.008823426366942316,
"acc_norm": 0.874,
"acc_norm_stderr": 0.010499249222408035
},
"siqa": {
"acc": 0.4109518935516888,
"acc_stderr": 0.011133193398910182,
"acc_norm": 0.4508700102354145,
"acc_norm_stderr": 0.011259319269273942
},
"winogrande": {
"acc": 0.6416732438831886,
"acc_stderr": 0.013476581172567524
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "pretrained=EleutherAI/gpt-j-6B,dtype=float16,trust_remote_code=True,low_cpu_mem_usage=True,use_fast=True",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/EleutherAI_gpt-neox-20B.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.378839590443686,
"acc_stderr": 0.014175915490000319,
"acc_norm": 0.40784982935153585,
"acc_norm_stderr": 0.014361097288449696
},
"arc_easy": {
"acc": 0.7289562289562289,
"acc_stderr": 0.0091209197417606,
"acc_norm": 0.6868686868686869,
"acc_norm_stderr": 0.009516303879309528
},
"boolq": {
"acc": 0.6948012232415902,
"acc_stderr": 0.00805404814192796
},
"hellaswag": {
"acc": 0.5356502688707429,
"acc_stderr": 0.004977081808179433,
"acc_norm": 0.714299940250946,
"acc_norm_stderr": 0.004508239594503832
},
"lambada_openai": {
"ppl": 3.6403044358845733,
"ppl_stderr": 0.0747797639775495,
"acc": 0.7197748884145159,
"acc_stderr": 0.0062569681407934575
},
"openbookqa": {
"acc": 0.298,
"acc_stderr": 0.02047511809298897,
"acc_norm": 0.402,
"acc_norm_stderr": 0.021948929609938602
},
"piqa": {
"acc": 0.7742110990206746,
"acc_stderr": 0.009754980670917316,
"acc_norm": 0.7845484221980413,
"acc_norm_stderr": 0.009592463115658116
},
"sciq": {
"acc": 0.931,
"acc_stderr": 0.008018934050315145,
"acc_norm": 0.89,
"acc_norm_stderr": 0.009899393819724444
},
"siqa": {
"acc": 0.4196519959058342,
"acc_stderr": 0.011167032303390547,
"acc_norm": 0.44728761514841353,
"acc_norm_stderr": 0.011251020423273035
},
"winogrande": {
"acc": 0.6614048934490924,
"acc_stderr": 0.01330016986584241
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "pretrained=EleutherAI/gpt-neox-20B,dtype=float16,trust_remote_code=True,low_cpu_mem_usage=True,use_fast=True",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/EleutherAI_pythia-12b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.318259385665529,
"acc_stderr": 0.013611993916971451,
"acc_norm": 0.3506825938566553,
"acc_norm_stderr": 0.013944635930726083
},
"arc_easy": {
"acc": 0.702020202020202,
"acc_stderr": 0.009385046066694871,
"acc_norm": 0.6372053872053872,
"acc_norm_stderr": 0.00986593675701393
},
"boolq": {
"acc": 0.6730886850152905,
"acc_stderr": 0.008204340208838751
},
"hellaswag": {
"acc": 0.5046803425612428,
"acc_stderr": 0.004989562798280524,
"acc_norm": 0.673770165305716,
"acc_norm_stderr": 0.004678743563766643
},
"lambada_openai": {
"ppl": 3.9264187287921497,
"ppl_stderr": 0.08375179732007268,
"acc": 0.7063846303124394,
"acc_stderr": 0.006344860619678724
},
"openbookqa": {
"acc": 0.264,
"acc_stderr": 0.019732885585922108,
"acc_norm": 0.372,
"acc_norm_stderr": 0.0216371979857224
},
"piqa": {
"acc": 0.7627856365614799,
"acc_stderr": 0.009924694933586366,
"acc_norm": 0.7698585418933623,
"acc_norm_stderr": 0.009820832826839798
},
"sciq": {
"acc": 0.902,
"acc_stderr": 0.009406619184621235,
"acc_norm": 0.851,
"acc_norm_stderr": 0.011266140684632175
},
"siqa": {
"acc": 0.4201637666325486,
"acc_stderr": 0.011168911571162012,
"acc_norm": 0.44268167860798363,
"acc_norm_stderr": 0.011239482425741961
},
"winogrande": {
"acc": 0.6400947119179163,
"acc_stderr": 0.013489609590266804
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "pretrained=EleutherAI/pythia-12b,dtype=float16,trust_remote_code=True,low_cpu_mem_usage=True,use_fast=True",
"num_fewshot": 0,
"batch_size": "1",
"batch_sizes": [],
"device": "cuda",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/EleutherAI_pythia-6.9b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.318259385665529,
"acc_stderr": 0.013611993916971451,
"acc_norm": 0.3532423208191126,
"acc_norm_stderr": 0.013967822714840055
},
"arc_easy": {
"acc": 0.6721380471380471,
"acc_stderr": 0.009632587076170013,
"acc_norm": 0.6106902356902357,
"acc_norm_stderr": 0.010005212782878145
},
"boolq": {
"acc": 0.6400611620795107,
"acc_stderr": 0.008394940698368873
},
"hellaswag": {
"acc": 0.48048197570205137,
"acc_stderr": 0.0049859782149379184,
"acc_norm": 0.6388169687313284,
"acc_norm_stderr": 0.004793617835645056
},
"lambada_openai": {
"ppl": 4.457423605452202,
"ppl_stderr": 0.10007464956313723,
"acc": 0.670095090238696,
"acc_stderr": 0.00655050345779628
},
"openbookqa": {
"acc": 0.258,
"acc_stderr": 0.019586711785215837,
"acc_norm": 0.372,
"acc_norm_stderr": 0.0216371979857224
},
"piqa": {
"acc": 0.750816104461371,
"acc_stderr": 0.01009188277012022,
"acc_norm": 0.76550598476605,
"acc_norm_stderr": 0.009885203143240538
},
"sciq": {
"acc": 0.898,
"acc_stderr": 0.00957536880165389,
"acc_norm": 0.84,
"acc_norm_stderr": 0.011598902298689004
},
"siqa": {
"acc": 0.40736949846468784,
"acc_stderr": 0.011118216651888717,
"acc_norm": 0.42988741044012285,
"acc_norm_stderr": 0.011202283451328794
},
"winogrande": {
"acc": 0.606156274664562,
"acc_stderr": 0.013732114472668745
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "pretrained=EleutherAI/pythia-6.9b,dtype=float16,trust_remote_code=True,low_cpu_mem_usage=True,use_fast=True",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/Qwen-Qwen-7B-Chat.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.45563139931740615,
"acc_stderr": 0.014553749939306863,
"acc_norm": 0.4667235494880546,
"acc_norm_stderr": 0.014578995859605797
},
"arc_easy": {
"acc": 0.6919191919191919,
"acc_stderr": 0.00947388707582633,
"acc_norm": 0.6447811447811448,
"acc_norm_stderr": 0.009820245899287122
},
"boolq": {
"acc": 0.7168195718654434,
"acc_stderr": 0.007880052012351937
},
"hellaswag": {
"acc": 0.6792471619199363,
"acc_stderr": 0.004658120152230809,
"acc_norm": 0.8497311292571201,
"acc_norm_stderr": 0.003566044777327419
},
"lambada_openai": {
"ppl": 4.258406715958618,
"ppl_stderr": 0.1263040238577051,
"acc": 0.6547642150203765,
"acc_stderr": 0.006623879809039193
},
"openbookqa": {
"acc": 0.356,
"acc_stderr": 0.021434712356072652,
"acc_norm": 0.462,
"acc_norm_stderr": 0.02231833811987053
},
"piqa": {
"acc": 0.7872687704026116,
"acc_stderr": 0.00954822312304733,
"acc_norm": 0.7992383025027203,
"acc_norm_stderr": 0.00934596167482341
},
"sciq": {
"acc": 0.907,
"acc_stderr": 0.00918887563499669,
"acc_norm": 0.806,
"acc_norm_stderr": 0.012510816141264359
},
"siqa": {
"acc": 0.4703172978505629,
"acc_stderr": 0.011294116144908552,
"acc_norm": 0.47389969293756395,
"acc_norm_stderr": 0.01129864516098083
},
"winogrande": {
"acc": 0.6858721389108129,
"acc_stderr": 0.01304541671607256
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "pretrained=Qwen/Qwen-7B-Chat,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:4",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/Qwen-Qwen-7B.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.4539249146757679,
"acc_stderr": 0.014549221105171865,
"acc_norm": 0.49146757679180886,
"acc_norm_stderr": 0.014609263165632186
},
"arc_easy": {
"acc": 0.6738215488215489,
"acc_stderr": 0.009619849417035167,
"acc_norm": 0.6519360269360269,
"acc_norm_stderr": 0.009774627600259012
},
"boolq": {
"acc": 0.745565749235474,
"acc_stderr": 0.007617690099234367
},
"hellaswag": {
"acc": 0.7305317665803625,
"acc_stderr": 0.004427767996301626,
"acc_norm": 0.8884684325831508,
"acc_norm_stderr": 0.0031414591751392695
},
"lambada_openai": {
"ppl": 4.014976946118802,
"ppl_stderr": 0.08858949240464632,
"acc": 0.6966815447312246,
"acc_stderr": 0.006404402872809113
},
"openbookqa": {
"acc": 0.322,
"acc_stderr": 0.02091666833001988,
"acc_norm": 0.448,
"acc_norm_stderr": 0.022261697292270143
},
"piqa": {
"acc": 0.7399347116430903,
"acc_stderr": 0.010234893249061308,
"acc_norm": 0.749183895538629,
"acc_norm_stderr": 0.010113869547069044
},
"sciq": {
"acc": 0.932,
"acc_stderr": 0.007964887911291603,
"acc_norm": 0.908,
"acc_norm_stderr": 0.009144376393151117
},
"siqa": {
"acc": 0.49437052200614123,
"acc_stderr": 0.011313353423379522,
"acc_norm": 0.5240532241555783,
"acc_norm_stderr": 0.01130097128912773
},
"winogrande": {
"acc": 0.65982636148382,
"acc_stderr": 0.013315218762417397
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=Qwen/Qwen-7B,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:4",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/baichuan-inc_Baichuan2-7B-Base.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.42235494880546076,
"acc_stderr": 0.014434138713379977,
"acc_norm": 0.431740614334471,
"acc_norm_stderr": 0.014474591427196202
},
"arc_easy": {
"acc": 0.75,
"acc_stderr": 0.008885233166386385,
"acc_norm": 0.7281144781144782,
"acc_norm_stderr": 0.0091297958673105
},
"boolq": {
"acc": 0.7308868501529052,
"acc_stderr": 0.007756844290794318
},
"hellaswag": {
"acc": 0.5366460864369648,
"acc_stderr": 0.004976361454341348,
"acc_norm": 0.7228639713204541,
"acc_norm_stderr": 0.004466695023677831
},
"lambada_openai": {
"ppl": 3.6834487322486384,
"ppl_stderr": 0.07653758845393456,
"acc": 0.7098777411216767,
"acc_stderr": 0.006322580641394925
},
"openbookqa": {
"acc": 0.304,
"acc_stderr": 0.02059164957122493,
"acc_norm": 0.394,
"acc_norm_stderr": 0.021874299301689253
},
"piqa": {
"acc": 0.7616974972796517,
"acc_stderr": 0.009940334245876209,
"acc_norm": 0.7736670293797606,
"acc_norm_stderr": 0.009763294246879418
},
"sciq": {
"acc": 0.946,
"acc_stderr": 0.007150883521295436,
"acc_norm": 0.913,
"acc_norm_stderr": 0.008916866630745916
},
"siqa": {
"acc": 0.41760491299897645,
"acc_stderr": 0.011159391894922486,
"acc_norm": 0.44779938587512796,
"acc_norm_stderr": 0.011252242102001767
},
"winogrande": {
"acc": 0.675611681136543,
"acc_stderr": 0.01315722572664164
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "pretrained=baichuan-inc/Baichuan2-7B-Base,dtype=bfloat16,trust_remote_code=True,low_cpu_mem_usage=True,use_fast=True",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/bigscience-bloom-3b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.27986348122866894,
"acc_stderr": 0.013119040897725922,
"acc_norm": 0.3037542662116041,
"acc_norm_stderr": 0.013438909184778757
},
"arc_easy": {
"acc": 0.5942760942760943,
"acc_stderr": 0.010075755540128871,
"acc_norm": 0.5328282828282829,
"acc_norm_stderr": 0.010237645778853869
},
"boolq": {
"acc": 0.617125382262997,
"acc_stderr": 0.008501734385335953
},
"hellaswag": {
"acc": 0.4137621987651862,
"acc_stderr": 0.004915003499517833,
"acc_norm": 0.545309699263095,
"acc_norm_stderr": 0.004969251445596341
},
"lambada_openai": {
"ppl": 9.094700256232823,
"ppl_stderr": 0.2652067493709512,
"acc": 0.5173685231903745,
"acc_stderr": 0.006961773596960152
},
"openbookqa": {
"acc": 0.218,
"acc_stderr": 0.01848337822317886,
"acc_norm": 0.322,
"acc_norm_stderr": 0.020916668330019882
},
"piqa": {
"acc": 0.705658324265506,
"acc_stderr": 0.010633311470347498,
"acc_norm": 0.7067464635473341,
"acc_norm_stderr": 0.010621818421101931
},
"sciq": {
"acc": 0.891,
"acc_stderr": 0.009859828407037191,
"acc_norm": 0.816,
"acc_norm_stderr": 0.012259457340938588
},
"siqa": {
"acc": 0.4017400204708291,
"acc_stderr": 0.011093444192711183,
"acc_norm": 0.4314227226202661,
"acc_norm_stderr": 0.011207148736838392
},
"winogrande": {
"acc": 0.584846093133386,
"acc_stderr": 0.01384868408665859
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=bigscience/bloom-3b,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": null,
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/bigscience-bloom-7b1.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.302901023890785,
"acc_stderr": 0.013428241573185349,
"acc_norm": 0.33532423208191126,
"acc_norm_stderr": 0.013796182947785564
},
"arc_easy": {
"acc": 0.6494107744107744,
"acc_stderr": 0.009791003829831557,
"acc_norm": 0.5736531986531986,
"acc_norm_stderr": 0.010147858603835144
},
"boolq": {
"acc": 0.6284403669724771,
"acc_stderr": 0.008451598145076589
},
"hellaswag": {
"acc": 0.4649472216689902,
"acc_stderr": 0.0049775044466089996,
"acc_norm": 0.6228838876717785,
"acc_norm_stderr": 0.004836738514051334
},
"lambada_openai": {
"ppl": 6.619927277080142,
"ppl_stderr": 0.1762520708430581,
"acc": 0.5755870366776635,
"acc_stderr": 0.006885918770006387
},
"openbookqa": {
"acc": 0.252,
"acc_stderr": 0.019435727282249536,
"acc_norm": 0.358,
"acc_norm_stderr": 0.021461434862859122
},
"piqa": {
"acc": 0.7274211099020674,
"acc_stderr": 0.010389256803296021,
"acc_norm": 0.7366702937976061,
"acc_norm_stderr": 0.010276185322196764
},
"sciq": {
"acc": 0.901,
"acc_stderr": 0.009449248027662765,
"acc_norm": 0.845,
"acc_norm_stderr": 0.011450157470799475
},
"siqa": {
"acc": 0.4211873080859775,
"acc_stderr": 0.011172633149198374,
"acc_norm": 0.4498464687819857,
"acc_norm_stderr": 0.011257008360485692
},
"winogrande": {
"acc": 0.6464088397790055,
"acc_stderr": 0.013436541262599948
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=bigscience/bloom-7b1,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:5",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/cerebras-btlm-3b-8k-base.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.34897610921501704,
"acc_stderr": 0.013928933461382496,
"acc_norm": 0.37627986348122866,
"acc_norm_stderr": 0.014157022555407168
},
"arc_easy": {
"acc": 0.7045454545454546,
"acc_stderr": 0.009361987126556457,
"acc_norm": 0.6708754208754208,
"acc_norm_stderr": 0.00964204805806098
},
"boolq": {
"acc": 0.6963302752293578,
"acc_stderr": 0.008042682539896304
},
"hellaswag": {
"acc": 0.5184226249751046,
"acc_stderr": 0.004986393266269161,
"acc_norm": 0.6977693686516631,
"acc_norm_stderr": 0.004582861219020893
},
"lambada_openai": {
"ppl": 4.720441000893734,
"ppl_stderr": 0.1100556659950519,
"acc": 0.6623326217737241,
"acc_stderr": 0.00658862361668043
},
"openbookqa": {
"acc": 0.276,
"acc_stderr": 0.02001121929807353,
"acc_norm": 0.408,
"acc_norm_stderr": 0.02200091089387719
},
"piqa": {
"acc": 0.7584330794341676,
"acc_stderr": 0.009986718001804472,
"acc_norm": 0.7720348204570185,
"acc_norm_stderr": 0.009788093832324906
},
"sciq": {
"acc": 0.929,
"acc_stderr": 0.008125578442487916,
"acc_norm": 0.895,
"acc_norm_stderr": 0.009698921026024966
},
"siqa": {
"acc": 0.4278403275332651,
"acc_stderr": 0.01119562541819821,
"acc_norm": 0.4611054247697032,
"acc_norm_stderr": 0.011279787032703659
},
"truthfulqa_mc": {
"mc1": 0.2252141982864137,
"mc1_stderr": 0.014623240768023498,
"mc2": 0.3599562107238256,
"mc2_stderr": 0.013576568348894856
},
"winogrande": {
"acc": 0.6495659037095501,
"acc_stderr": 0.013409047676670189
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=cerebras/btlm-3b-8k-base,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto","num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:4",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/facebook-opt-2.7b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.2687713310580205,
"acc_stderr": 0.012955065963710691,
"acc_norm": 0.31313993174061433,
"acc_norm_stderr": 0.013552671543623504
},
"arc_easy": {
"acc": 0.6077441077441077,
"acc_stderr": 0.010018744689650043,
"acc_norm": 0.5429292929292929,
"acc_norm_stderr": 0.01022189756425603
},
"boolq": {
"acc": 0.6033639143730887,
"acc_stderr": 0.008556148582032
},
"hellaswag": {
"acc": 0.4584744074885481,
"acc_stderr": 0.004972543127767873,
"acc_norm": 0.6059549890460068,
"acc_norm_stderr": 0.0048764594346198
},
"lambada_openai": {
"ppl": 5.119857738610855,
"ppl_stderr": 0.11991227443177162,
"acc": 0.6357461672811954,
"acc_stderr": 0.006704339729528894
},
"openbookqa": {
"acc": 0.25,
"acc_stderr": 0.019384310743640384,
"acc_norm": 0.352,
"acc_norm_stderr": 0.021380042385946048
},
"piqa": {
"acc": 0.7383025027203483,
"acc_stderr": 0.01025563077270823,
"acc_norm": 0.7480957562568009,
"acc_norm_stderr": 0.010128421335088685
},
"sciq": {
"acc": 0.858,
"acc_stderr": 0.011043457699378227,
"acc_norm": 0.79,
"acc_norm_stderr": 0.012886662332274536
},
"siqa": {
"acc": 0.4083930399181167,
"acc_stderr": 0.011122558066098069,
"acc_norm": 0.44268167860798363,
"acc_norm_stderr": 0.011239482425741961
},
"winogrande": {
"acc": 0.6101026045777427,
"acc_stderr": 0.013707547317008462
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=facebook/opt-2.7b,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:2",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/facebook-opt-6.7b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.30716723549488056,
"acc_stderr": 0.013481034054980945,
"acc_norm": 0.34812286689419797,
"acc_norm_stderr": 0.013921008595179338
},
"arc_easy": {
"acc": 0.6565656565656566,
"acc_stderr": 0.00974381736896,
"acc_norm": 0.6014309764309764,
"acc_norm_stderr": 0.010046455400477947
},
"boolq": {
"acc": 0.6602446483180429,
"acc_stderr": 0.00828377201314756
},
"hellaswag": {
"acc": 0.5051782513443537,
"acc_stderr": 0.004989513809408587,
"acc_norm": 0.6719776936865166,
"acc_norm_stderr": 0.004685334844038652
},
"lambada_openai": {
"ppl": 4.252533670173101,
"ppl_stderr": 0.0927155022353668,
"acc": 0.6764991267222977,
"acc_stderr": 0.006517535744360227
},
"openbookqa": {
"acc": 0.276,
"acc_stderr": 0.020011219298073524,
"acc_norm": 0.372,
"acc_norm_stderr": 0.0216371979857224
},
"piqa": {
"acc": 0.7633297062023939,
"acc_stderr": 0.009916841655042804,
"acc_norm": 0.766050054406964,
"acc_norm_stderr": 0.009877236895137437
},
"sciq": {
"acc": 0.901,
"acc_stderr": 0.009449248027662761,
"acc_norm": 0.853,
"acc_norm_stderr": 0.011203415395160333
},
"siqa": {
"acc": 0.4263050153531218,
"acc_stderr": 0.011190503463264742,
"acc_norm": 0.4595701125895599,
"acc_norm_stderr": 0.011277022486079959
},
"winogrande": {
"acc": 0.6535122336227308,
"acc_stderr": 0.013373773411685644
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=facebook/opt-6.7b,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:2",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/huggyllama-llama-7b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.4189419795221843,
"acc_stderr": 0.014418106953639013,
"acc_norm": 0.4462457337883959,
"acc_norm_stderr": 0.014526705548539982
},
"arc_easy": {
"acc": 0.7525252525252525,
"acc_stderr": 0.00885511441483471,
"acc_norm": 0.7285353535353535,
"acc_norm_stderr": 0.009125362970360627
},
"boolq": {
"acc": 0.7504587155963303,
"acc_stderr": 0.0075688020241860285
},
"hellaswag": {
"acc": 0.5696076478789086,
"acc_stderr": 0.004941191607317913,
"acc_norm": 0.7621987651862179,
"acc_norm_stderr": 0.004248666961833349
},
"lambada_openai": {
"ppl": 3.4882593690160553,
"ppl_stderr": 0.068517853205977,
"acc": 0.7354938870560839,
"acc_stderr": 0.006144965702579053
},
"openbookqa": {
"acc": 0.344,
"acc_stderr": 0.02126575803797874,
"acc_norm": 0.444,
"acc_norm_stderr": 0.02224224437573102
},
"piqa": {
"acc": 0.7867247007616975,
"acc_stderr": 0.00955712122586133,
"acc_norm": 0.7916213275299239,
"acc_norm_stderr": 0.009476125383049459
},
"sciq": {
"acc": 0.946,
"acc_stderr": 0.007150883521295433,
"acc_norm": 0.928,
"acc_norm_stderr": 0.008178195576218681
},
"siqa": {
"acc": 0.4483111566018424,
"acc_stderr": 0.01125345173122258,
"acc_norm": 0.4703172978505629,
"acc_norm_stderr": 0.011294116144908554
},
"winogrande": {
"acc": 0.6992896606156275,
"acc_stderr": 0.012888010494704723
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=False,pretrained=huggyllama/llama-7b,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:2",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/kittn_mistral-7B-v0.1-hf.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.5025597269624573,
"acc_stderr": 0.014611199329843777,
"acc_norm": 0.5435153583617748,
"acc_norm_stderr": 0.014555949760496439
},
"arc_easy": {
"acc": 0.8080808080808081,
"acc_stderr": 0.00808080808080796,
"acc_norm": 0.7954545454545454,
"acc_norm_stderr": 0.008276958800002997
},
"boolq": {
"acc": 0.8363914373088684,
"acc_stderr": 0.006469941343840763
},
"hellaswag": {
"acc": 0.6127265484963155,
"acc_stderr": 0.0048613146132868434,
"acc_norm": 0.8105954989046007,
"acc_norm_stderr": 0.003910288117015163
},
"lambada_openai": {
"ppl": 3.181388367482107,
"ppl_stderr": 0.058369197945258765,
"acc": 0.7568406753347564,
"acc_stderr": 0.0059766767751295085
},
"openbookqa": {
"acc": 0.328,
"acc_stderr": 0.021017027165175485,
"acc_norm": 0.44,
"acc_norm_stderr": 0.022221331534143057
},
"piqa": {
"acc": 0.8079434167573449,
"acc_stderr": 0.00919074029512649,
"acc_norm": 0.8215451577801959,
"acc_norm_stderr": 0.008933575463062072
},
"sciq": {
"acc": 0.959,
"acc_stderr": 0.006273624021118748,
"acc_norm": 0.938,
"acc_norm_stderr": 0.0076298239962803065
},
"siqa": {
"acc": 0.4273285568065507,
"acc_stderr": 0.01119393034055127,
"acc_norm": 0.4570112589559877,
"acc_norm_stderr": 0.01127217546233142
},
"winogrande": {
"acc": 0.7403314917127072,
"acc_stderr": 0.012322700705552673
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "pretrained=kittn/mistral-7B-v0.1-hf,dtype=bfloat16,trust_remote_code=True,low_cpu_mem_usage=True,use_fast=False",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/meta-llama-Llama-2-13b-hf.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.4863481228668942,
"acc_stderr": 0.014605943429860945,
"acc_norm": 0.492320819112628,
"acc_norm_stderr": 0.01460966744089257
},
"arc_easy": {
"acc": 0.79503367003367,
"acc_stderr": 0.008283277600626398,
"acc_norm": 0.7760942760942761,
"acc_norm_stderr": 0.008553779114531757
},
"boolq": {
"acc": 0.8051987767584098,
"acc_stderr": 0.006926916185348359
},
"hellaswag": {
"acc": 0.6011750647281418,
"acc_stderr": 0.004886559008754982,
"acc_norm": 0.7935670185222067,
"acc_norm_stderr": 0.004039176806180289
},
"lambada_openai": {
"ppl": 3.0445239988325836,
"ppl_stderr": 0.05613639700817276,
"acc": 0.767708131185717,
"acc_stderr": 0.005883383348944357
},
"openbookqa": {
"acc": 0.354,
"acc_stderr": 0.021407582047916447,
"acc_norm": 0.454,
"acc_norm_stderr": 0.02228814759117695
},
"piqa": {
"acc": 0.7905331882480957,
"acc_stderr": 0.0094943029798198,
"acc_norm": 0.8063112078346029,
"acc_norm_stderr": 0.009220384152336643
},
"sciq": {
"acc": 0.945,
"acc_stderr": 0.007212976294639239,
"acc_norm": 0.933,
"acc_norm_stderr": 0.007910345983177549
},
"siqa": {
"acc": 0.4278403275332651,
"acc_stderr": 0.01119562541819821,
"acc_norm": 0.4503582395087001,
"acc_norm_stderr": 0.01125816983012229
},
"winogrande": {
"acc": 0.7221783741120757,
"acc_stderr": 0.012588918183871603
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=False,pretrained=meta-llama/Llama-2-13b-hf,trust_remote_code=True,low_cpu_mem_usage=True,dtype=bfloat16",
"num_fewshot": 0,
"batch_size": "16",
"batch_sizes": [],
"device": "cuda:5",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/meta-llama-Llama-2-7b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.4300341296928328,
"acc_stderr": 0.014467631559137993,
"acc_norm": 0.4616040955631399,
"acc_norm_stderr": 0.014568245550296363
},
"arc_easy": {
"acc": 0.7626262626262627,
"acc_stderr": 0.008730525906362434,
"acc_norm": 0.7453703703703703,
"acc_norm_stderr": 0.008939407288589414
},
"boolq": {
"acc": 0.7773700305810397,
"acc_stderr": 0.007276093141006333
},
"hellaswag": {
"acc": 0.5720971917944633,
"acc_stderr": 0.00493763511283029,
"acc_norm": 0.7594104760007967,
"acc_norm_stderr": 0.004265678940698863
},
"lambada_openai": {
"ppl": 3.3970918836338027,
"ppl_stderr": 0.06684659102563836,
"acc": 0.7347176402095866,
"acc_stderr": 0.0061507275830540355
},
"openbookqa": {
"acc": 0.314,
"acc_stderr": 0.020776701920308997,
"acc_norm": 0.436,
"acc_norm_stderr": 0.0221989546414768
},
"piqa": {
"acc": 0.7774755168661589,
"acc_stderr": 0.009704600975718245,
"acc_norm": 0.7878128400435256,
"acc_norm_stderr": 0.009539299828174048
},
"sciq": {
"acc": 0.936,
"acc_stderr": 0.007743640226919308,
"acc_norm": 0.908,
"acc_norm_stderr": 0.009144376393151106
},
"siqa": {
"acc": 0.43500511770726713,
"acc_stderr": 0.011218074465506494,
"acc_norm": 0.47389969293756395,
"acc_norm_stderr": 0.011298645160980832
},
"truthfulqa_mc": {
"mc1": 0.2521419828641371,
"mc1_stderr": 0.01520152224629997,
"mc2": 0.38967559882659686,
"mc2_stderr": 0.01357922221561985
},
"winogrande": {
"acc": 0.6961325966850829,
"acc_stderr": 0.012926209475483574
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=False,pretrained=meta-llama/Llama-2-7b-hf,trust_remote_code=True,low_cpu_mem_usage=True,dtype=bfloat16",
"num_fewshot": 0,
"batch_size": "2",
"batch_sizes": [],
"device": "cuda:0",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/microsoft-phi-1_5.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.4445392491467577,
"acc_stderr": 0.014521226405627079,
"acc_norm": 0.4803754266211604,
"acc_norm_stderr": 0.014600132075947092
},
"arc_easy": {
"acc": 0.7613636363636364,
"acc_stderr": 0.008746465140706127,
"acc_norm": 0.7314814814814815,
"acc_norm_stderr": 0.009094042554994847
},
"boolq": {
"acc": 0.7452599388379205,
"acc_stderr": 0.007620703281690057
},
"hellaswag": {
"acc": 0.47988448516231824,
"acc_stderr": 0.004985741706385726,
"acc_norm": 0.6261700856403107,
"acc_norm_stderr": 0.004828305041904399
},
"lambada_openai": {
"ppl": 8.945202930351474,
"ppl_stderr": 0.3005661666926477,
"acc": 0.527459732194838,
"acc_stderr": 0.0069554645156210786
},
"openbookqa": {
"acc": 0.376,
"acc_stderr": 0.021683827539286115,
"acc_norm": 0.482,
"acc_norm_stderr": 0.02236856511738799
},
"piqa": {
"acc": 0.7633297062023939,
"acc_stderr": 0.009916841655042806,
"acc_norm": 0.7578890097932536,
"acc_norm_stderr": 0.009994371269104381
},
"sciq": {
"acc": 0.932,
"acc_stderr": 0.007964887911291603,
"acc_norm": 0.916,
"acc_norm_stderr": 0.00877616208949113
},
"siqa": {
"acc": 0.5537359263050153,
"acc_stderr": 0.01124854090154796,
"acc_norm": 0.5957011258955988,
"acc_norm_stderr": 0.0111048923983008
},
"winogrande": {
"acc": 0.7253354380426204,
"acc_stderr": 0.012544516005117187
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=False,pretrained=microsoft/phi-1_5,trust_remote_code=True,low_cpu_mem_usage=True,dtype=bfloat16",
"num_fewshot": 0,
"batch_size": "16",
"batch_sizes": [],
"device": "cuda:5",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/mosaicml-mpt-7b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.4052901023890785,
"acc_stderr": 0.01434686906022933,
"acc_norm": 0.4189419795221843,
"acc_norm_stderr": 0.014418106953639011
},
"arc_easy": {
"acc": 0.7491582491582491,
"acc_stderr": 0.008895183010487386,
"acc_norm": 0.7003367003367004,
"acc_norm_stderr": 0.009400228586205971
},
"boolq": {
"acc": 0.7394495412844037,
"acc_stderr": 0.007677021072511165
},
"hellaswag": {
"acc": 0.571400119498108,
"acc_stderr": 0.004938643787869543,
"acc_norm": 0.7617008564031069,
"acc_norm_stderr": 0.00425172316377217
},
"lambada_openai": {
"ppl": 3.8689102314884214,
"ppl_stderr": 0.0808940319922043,
"acc": 0.6863962740151368,
"acc_stderr": 0.006463833164285203
},
"openbookqa": {
"acc": 0.314,
"acc_stderr": 0.020776701920308997,
"acc_norm": 0.428,
"acc_norm_stderr": 0.022149790663861923
},
"piqa": {
"acc": 0.7889009793253536,
"acc_stderr": 0.00952137737873414,
"acc_norm": 0.8063112078346029,
"acc_norm_stderr": 0.009220384152336641
},
"sciq": {
"acc": 0.937,
"acc_stderr": 0.007687007876286428,
"acc_norm": 0.888,
"acc_norm_stderr": 0.009977753031397236
},
"siqa": {
"acc": 0.45138178096212894,
"acc_stderr": 0.01126045668162444,
"acc_norm": 0.48311156601842375,
"acc_norm_stderr": 0.011307614732827416
},
"truthfulqa_mc": {
"mc1": 0.20930232558139536,
"mc1_stderr": 0.014241219434785823,
"mc2": 0.3348523259251629,
"mc2_stderr": 0.01313621094524683
},
"winogrande": {
"acc": 0.6803472770323599,
"acc_stderr": 0.01310652851766514
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=mosaicml/mpt-7b,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:2",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/openlm-research-open_llama_13b
================================================
{
"results": {
"lambada_openai": {
"ppl": 3.569586285393247,
"ppl_stderr": 0.07265272740503036,
"acc": 0.7209392586842616,
"acc_stderr": 0.006249003708978234
},
"sciq": {
"acc": 0.941,
"acc_stderr": 0.007454835650406722,
"acc_norm": 0.914,
"acc_norm_stderr": 0.008870325962594766
}
},
"versions": {
"lambada_openai": 0,
"sciq": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=False,pretrained=openlm-research/open_llama_13b,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "2",
"batch_sizes": [],
"device": "cuda:3",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/openlm-research-open_llama_3b_v2.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.3387372013651877,
"acc_stderr": 0.013830568927974332,
"acc_norm": 0.3609215017064846,
"acc_norm_stderr": 0.014034761386175458
},
"arc_easy": {
"acc": 0.6759259259259259,
"acc_stderr": 0.00960372885009539,
"acc_norm": 0.63510101010101,
"acc_norm_stderr": 0.009878157021155649
},
"boolq": {
"acc": 0.6568807339449542,
"acc_stderr": 0.008303445777655941
},
"hellaswag": {
"acc": 0.5223063134833699,
"acc_stderr": 0.0049848133910162145,
"acc_norm": 0.6998605855407289,
"acc_norm_stderr": 0.004573817163007456
},
"lambada_openai": {
"ppl": 4.565625743504039,
"ppl_stderr": 0.1034965673734545,
"acc": 0.6673782262759558,
"acc_stderr": 0.006564073374961233
},
"openbookqa": {
"acc": 0.26,
"acc_stderr": 0.019635965529725512,
"acc_norm": 0.376,
"acc_norm_stderr": 0.021683827539286122
},
"piqa": {
"acc": 0.7665941240478781,
"acc_stderr": 0.009869247889521001,
"acc_norm": 0.778563656147987,
"acc_norm_stderr": 0.009687616456840284
},
"sciq": {
"acc": 0.924,
"acc_stderr": 0.008384169266796386,
"acc_norm": 0.878,
"acc_norm_stderr": 0.010354864712936698
},
"siqa": {
"acc": 0.4119754350051177,
"acc_stderr": 0.011137360400975268,
"acc_norm": 0.4524053224155578,
"acc_norm_stderr": 0.011262695440459566
},
"truthfulqa_mc": {
"mc1": 0.21297429620563035,
"mc1_stderr": 0.014332203787059685,
"mc2": 0.3458747299959986,
"mc2_stderr": 0.013215129281312441
},
"winogrande": {
"acc": 0.6290449881610103,
"acc_stderr": 0.01357639990223157
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=False,pretrained=openlm-research/open_llama_3b_v2,trust_remote_code=True,low_cpu_mem_usage=True,dtype=bfloat16",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:4",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/openlm-research-open_llama_7b_v2.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.38822525597269625,
"acc_stderr": 0.014241614207414034,
"acc_norm": 0.42406143344709896,
"acc_norm_stderr": 0.014441889627464392
},
"arc_easy": {
"acc": 0.7192760942760943,
"acc_stderr": 0.009220526174711363,
"acc_norm": 0.6965488215488216,
"acc_norm_stderr": 0.009433837434252279
},
"boolq": {
"acc": 0.7140672782874617,
"acc_stderr": 0.00790303735916362
},
"hellaswag": {
"acc": 0.5569607647878908,
"acc_stderr": 0.004957296691391575,
"acc_norm": 0.7464648476399124,
"acc_norm_stderr": 0.004341454841892331
},
"lambada_openai": {
"ppl": 3.8265934555236463,
"ppl_stderr": 0.07878439404908999,
"acc": 0.7104599262565496,
"acc_stderr": 0.006318823234213216
},
"openbookqa": {
"acc": 0.302,
"acc_stderr": 0.020553269174209184,
"acc_norm": 0.402,
"acc_norm_stderr": 0.021948929609938606
},
"piqa": {
"acc": 0.7916213275299239,
"acc_stderr": 0.009476125383049447,
"acc_norm": 0.8030467899891186,
"acc_norm_stderr": 0.009278918898006383
},
"sciq": {
"acc": 0.938,
"acc_stderr": 0.007629823996280304,
"acc_norm": 0.901,
"acc_norm_stderr": 0.009449248027662747
},
"siqa": {
"acc": 0.4196519959058342,
"acc_stderr": 0.011167032303390547,
"acc_norm": 0.4600818833162743,
"acc_norm_stderr": 0.011277955967920398
},
"truthfulqa_mc": {
"mc1": 0.22643818849449204,
"mc1_stderr": 0.014651337324602576,
"mc2": 0.3456877328963021,
"mc2_stderr": 0.013482248222806824
},
"winogrande": {
"acc": 0.6582478295185478,
"acc_stderr": 0.013330103018622861
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=False,pretrained=openlm-research/open_llama_7b_v2,trust_remote_code=True,low_cpu_mem_usage=True,dtype=bfloat16",
"num_fewshot": 0,
"batch_size": "2",
"batch_sizes": [],
"device": "cuda:0",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/tiiuae_falcon-7b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.40273037542662116,
"acc_stderr": 0.014332236306790149,
"acc_norm": 0.43686006825938567,
"acc_norm_stderr": 0.014494421584256519
},
"arc_easy": {
"acc": 0.7441077441077442,
"acc_stderr": 0.008953950243013991,
"acc_norm": 0.7079124579124579,
"acc_norm_stderr": 0.00933070561656907
},
"boolq": {
"acc": 0.735474006116208,
"acc_stderr": 0.007714546144910642
},
"hellaswag": {
"acc": 0.5771758613821948,
"acc_stderr": 0.004929983692795067,
"acc_norm": 0.7634933280223063,
"acc_norm_stderr": 0.004240683281093403
},
"lambada_openai": {
"ppl": 3.3698354504387775,
"ppl_stderr": 0.06490807599596318,
"acc": 0.7455850960605472,
"acc_stderr": 0.006067809764031527
},
"openbookqa": {
"acc": 0.306,
"acc_stderr": 0.020629569998345403,
"acc_norm": 0.442,
"acc_norm_stderr": 0.02223197069632112
},
"piqa": {
"acc": 0.794885745375408,
"acc_stderr": 0.009420971671017915,
"acc_norm": 0.8057671381936888,
"acc_norm_stderr": 0.009230209366168272
},
"sciq": {
"acc": 0.94,
"acc_stderr": 0.00751375115747492,
"acc_norm": 0.915,
"acc_norm_stderr": 0.008823426366942328
},
"siqa": {
"acc": 0.42067553735926305,
"acc_stderr": 0.011170778517705619,
"acc_norm": 0.4600818833162743,
"acc_norm_stderr": 0.011277955967920396
},
"winogrande": {
"acc": 0.6724546172059984,
"acc_stderr": 0.013190169546797016
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "pretrained=tiiuae/falcon-7b,dtype=bfloat16,trust_remote_code=True,low_cpu_mem_usage=True,use_fast=True",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/external/togethercomputer-RedPajama-INCITE-7B-Base2.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.3771331058020478,
"acc_stderr": 0.014163366896192593,
"acc_norm": 0.39419795221843,
"acc_norm_stderr": 0.014280522667467323
},
"arc_easy": {
"acc": 0.7234848484848485,
"acc_stderr": 0.00917788010146828,
"acc_norm": 0.6919191919191919,
"acc_norm_stderr": 0.009473887075826333
},
"boolq": {
"acc": 0.7076452599388379,
"acc_stderr": 0.00795527890990574
},
"hellaswag": {
"acc": 0.5256920932085242,
"acc_stderr": 0.004983189711208505,
"acc_norm": 0.7033459470225055,
"acc_norm_stderr": 0.004558491550673688
},
"lambada_openai": {
"ppl": 3.9176883876529036,
"ppl_stderr": 0.08360630042095321,
"acc": 0.713370851930914,
"acc_stderr": 0.006299845944000654
},
"openbookqa": {
"acc": 0.29,
"acc_stderr": 0.020313179231745186,
"acc_norm": 0.406,
"acc_norm_stderr": 0.021983962090086333
},
"piqa": {
"acc": 0.7714907508161044,
"acc_stderr": 0.009796313511829524,
"acc_norm": 0.7736670293797606,
"acc_norm_stderr": 0.009763294246879415
},
"sciq": {
"acc": 0.927,
"acc_stderr": 0.008230354715244055,
"acc_norm": 0.897,
"acc_norm_stderr": 0.009616833339695796
},
"siqa": {
"acc": 0.4257932446264074,
"acc_stderr": 0.011188771652377858,
"acc_norm": 0.44882292732855683,
"acc_norm_stderr": 0.011254649314820132
},
"truthfulqa_mc": {
"mc1": 0.23011015911872704,
"mc1_stderr": 0.014734557959807769,
"mc2": 0.3301333660822527,
"mc2_stderr": 0.012995959559501016
},
"winogrande": {
"acc": 0.6432517758484609,
"acc_stderr": 0.013463393958028733
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=togethercomputer/RedPajama-INCITE-7B-Base,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:2",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/open_llm_leaderboard/stablelm-beta-3b-v2-arc-challenge.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.36689419795221845,
"acc_norm": 0.3967576791808874
}
},
"versions": {
"arc_challenge": 0
},
"config": {
"model": "stabilityai/stablelm-base-alpha-3b-v2",
"num_fewshot": 25,
"batch_size": 8,
"device": "cuda:0",
"no_cache": true,
"limit": null,
"bootstrap_iters": 10000,
"description_dict": null
}
}
================================================
FILE: evals/open_llm_leaderboard/stablelm-beta-3b-v2-hellaswag.json
================================================
{
"results": {
"hellaswag": {
"acc": 0.5239992033459471,
"acc_norm": 0.7066321449910377
}
},
"versions": {
"hellaswag": 0
},
"config": {
"model": "stabilityai/stablelm-base-alpha-3b-v2",
"num_fewshot": 10,
"batch_size": 8,
"device": "cuda:0",
"no_cache": true,
"limit": null,
"bootstrap_iters": 10000,
"description_dict": null
}
}
================================================
FILE: evals/open_llm_leaderboard/stablelm-beta-3b-v2-mmmlu.json
================================================
{
"results": {
"hendrycksTest-professional_psychology": {
"acc": 0.2826797385620915,
"acc_norm": 0.27124183006535946
},
"hendrycksTest-abstract_algebra": {
"acc": 0.26,
"acc_norm": 0.29
},
"hendrycksTest-prehistory": {
"acc": 0.3271604938271605,
"acc_norm": 0.26851851851851855
},
"hendrycksTest-international_law": {
"acc": 0.23140495867768596,
"acc_norm": 0.36363636363636365
},
"hendrycksTest-logical_fallacies": {
"acc": 0.20245398773006135,
"acc_norm": 0.22085889570552147
},
"hendrycksTest-professional_medicine": {
"acc": 0.30514705882352944,
"acc_norm": 0.33455882352941174
},
"hendrycksTest-high_school_european_history": {
"acc": 0.3151515151515151,
"acc_norm": 0.296969696969697
},
"hendrycksTest-high_school_physics": {
"acc": 0.2251655629139073,
"acc_norm": 0.2582781456953642
},
"hendrycksTest-management": {
"acc": 0.30097087378640774,
"acc_norm": 0.30097087378640774
},
"hendrycksTest-college_mathematics": {
"acc": 0.17,
"acc_norm": 0.24
},
"hendrycksTest-college_computer_science": {
"acc": 0.29,
"acc_norm": 0.24
},
"hendrycksTest-human_sexuality": {
"acc": 0.40458015267175573,
"acc_norm": 0.3893129770992366
},
"hendrycksTest-college_biology": {
"acc": 0.2708333333333333,
"acc_norm": 0.2569444444444444
},
"hendrycksTest-high_school_computer_science": {
"acc": 0.34,
"acc_norm": 0.3
},
"hendrycksTest-high_school_psychology": {
"acc": 0.3467889908256881,
"acc_norm": 0.326605504587156
},
"hendrycksTest-high_school_chemistry": {
"acc": 0.2413793103448276,
"acc_norm": 0.28078817733990147
},
"hendrycksTest-astronomy": {
"acc": 0.3355263157894737,
"acc_norm": 0.3355263157894737
},
"hendrycksTest-medical_genetics": {
"acc": 0.39,
"acc_norm": 0.4
},
"hendrycksTest-nutrition": {
"acc": 0.35947712418300654,
"acc_norm": 0.3758169934640523
},
"hendrycksTest-moral_disputes": {
"acc": 0.3554913294797688,
"acc_norm": 0.33815028901734107
},
"hendrycksTest-computer_security": {
"acc": 0.38,
"acc_norm": 0.39
},
"hendrycksTest-anatomy": {
"acc": 0.32592592592592595,
"acc_norm": 0.2814814814814815
},
"hendrycksTest-formal_logic": {
"acc": 0.30952380952380953,
"acc_norm": 0.3253968253968254
},
"hendrycksTest-high_school_us_history": {
"acc": 0.3088235294117647,
"acc_norm": 0.29901960784313725
},
"hendrycksTest-security_studies": {
"acc": 0.3224489795918367,
"acc_norm": 0.23265306122448978
},
"hendrycksTest-high_school_mathematics": {
"acc": 0.23703703703703705,
"acc_norm": 0.2814814814814815
},
"hendrycksTest-high_school_macroeconomics": {
"acc": 0.2794871794871795,
"acc_norm": 0.2794871794871795
},
"hendrycksTest-clinical_knowledge": {
"acc": 0.26037735849056604,
"acc_norm": 0.32075471698113206
},
"hendrycksTest-us_foreign_policy": {
"acc": 0.39,
"acc_norm": 0.41
},
"hendrycksTest-virology": {
"acc": 0.35542168674698793,
"acc_norm": 0.3493975903614458
},
"hendrycksTest-public_relations": {
"acc": 0.34545454545454546,
"acc_norm": 0.3
},
"hendrycksTest-world_religions": {
"acc": 0.4093567251461988,
"acc_norm": 0.4619883040935672
},
"hendrycksTest-college_physics": {
"acc": 0.3137254901960784,
"acc_norm": 0.3235294117647059
},
"hendrycksTest-high_school_biology": {
"acc": 0.267741935483871,
"acc_norm": 0.2870967741935484
},
"hendrycksTest-business_ethics": {
"acc": 0.33,
"acc_norm": 0.28
},
"hendrycksTest-high_school_government_and_politics": {
"acc": 0.30569948186528495,
"acc_norm": 0.2849740932642487
},
"hendrycksTest-high_school_world_history": {
"acc": 0.2742616033755274,
"acc_norm": 0.27848101265822783
},
"hendrycksTest-jurisprudence": {
"acc": 0.23148148148148148,
"acc_norm": 0.24074074074074073
},
"hendrycksTest-miscellaneous": {
"acc": 0.4227330779054917,
"acc_norm": 0.41762452107279696
},
"hendrycksTest-marketing": {
"acc": 0.3974358974358974,
"acc_norm": 0.39316239316239315
},
"hendrycksTest-high_school_microeconomics": {
"acc": 0.29411764705882354,
"acc_norm": 0.3445378151260504
},
"hendrycksTest-econometrics": {
"acc": 0.2543859649122807,
"acc_norm": 0.24561403508771928
},
"hendrycksTest-conceptual_physics": {
"acc": 0.3276595744680851,
"acc_norm": 0.28936170212765955
},
"hendrycksTest-high_school_statistics": {
"acc": 0.3287037037037037,
"acc_norm": 0.33796296296296297
},
"hendrycksTest-sociology": {
"acc": 0.3333333333333333,
"acc_norm": 0.31343283582089554
},
"hendrycksTest-electrical_engineering": {
"acc": 0.2689655172413793,
"acc_norm": 0.25517241379310346
},
"hendrycksTest-elementary_mathematics": {
"acc": 0.335978835978836,
"acc_norm": 0.328042328042328
},
"hendrycksTest-high_school_geography": {
"acc": 0.3181818181818182,
"acc_norm": 0.3333333333333333
},
"hendrycksTest-philosophy": {
"acc": 0.31511254019292606,
"acc_norm": 0.3247588424437299
},
"hendrycksTest-moral_scenarios": {
"acc": 0.26927374301675977,
"acc_norm": 0.27150837988826815
},
"hendrycksTest-college_chemistry": {
"acc": 0.34,
"acc_norm": 0.36
},
"hendrycksTest-machine_learning": {
"acc": 0.2767857142857143,
"acc_norm": 0.2767857142857143
},
"hendrycksTest-professional_accounting": {
"acc": 0.29432624113475175,
"acc_norm": 0.3049645390070922
},
"hendrycksTest-professional_law": {
"acc": 0.25488917861799215,
"acc_norm": 0.258148631029987
},
"hendrycksTest-college_medicine": {
"acc": 0.35260115606936415,
"acc_norm": 0.3468208092485549
},
"hendrycksTest-global_facts": {
"acc": 0.26,
"acc_norm": 0.21
},
"hendrycksTest-human_aging": {
"acc": 0.28699551569506726,
"acc_norm": 0.2556053811659193
}
},
"versions": {
"hendrycksTest-professional_psychology": 0,
"hendrycksTest-abstract_algebra": 0,
"hendrycksTest-prehistory": 0,
"hendrycksTest-international_law": 0,
"hendrycksTest-logical_fallacies": 0,
"hendrycksTest-professional_medicine": 0,
"hendrycksTest-high_school_european_history": 0,
"hendrycksTest-high_school_physics": 0,
"hendrycksTest-management": 0,
"hendrycksTest-college_mathematics": 0,
"hendrycksTest-college_computer_science": 0,
"hendrycksTest-human_sexuality": 0,
"hendrycksTest-college_biology": 0,
"hendrycksTest-high_school_computer_science": 0,
"hendrycksTest-high_school_psychology": 0,
"hendrycksTest-high_school_chemistry": 0,
"hendrycksTest-astronomy": 0,
"hendrycksTest-medical_genetics": 0,
"hendrycksTest-nutrition": 0,
"hendrycksTest-moral_disputes": 0,
"hendrycksTest-computer_security": 0,
"hendrycksTest-anatomy": 0,
"hendrycksTest-formal_logic": 0,
"hendrycksTest-high_school_us_history": 0,
"hendrycksTest-security_studies": 0,
"hendrycksTest-high_school_mathematics": 0,
"hendrycksTest-high_school_macroeconomics": 0,
"hendrycksTest-clinical_knowledge": 0,
"hendrycksTest-us_foreign_policy": 0,
"hendrycksTest-virology": 0,
"hendrycksTest-public_relations": 0,
"hendrycksTest-world_religions": 0,
"hendrycksTest-college_physics": 0,
"hendrycksTest-high_school_biology": 0,
"hendrycksTest-business_ethics": 0,
"hendrycksTest-high_school_government_and_politics": 0,
"hendrycksTest-high_school_world_history": 0,
"hendrycksTest-jurisprudence": 0,
"hendrycksTest-miscellaneous": 0,
"hendrycksTest-marketing": 0,
"hendrycksTest-high_school_microeconomics": 0,
"hendrycksTest-econometrics": 0,
"hendrycksTest-conceptual_physics": 0,
"hendrycksTest-high_school_statistics": 0,
"hendrycksTest-sociology": 0,
"hendrycksTest-electrical_engineering": 0,
"hendrycksTest-elementary_mathematics": 0,
"hendrycksTest-high_school_geography": 0,
"hendrycksTest-philosophy": 0,
"hendrycksTest-moral_scenarios": 0,
"hendrycksTest-college_chemistry": 0,
"hendrycksTest-machine_learning": 0,
"hendrycksTest-professional_accounting": 0,
"hendrycksTest-professional_law": 0,
"hendrycksTest-college_medicine": 0,
"hendrycksTest-global_facts": 0,
"hendrycksTest-human_aging": 0
},
"config": {
"model": "stabilityai/stablelm-base-alpha-3b-v2",
"num_fewshot": 5,
"batch_size": 8,
"device": "cuda:0",
"no_cache": true,
"limit": null,
"bootstrap_iters": 10000,
"description_dict": null
}
}
================================================
FILE: evals/open_llm_leaderboard/stablelm-beta-3b-v2-truthfulqa_mc.json
================================================
{
"results": {
"truthfulqa_mc": {
"mc1": 0.22399020807833536,
"mc2": 0.35873137294835583
}
},
"versions": {
"truthfulqa_mc": 1
},
"config": {
"model": "stabilityai/stablelm-base-alpha-3b-v2",
"num_fewshot": 0,
"batch_size": 8,
"device": "cuda:0",
"no_cache": true,
"limit": null,
"bootstrap_iters": 10000,
"description_dict": null
}
}
================================================
FILE: evals/open_llm_leaderboard/stablelm-beta-7b-v2-arc-challenge.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.431740614334471,
"acc_norm": 0.4726962457337884
}
},
"versions": {
"arc_challenge": 0
},
"config": {
"model": "stabilityai/stablelm-base-alpha-7b-v2",
"num_fewshot": 25,
"batch_size": 8,
"device": "cuda:0",
"no_cache": true,
"limit": null,
"bootstrap_iters": 10000,
"description_dict": null
}
}
================================================
FILE: evals/open_llm_leaderboard/stablelm-beta-7b-v2-hellaswag.json
================================================
{
"results": {
"hellaswag": {
"acc": 0.5721967735510854,
"acc_norm": 0.7706632144991038
}
},
"versions": {
"hellaswag": 0
},
"config": {
"model": "stabilityai/stablelm-base-alpha-7b-v2",
"num_fewshot": 10,
"batch_size": 8,
"device": "cuda:0",
"no_cache": true,
"limit": null,
"bootstrap_iters": 10000,
"description_dict": null
}
}
================================================
FILE: evals/open_llm_leaderboard/stablelm-beta-7b-v2-mmmlu.json
================================================
{
"results": {
"hendrycksTest-high_school_european_history": {
"acc": 0.48484848484848486,
"acc_norm": 0.47878787878787876
},
"hendrycksTest-college_chemistry": {
"acc": 0.29,
"acc_norm": 0.28
},
"hendrycksTest-international_law": {
"acc": 0.38016528925619836,
"acc_norm": 0.4214876033057851
},
"hendrycksTest-high_school_macroeconomics": {
"acc": 0.36153846153846153,
"acc_norm": 0.35128205128205126
},
"hendrycksTest-miscellaneous": {
"acc": 0.6513409961685823,
"acc_norm": 0.644955300127714
},
"hendrycksTest-professional_law": {
"acc": 0.31877444589308995,
"acc_norm": 0.3116036505867014
},
"hendrycksTest-medical_genetics": {
"acc": 0.51,
"acc_norm": 0.55
},
"hendrycksTest-high_school_world_history": {
"acc": 0.3881856540084388,
"acc_norm": 0.4050632911392405
},
"hendrycksTest-professional_medicine": {
"acc": 0.4742647058823529,
"acc_norm": 0.44485294117647056
},
"hendrycksTest-moral_disputes": {
"acc": 0.42196531791907516,
"acc_norm": 0.430635838150289
},
"hendrycksTest-high_school_geography": {
"acc": 0.5404040404040404,
"acc_norm": 0.5454545454545454
},
"hendrycksTest-high_school_microeconomics": {
"acc": 0.42857142857142855,
"acc_norm": 0.42436974789915966
},
"hendrycksTest-machine_learning": {
"acc": 0.32142857142857145,
"acc_norm": 0.32142857142857145
},
"hendrycksTest-security_studies": {
"acc": 0.3510204081632653,
"acc_norm": 0.2612244897959184
},
"hendrycksTest-world_religions": {
"acc": 0.6608187134502924,
"acc_norm": 0.695906432748538
},
"hendrycksTest-conceptual_physics": {
"acc": 0.40425531914893614,
"acc_norm": 0.3659574468085106
},
"hendrycksTest-high_school_physics": {
"acc": 0.2582781456953642,
"acc_norm": 0.2980132450331126
},
"hendrycksTest-nutrition": {
"acc": 0.43137254901960786,
"acc_norm": 0.45098039215686275
},
"hendrycksTest-high_school_psychology": {
"acc": 0.581651376146789,
"acc_norm": 0.5614678899082569
},
"hendrycksTest-professional_accounting": {
"acc": 0.2907801418439716,
"acc_norm": 0.30141843971631205
},
"hendrycksTest-human_aging": {
"acc": 0.4304932735426009,
"acc_norm": 0.39461883408071746
},
"hendrycksTest-college_physics": {
"acc": 0.35294117647058826,
"acc_norm": 0.3627450980392157
},
"hendrycksTest-high_school_chemistry": {
"acc": 0.28078817733990147,
"acc_norm": 0.3054187192118227
},
"hendrycksTest-high_school_biology": {
"acc": 0.4290322580645161,
"acc_norm": 0.43548387096774194
},
"hendrycksTest-us_foreign_policy": {
"acc": 0.57,
"acc_norm": 0.53
},
"hendrycksTest-philosophy": {
"acc": 0.4758842443729904,
"acc_norm": 0.4662379421221865
},
"hendrycksTest-logical_fallacies": {
"acc": 0.4110429447852761,
"acc_norm": 0.4294478527607362
},
"hendrycksTest-anatomy": {
"acc": 0.48148148148148145,
"acc_norm": 0.4888888888888889
},
"hendrycksTest-jurisprudence": {
"acc": 0.4351851851851852,
"acc_norm": 0.5
},
"hendrycksTest-high_school_computer_science": {
"acc": 0.41,
"acc_norm": 0.46
},
"hendrycksTest-elementary_mathematics": {
"acc": 0.32275132275132273,
"acc_norm": 0.32275132275132273
},
"hendrycksTest-abstract_algebra": {
"acc": 0.27,
"acc_norm": 0.28
},
"hendrycksTest-prehistory": {
"acc": 0.44753086419753085,
"acc_norm": 0.44135802469135804
},
"hendrycksTest-moral_scenarios": {
"acc": 0.2581005586592179,
"acc_norm": 0.26033519553072626
},
"hendrycksTest-college_medicine": {
"acc": 0.4161849710982659,
"acc_norm": 0.41040462427745666
},
"hendrycksTest-econometrics": {
"acc": 0.2982456140350877,
"acc_norm": 0.2807017543859649
},
"hendrycksTest-human_sexuality": {
"acc": 0.5038167938931297,
"acc_norm": 0.4961832061068702
},
"hendrycksTest-management": {
"acc": 0.5242718446601942,
"acc_norm": 0.5339805825242718
},
"hendrycksTest-computer_security": {
"acc": 0.57,
"acc_norm": 0.54
},
"hendrycksTest-college_computer_science": {
"acc": 0.38,
"acc_norm": 0.34
},
"hendrycksTest-public_relations": {
"acc": 0.5636363636363636,
"acc_norm": 0.5454545454545454
},
"hendrycksTest-sociology": {
"acc": 0.5522388059701493,
"acc_norm": 0.5124378109452736
},
"hendrycksTest-global_facts": {
"acc": 0.29,
"acc_norm": 0.28
},
"hendrycksTest-astronomy": {
"acc": 0.42105263157894735,
"acc_norm": 0.42105263157894735
},
"hendrycksTest-high_school_statistics": {
"acc": 0.27314814814814814,
"acc_norm": 0.2916666666666667
},
"hendrycksTest-professional_psychology": {
"acc": 0.40522875816993464,
"acc_norm": 0.39052287581699346
},
"hendrycksTest-high_school_us_history": {
"acc": 0.37745098039215685,
"acc_norm": 0.39705882352941174
},
"hendrycksTest-business_ethics": {
"acc": 0.42,
"acc_norm": 0.44
},
"hendrycksTest-clinical_knowledge": {
"acc": 0.42641509433962266,
"acc_norm": 0.4226415094339623
},
"hendrycksTest-college_biology": {
"acc": 0.4236111111111111,
"acc_norm": 0.4166666666666667
},
"hendrycksTest-formal_logic": {
"acc": 0.42063492063492064,
"acc_norm": 0.40476190476190477
},
"hendrycksTest-marketing": {
"acc": 0.6965811965811965,
"acc_norm": 0.6837606837606838
},
"hendrycksTest-college_mathematics": {
"acc": 0.23,
"acc_norm": 0.29
},
"hendrycksTest-electrical_engineering": {
"acc": 0.4827586206896552,
"acc_norm": 0.4
},
"hendrycksTest-high_school_government_and_politics": {
"acc": 0.5284974093264249,
"acc_norm": 0.5492227979274611
},
"hendrycksTest-virology": {
"acc": 0.37349397590361444,
"acc_norm": 0.39759036144578314
},
"hendrycksTest-high_school_mathematics": {
"acc": 0.2518518518518518,
"acc_norm": 0.28888888888888886
}
},
"versions": {
"hendrycksTest-high_school_european_history": 0,
"hendrycksTest-college_chemistry": 0,
"hendrycksTest-international_law": 0,
"hendrycksTest-high_school_macroeconomics": 0,
"hendrycksTest-miscellaneous": 0,
"hendrycksTest-professional_law": 0,
"hendrycksTest-medical_genetics": 0,
"hendrycksTest-high_school_world_history": 0,
"hendrycksTest-professional_medicine": 0,
"hendrycksTest-moral_disputes": 0,
"hendrycksTest-high_school_geography": 0,
"hendrycksTest-high_school_microeconomics": 0,
"hendrycksTest-machine_learning": 0,
"hendrycksTest-security_studies": 0,
"hendrycksTest-world_religions": 0,
"hendrycksTest-conceptual_physics": 0,
"hendrycksTest-high_school_physics": 0,
"hendrycksTest-nutrition": 0,
"hendrycksTest-high_school_psychology": 0,
"hendrycksTest-professional_accounting": 0,
"hendrycksTest-human_aging": 0,
"hendrycksTest-college_physics": 0,
"hendrycksTest-high_school_chemistry": 0,
"hendrycksTest-high_school_biology": 0,
"hendrycksTest-us_foreign_policy": 0,
"hendrycksTest-philosophy": 0,
"hendrycksTest-logical_fallacies": 0,
"hendrycksTest-anatomy": 0,
"hendrycksTest-jurisprudence": 0,
"hendrycksTest-high_school_computer_science": 0,
"hendrycksTest-elementary_mathematics": 0,
"hendrycksTest-abstract_algebra": 0,
"hendrycksTest-prehistory": 0,
"hendrycksTest-moral_scenarios": 0,
"hendrycksTest-college_medicine": 0,
"hendrycksTest-econometrics": 0,
"hendrycksTest-human_sexuality": 0,
"hendrycksTest-management": 0,
"hendrycksTest-computer_security": 0,
"hendrycksTest-college_computer_science": 0,
"hendrycksTest-public_relations": 0,
"hendrycksTest-sociology": 0,
"hendrycksTest-global_facts": 0,
"hendrycksTest-astronomy": 0,
"hendrycksTest-high_school_statistics": 0,
"hendrycksTest-professional_psychology": 0,
"hendrycksTest-high_school_us_history": 0,
"hendrycksTest-business_ethics": 0,
"hendrycksTest-clinical_knowledge": 0,
"hendrycksTest-college_biology": 0,
"hendrycksTest-formal_logic": 0,
"hendrycksTest-marketing": 0,
"hendrycksTest-college_mathematics": 0,
"hendrycksTest-electrical_engineering": 0,
"hendrycksTest-high_school_government_and_politics": 0,
"hendrycksTest-virology": 0,
"hendrycksTest-high_school_mathematics": 0
},
"config": {
"model": "stabilityai/stablelm-base-alpha-7b-v2",
"num_fewshot": 5,
"batch_size": 8,
"device": "cuda:0",
"no_cache": true,
"limit": null,
"bootstrap_iters": 10000,
"description_dict": null
}
}
================================================
FILE: evals/open_llm_leaderboard/stablelm-beta-7b-v2-truthfulqa_mc.json
================================================
{
"results": {
"truthfulqa_mc": {
"mc1": 0.2423500611995104,
"mc2": 0.3645856452722409
}
},
"versions": {
"truthfulqa_mc": 1
},
"config": {
"model": "stabilityai/stablelm-base-alpha-7b-v2",
"num_fewshot": 0,
"batch_size": 8,
"device": "cuda:0",
"no_cache": true,
"limit": null,
"bootstrap_iters": 10000,
"description_dict": null
}
}
================================================
FILE: evals/stablelm-3b-4e1t.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.3779863481228669,
"acc_stderr": 0.014169664520303103,
"acc_norm": 0.40017064846416384,
"acc_norm_stderr": 0.014317197787809169
},
"arc_easy": {
"acc": 0.7247474747474747,
"acc_stderr": 0.009164888895174743,
"acc_norm": 0.6771885521885522,
"acc_norm_stderr": 0.009593950220366743
},
"boolq": {
"acc": 0.7562691131498471,
"acc_stderr": 0.007509067459407977
},
"hellaswag": {
"acc": 0.5482971519617607,
"acc_stderr": 0.004966448380104203,
"acc_norm": 0.7389962158932484,
"acc_norm_stderr": 0.004382844128643414
},
"lambada_openai": {
"ppl": 3.827229437157901,
"ppl_stderr": 0.07919367133146168,
"acc": 0.7063846303124394,
"acc_stderr": 0.006344860619678723
},
"openbookqa": {
"acc": 0.314,
"acc_stderr": 0.020776701920308997,
"acc_norm": 0.398,
"acc_norm_stderr": 0.02191237788577997
},
"piqa": {
"acc": 0.7921653971708379,
"acc_stderr": 0.009466997964536423,
"acc_norm": 0.7976060935799782,
"acc_norm_stderr": 0.00937428968280767
},
"sciq": {
"acc": 0.948,
"acc_stderr": 0.007024624213817142,
"acc_norm": 0.919,
"acc_norm_stderr": 0.00863212103213998
},
"siqa": {
"acc": 0.41914022517911975,
"acc_stderr": 0.011165140708170328,
"acc_norm": 0.44319344933469806,
"acc_norm_stderr": 0.011240812731564952
},
"winogrande": {
"acc": 0.665351223362273,
"acc_stderr": 0.013261823629558375
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "pretrained=stablelm-3b-4e1t,dtype=bfloat16,trust_remote_code=True,low_cpu_mem_usage=True,use_fast=True",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/stablelm-base-alpha-3b-v2.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.3242320819112628,
"acc_stderr": 0.013678810399518815,
"acc_norm": 0.3506825938566553,
"acc_norm_stderr": 0.013944635930726085
},
"arc_easy": {
"acc": 0.6725589225589226,
"acc_stderr": 0.009629415859100609,
"acc_norm": 0.6325757575757576,
"acc_norm_stderr": 0.009892552616211548
},
"boolq": {
"acc": 0.645565749235474,
"acc_stderr": 0.008366245832688784
},
"hellaswag": {
"acc": 0.5105556662019518,
"acc_stderr": 0.004988669343786956,
"acc_norm": 0.6858195578570006,
"acc_norm_stderr": 0.004632399677490817
},
"lambada_openai": {
"ppl": 3.995184075578421,
"ppl_stderr": 0.08502310534019014,
"acc": 0.7025033960799534,
"acc_stderr": 0.006369088639380684
},
"openbookqa": {
"acc": 0.264,
"acc_stderr": 0.019732885585922098,
"acc_norm": 0.386,
"acc_norm_stderr": 0.021793529219281165
},
"piqa": {
"acc": 0.7600652883569097,
"acc_stderr": 0.009963625892809545,
"acc_norm": 0.780195865070729,
"acc_norm_stderr": 0.00966195861665176
},
"sciq": {
"acc": 0.921,
"acc_stderr": 0.008534156773333438,
"acc_norm": 0.868,
"acc_norm_stderr": 0.01070937396352803
},
"siqa": {
"acc": 0.4247697031729785,
"acc_stderr": 0.011185271257671346,
"acc_norm": 0.4600818833162743,
"acc_norm_stderr": 0.011277955967920396
},
"truthfulqa_mc": {
"mc1": 0.22399020807833536,
"mc1_stderr": 0.014594964329474202,
"mc2": 0.35868737415331753,
"mc2_stderr": 0.013670666454421172
},
"winogrande": {
"acc": 0.6211523283346487,
"acc_stderr": 0.013633724603180328
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=stabilityai/stablelm-base-alpha-3b-v2,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": "cuda:2",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/stablelm-base-alpha-3b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.2363481228668942,
"acc_stderr": 0.012414960524301829,
"acc_norm": 0.257679180887372,
"acc_norm_stderr": 0.012780770562768407
},
"arc_easy": {
"acc": 0.4473905723905724,
"acc_stderr": 0.010202832385415646,
"acc_norm": 0.42045454545454547,
"acc_norm_stderr": 0.010129114278546535
},
"boolq": {
"acc": 0.5764525993883792,
"acc_stderr": 0.008642220663071512
},
"hellaswag": {
"acc": 0.329416450906194,
"acc_stderr": 0.004690407826933909,
"acc_norm": 0.38309101772555265,
"acc_norm_stderr": 0.004851466623601455
},
"lambada_openai": {
"ppl": 20.187359473367042,
"ppl_stderr": 0.7391414436494796,
"acc": 0.4172326799922375,
"acc_stderr": 0.006869874864639983
},
"openbookqa": {
"acc": 0.17,
"acc_stderr": 0.016815633531393422,
"acc_norm": 0.294,
"acc_norm_stderr": 0.020395095484936614
},
"sciq": {
"acc": 0.717,
"acc_stderr": 0.01425181090648174,
"acc_norm": 0.649,
"acc_norm_stderr": 0.015100563798316403
},
"siqa": {
"acc": 0.3561924257932446,
"acc_stderr": 0.010836006561369118,
"acc_norm": 0.4094165813715456,
"acc_norm_stderr": 0.01112684957658903
},
"piqa": {
"acc": 0.6381936887921654,
"acc_stderr": 0.011211397313020371,
"acc_norm": 0.6273122959738846,
"acc_norm_stderr": 0.01128131833289774
},
"truthfulqa_mc": {
"mc1": 0.22399020807833536,
"mc1_stderr": 0.014594964329474205,
"mc2": 0.4052844601694033,
"mc2_stderr": 0.014547007787950397
},
"winogrande": {
"acc": 0.526440410418311,
"acc_stderr": 0.014032823874407229
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=stabilityai/stablelm-base-alpha-3b,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "8",
"batch_sizes": [],
"device": null,
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/stablelm-base-alpha-7b-v2.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.3848122866894198,
"acc_stderr": 0.014218371065251107,
"acc_norm": 0.4052901023890785,
"acc_norm_stderr": 0.014346869060229334
},
"arc_easy": {
"acc": 0.7319023569023569,
"acc_stderr": 0.009089526578213694,
"acc_norm": 0.6910774410774411,
"acc_norm_stderr": 0.009481048387761353
},
"boolq": {
"acc": 0.7030581039755351,
"acc_stderr": 0.007991418738281637
},
"hellaswag": {
"acc": 0.5553674566819359,
"acc_stderr": 0.004959094146471525,
"acc_norm": 0.7426807408882693,
"acc_norm_stderr": 0.004362633637374482
},
"lambada_openai": {
"ppl": 3.366151815984524,
"ppl_stderr": 0.06573496070440447,
"acc": 0.7418979235396856,
"acc_stderr": 0.006096490478492321
},
"openbookqa": {
"acc": 0.304,
"acc_stderr": 0.020591649571224925,
"acc_norm": 0.418,
"acc_norm_stderr": 0.022080014812228134
},
"piqa": {
"acc": 0.7845484221980413,
"acc_stderr": 0.009592463115658117,
"acc_norm": 0.8019586507072906,
"acc_norm_stderr": 0.009298209954776726
},
"sciq": {
"acc": 0.939,
"acc_stderr": 0.007572076091557423,
"acc_norm": 0.917,
"acc_norm_stderr": 0.00872852720607479
},
"siqa": {
"acc": 0.4242579324462641,
"acc_stderr": 0.011183502662341787,
"acc_norm": 0.4692937563971341,
"acc_norm_stderr": 0.011292714928103489
},
"truthfulqa_mc": {
"mc1": 0.2423500611995104,
"mc1_stderr": 0.01500067437357034,
"mc2": 0.36457253410937596,
"mc2_stderr": 0.0135567172376818
},
"winogrande": {
"acc": 0.6882399368587214,
"acc_stderr": 0.013018571197638537
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=stabilityai/stablelm-base-alpha-7b-v2,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "2",
"batch_sizes": [],
"device": "cuda:2",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: evals/stablelm-base-alpha-7b.json
================================================
{
"results": {
"arc_challenge": {
"acc": 0.24829351535836178,
"acc_stderr": 0.012624912868089755,
"acc_norm": 0.27047781569965873,
"acc_norm_stderr": 0.012980954547659556
},
"arc_easy": {
"acc": 0.5054713804713805,
"acc_stderr": 0.01025916922861504,
"acc_norm": 0.44865319865319864,
"acc_norm_stderr": 0.010205540414612876
},
"boolq": {
"acc": 0.600611620795107,
"acc_stderr": 0.008566178448007831
},
"hellaswag": {
"acc": 0.3467436765584545,
"acc_stderr": 0.004749606196363328,
"acc_norm": 0.41216889065923124,
"acc_norm_stderr": 0.004912192800263316
},
"lambada_openai": {
"ppl": 9.459674222745228,
"ppl_stderr": 0.33183598007493614,
"acc": 0.5511352610130021,
"acc_stderr": 0.006929452414790843
},
"openbookqa": {
"acc": 0.214,
"acc_stderr": 0.018359797502387025,
"acc_norm": 0.32,
"acc_norm_stderr": 0.02088234048876181
},
"piqa": {
"acc": 0.6675734494015234,
"acc_stderr": 0.010991141557445587,
"acc_norm": 0.6735582154515778,
"acc_norm_stderr": 0.010940467046177304
},
"sciq": {
"acc": 0.801,
"acc_stderr": 0.012631649083099177,
"acc_norm": 0.697,
"acc_norm_stderr": 0.014539683710535257
},
"siqa": {
"acc": 0.394575230296827,
"acc_stderr": 0.011059713589720797,
"acc_norm": 0.4140225179119754,
"acc_norm_stderr": 0.011145545345176117
},
"truthfulqa_mc": {
"mc1": 0.23745410036719705,
"mc1_stderr": 0.014896277441041836,
"mc2": 0.3995908363542637,
"mc2_stderr": 0.014371652685680641
},
"winogrande": {
"acc": 0.5011838989739542,
"acc_stderr": 0.014052446290529012
}
},
"versions": {
"arc_challenge": 0,
"arc_easy": 0,
"boolq": 1,
"hellaswag": 0,
"lambada_openai": 0,
"openbookqa": 0,
"piqa": 0,
"sciq": 0,
"siqa": 0,
"truthfulqa_mc": 1,
"winogrande": 0
},
"config": {
"model": "gpt2",
"model_args": "use_fast=True,pretrained=stabilityai/stablelm-base-alpha-7b,trust_remote_code=True,low_cpu_mem_usage=True,dtype=auto",
"num_fewshot": 0,
"batch_size": "2",
"batch_sizes": [],
"device": "cuda:1",
"no_cache": true,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}
================================================
FILE: notebooks/stablelm-alpha.ipynb
================================================
{
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "4weyZUFfgUlD"
},
"source": [
"# StableLM-Alpha\n",
"\n",
"[](https://colab.research.google.com/github/Stability-AI/StableLM//blob/main/notebooks/stablelm-alpha.ipynb)\n",
"\n",
"<img src=\"https://raw.githubusercontent.com/Stability-AI/StableLM/main/assets/mascot.png?token=GHSAT0AAAAAABWTZAV7EFSADKXWO3HDNKPYZBZ6Z7A\"/>\n",
"\n",
"This notebook is designed to let you quickly generate text with the latest StableLM models (**StableLM-Alpha**) using Hugging Face's `transformers` library."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "8xicyuk_Ezuw"
},
"outputs": [],
"source": [
"!nvidia-smi"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "V1Da2YDX71IF"
},
"outputs": [],
"source": [
"!pip install -U pip\n",
"!pip install accelerate bitsandbytes torch transformers"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"cellView": "form",
"id": "sSifeGXKlIgY"
},
"outputs": [],
"source": [
"#@title Setup\n",
"\n",
"import torch\n",
"from transformers import AutoModelForCausalLM, AutoTokenizer, StoppingCriteria, StoppingCriteriaList\n",
"\n",
"from IPython.display import Markdown, display\n",
"def hr(): display(Markdown('---'))\n",
"def cprint(msg: str, color: str = \"blue\", **kwargs) -> None:\n",
" color_codes = {\n",
" \"blue\": \"\\033[34m\",\n",
" \"red\": \"\\033[31m\",\n",
" \"green\": \"\\033[32m\",\n",
" \"yellow\": \"\\033[33m\",\n",
" \"purple\": \"\\033[35m\",\n",
" \"cyan\": \"\\033[36m\",\n",
" }\n",
" \n",
" if color not in color_codes:\n",
" raise ValueError(f\"Invalid info color: `{color}`\")\n",
" \n",
" print(color_codes[color] + msg + \"\\033[0m\", **kwargs)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"cellView": "form",
"id": "dQZCeE-ujdzW"
},
"outputs": [],
"source": [
"#@title Pick Your Model\n",
"#@markdown Refer to Hugging Face docs for more information the parameters below: https://huggingface.co/docs/transformers/main/en/main_classes/model#transformers.PreTrainedModel.from_pretrained\n",
"\n",
"# Choose model name\n",
"model_name = \"stabilityai/stablelm-tuned-alpha-7b\" #@param [\"stabilityai/stablelm-tuned-alpha-7b\", \"stabilityai/stablelm-base-alpha-7b\", \"stabilityai/stablelm-tuned-alpha-3b\", \"stabilityai/stablelm-base-alpha-3b\"]\n",
"\n",
"cprint(f\"Using `{model_name}`\", color=\"blue\")\n",
"\n",
"# Select \"big model inference\" parameters\n",
"torch_dtype = \"float16\" #@param [\"float16\", \"bfloat16\", \"float\"]\n",
"load_in_8bit = False #@param {type:\"boolean\"}\n",
"device_map = \"auto\"\n",
"\n",
"cprint(f\"Loading with: `{torch_dtype=}, {load_in_8bit=}, {device_map=}`\")\n",
"\n",
"tokenizer = AutoTokenizer.from_pretrained(model_name)\n",
"model = AutoModelForCausalLM.from_pretrained(\n",
" model_name,\n",
" torch_dtype=getattr(torch, torch_dtype),\n",
" load_in_8bit=load_in_8bit,\n",
" device_map=device_map,\n",
" offload_folder=\"./offload\",\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 35,
"metadata": {
"cellView": "form",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 327
},
"id": "P01Db-SVwtPO",
"outputId": "9911dead-44b8-43e2-de73-c40857131065"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[34mSampling with: `max_new_tokens=128, temperature=0.7, top_k=0, top_p=0.9, do_sample=True`\u001b[0m\n"
]
},
{
"data": {
"text/markdown": [
"---"
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Can you write a song about a pirate at sea? \u001b[32mSure, here's a song about a pirate at sea:\n",
"\n",
"Verse 1:\n",
"There he was, a pirate so bold\n",
"Sailing the seas, his story untold\n",
"His name was Captain Jack, and he ruled the waves\n",
"A legend in the seas, he conquered all his foes\n",
"\n",
"Chorus:\n",
"Oh, Captain Jack, the pirate of the sea\n",
"Your bravery and your daring, set us all free\n",
"From the tyranny of the sea, you led us to glory\n",
"A legend in our hearts, you'll be remembered as our story\n",
"\n",
"Verse 2:\n",
"He sailed the\u001b[0m\n"
]
}
],
"source": [
"#@title Generate Text\n",
"#@markdown <b>Note: The model response is colored in green</b>\n",
"\n",
"class StopOnTokens(StoppingCriteria):\n",
" def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool:\n",
" stop_ids = [50278, 50279, 50277, 1, 0]\n",
" for stop_id in stop_ids:\n",
" if input_ids[0][-1] == stop_id:\n",
" return True\n",
" return False\n",
"\n",
"# Process the user prompt\n",
"user_prompt = \"Can you write a song about a pirate at sea?\" #@param {type:\"string\"}\n",
"if \"tuned\" in model_name:\n",
" # Add system prompt for chat tuned models\n",
" system_prompt = \"\"\"<|SYSTEM|># StableLM Tuned (Alpha version)\n",
" - StableLM is a helpful and harmless open-source AI language model developed by StabilityAI.\n",
" - StableLM is excited to be able to help the user, but will refuse to do anything that could be considered harmful to the user.\n",
" - StableLM is more than just an information source, StableLM is also able to write poetry, short stories, and make jokes.\n",
" - StableLM will refuse to participate in anything that could harm a human.\n",
" \"\"\"\n",
" prompt = f\"{system_prompt}<|USER|>{user_prompt}<|ASSISTANT|>\"\n",
"else:\n",
" prompt = user_prompt\n",
"\n",
"# Sampling args\n",
"max_new_tokens = 128 #@param {type:\"slider\", min:32.0, max:3072.0, step:32}\n",
"temperature = 0.7 #@param {type:\"slider\", min:0.0, max:1.25, step:0.05}\n",
"top_k = 0 #@param {type:\"slider\", min:0.0, max:1.0, step:0.05}\n",
"top_p = 0.9 #@param {type:\"slider\", min:0.0, max:1.0, step:0.05}\n",
"do_sample = True #@param {type:\"boolean\"}\n",
"\n",
"cprint(f\"Sampling with: `{max_new_tokens=}, {temperature=}, {top_k=}, {top_p=}, {do_sample=}`\")\n",
"hr()\n",
"\n",
"# Create `generate` inputs\n",
"inputs = tokenizer(prompt, return_tensors=\"pt\")\n",
"inputs.to(model.device)\n",
"\n",
"# Generate\n",
"tokens = model.generate(\n",
" **inputs,\n",
" max_new_tokens=max_new_tokens,\n",
" temperature=temperature,\n",
" top_k=top_k,\n",
" top_p=top_p,\n",
" do_sample=do_sample,\n",
" pad_token_id=tokenizer.eos_token_id,\n",
" stopping_criteria=StoppingCriteriaList([StopOnTokens()])\n",
")\n",
"\n",
"# Extract out only the completion tokens\n",
"completion_tokens = tokens[0][inputs['input_ids'].size(1):]\n",
"completion = tokenizer.decode(completion_tokens, skip_special_tokens=True)\n",
"\n",
"# Display\n",
"print(user_prompt + \" \", end=\"\")\n",
"cprint(completion, color=\"green\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "rIZm5uwaQLa4"
},
"source": [
"## License (Apache 2.0)\n",
"\n",
"Copyright (c) 2023 by [StabilityAI LTD](https://stability.ai/)\n",
"\n",
"Licensed under the Apache License, Version 2.0 (the \"License\");\n",
"you may not use this file except in compliance with the License.\n",
"You may obtain a copy of the License at\n",
"\n",
" http://www.apache.org/licenses/LICENSE-2.0\n",
"\n",
"Unless required by applicable law or agreed to in writing, software\n",
"distributed under the License is distributed on an \"AS IS\" BASIS,\n",
"WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
"See the License for the specific language governing permissions and\n",
"limitations under the License."
]
}
],
"metadata": {
"accelerator": "GPU",
"colab": {
"machine_shape": "hm",
"provenance": []
},
"gpuClass": "standard",
"kernelspec": {
"display_name": "Python 3",
"name": "python3"
},
"language_info": {
"name": "python"
}
},
"nbformat": 4,
"nbformat_minor": 0
}
gitextract_yw65xlws/
├── LICENSE
├── README.md
├── configs/
│ ├── stablelm-2-12b.yml
│ ├── stablelm-2-1_6b.yml
│ ├── stablelm-3b-4e1t.yml
│ ├── stablelm-base-alpha-3b-v2-4k-extension.yml
│ ├── stablelm-base-alpha-3b-v2.yml
│ ├── stablelm-base-alpha-3b.yml
│ ├── stablelm-base-alpha-7b-v2-4k-extension.yml
│ ├── stablelm-base-alpha-7b-v2.yml
│ └── stablelm-base-alpha-7b.yml
├── evals/
│ ├── external/
│ │ ├── EleutherAI-pythia-2.8b-deduped.json
│ │ ├── EleutherAI_gpt-j-6B.json
│ │ ├── EleutherAI_gpt-neox-20B.json
│ │ ├── EleutherAI_pythia-12b.json
│ │ ├── EleutherAI_pythia-6.9b.json
│ │ ├── Qwen-Qwen-7B-Chat.json
│ │ ├── Qwen-Qwen-7B.json
│ │ ├── baichuan-inc_Baichuan2-7B-Base.json
│ │ ├── bigscience-bloom-3b.json
│ │ ├── bigscience-bloom-7b1.json
│ │ ├── cerebras-btlm-3b-8k-base.json
│ │ ├── facebook-opt-2.7b.json
│ │ ├── facebook-opt-6.7b.json
│ │ ├── huggyllama-llama-7b.json
│ │ ├── kittn_mistral-7B-v0.1-hf.json
│ │ ├── meta-llama-Llama-2-13b-hf.json
│ │ ├── meta-llama-Llama-2-7b.json
│ │ ├── microsoft-phi-1_5.json
│ │ ├── mosaicml-mpt-7b.json
│ │ ├── openlm-research-open_llama_13b
│ │ ├── openlm-research-open_llama_3b_v2.json
│ │ ├── openlm-research-open_llama_7b_v2.json
│ │ ├── tiiuae_falcon-7b.json
│ │ └── togethercomputer-RedPajama-INCITE-7B-Base2.json
│ ├── open_llm_leaderboard/
│ │ ├── stablelm-beta-3b-v2-arc-challenge.json
│ │ ├── stablelm-beta-3b-v2-hellaswag.json
│ │ ├── stablelm-beta-3b-v2-mmmlu.json
│ │ ├── stablelm-beta-3b-v2-truthfulqa_mc.json
│ │ ├── stablelm-beta-7b-v2-arc-challenge.json
│ │ ├── stablelm-beta-7b-v2-hellaswag.json
│ │ ├── stablelm-beta-7b-v2-mmmlu.json
│ │ └── stablelm-beta-7b-v2-truthfulqa_mc.json
│ ├── stablelm-3b-4e1t.json
│ ├── stablelm-base-alpha-3b-v2.json
│ ├── stablelm-base-alpha-3b.json
│ ├── stablelm-base-alpha-7b-v2.json
│ └── stablelm-base-alpha-7b.json
└── notebooks/
└── stablelm-alpha.ipynb
Condensed preview — 49 files, each showing path, character count, and a content snippet. Download the .json file or copy for the full structured content (200K chars).
[
{
"path": "LICENSE",
"chars": 11357,
"preview": " Apache License\n Version 2.0, January 2004\n "
},
{
"path": "README.md",
"chars": 36467,
"preview": "# StableLM: Stability AI Language Models\n\n\n<br/>*“A Stochastic Parrot, flat des"
},
{
"path": "configs/stablelm-2-12b.yml",
"chars": 3953,
"preview": "{\n # parallelism settings\n \"pipe-parallel-size\": 0,\n \"model-parallel-size\": 2,\n\n # model settings\n \"num-layers\": 40"
},
{
"path": "configs/stablelm-2-1_6b.yml",
"chars": 3783,
"preview": "{\n # parallelism settings\n \"pipe-parallel-size\": 0,\n \"model-parallel-size\": 1,\n\n # model settings\n \"num-layers\": 24"
},
{
"path": "configs/stablelm-3b-4e1t.yml",
"chars": 2957,
"preview": "{\n # parallelism settings\n \"pipe-parallel-size\": 1,\n \"model-parallel-size\": 1,\n\n # model settings\n \"num-layers\": 32"
},
{
"path": "configs/stablelm-base-alpha-3b-v2-4k-extension.yml",
"chars": 3424,
"preview": "{\n # parallelism settings\n \"pipe-parallel-size\": 1,\n \"model-parallel-size\": 2,\n\n # model settings\n \"num-layers\": 32"
},
{
"path": "configs/stablelm-base-alpha-3b-v2.yml",
"chars": 3337,
"preview": "{\n # parallelism settings\n \"pipe-parallel-size\": 1,\n \"model-parallel-size\": 2,\n\n # model settings\n \"num-layers\": 32"
},
{
"path": "configs/stablelm-base-alpha-3b.yml",
"chars": 3524,
"preview": "{\n # parallelism settings\n \"pipe-parallel-size\": 1,\n \"model-parallel-size\": 4,\n\n # model settings\n \"num-layers\": 16"
},
{
"path": "configs/stablelm-base-alpha-7b-v2-4k-extension.yml",
"chars": 3422,
"preview": "{\n # parallelism settings\n \"pipe-parallel-size\": 1,\n \"model-parallel-size\": 2,\n\n # model settings\n \"num-layers\": 32"
},
{
"path": "configs/stablelm-base-alpha-7b-v2.yml",
"chars": 3155,
"preview": "{\n # parallelism settings\n \"pipe-parallel-size\": 1,\n \"model-parallel-size\": 2,\n\n # model settings\n \"num-layers\": 32"
},
{
"path": "configs/stablelm-base-alpha-7b.yml",
"chars": 3440,
"preview": "{\n # parallelism settings\n \"pipe-parallel-size\": 1,\n \"model-parallel-size\": 2,\n\n # model settings\n \"num-layers\": 16"
},
{
"path": "evals/external/EleutherAI-pythia-2.8b-deduped.json",
"chars": 2821,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.30119453924914674,\n \"acc_stderr\": 0.013406"
},
{
"path": "evals/external/EleutherAI_gpt-j-6B.json",
"chars": 2191,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.3395904436860068,\n \"acc_stderr\": 0.01383903976282016,\n "
},
{
"path": "evals/external/EleutherAI_gpt-neox-20B.json",
"chars": 2193,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.378839590443686,\n \"acc_stderr\": 0.014175915490000319,\n "
},
{
"path": "evals/external/EleutherAI_pythia-12b.json",
"chars": 2192,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.318259385665529,\n \"acc_stderr\": 0.013611993916971451,\n "
},
{
"path": "evals/external/EleutherAI_pythia-6.9b.json",
"chars": 2189,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.318259385665529,\n \"acc_stderr\": 0.013611993916971451,\n "
},
{
"path": "evals/external/Qwen-Qwen-7B-Chat.json",
"chars": 2172,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.45563139931740615,\n \"acc_stderr\": 0.014553749939306863,\n "
},
{
"path": "evals/external/Qwen-Qwen-7B.json",
"chars": 2572,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.4539249146757679,\n \"acc_stderr\": 0.0145492"
},
{
"path": "evals/external/baichuan-inc_Baichuan2-7B-Base.json",
"chars": 2190,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.42235494880546076,\n \"acc_stderr\": 0.014434138713379977,\n "
},
{
"path": "evals/external/bigscience-bloom-3b.json",
"chars": 2571,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.27986348122866894,\n \"acc_stderr\": 0.013119"
},
{
"path": "evals/external/bigscience-bloom-7b1.json",
"chars": 2194,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.302901023890785,\n \"acc_stderr\": 0.013428241573185349,\n "
},
{
"path": "evals/external/cerebras-btlm-3b-8k-base.json",
"chars": 2812,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.34897610921501704,\n \"acc_stderr\": 0.013928"
},
{
"path": "evals/external/facebook-opt-2.7b.json",
"chars": 2185,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.2687713310580205,\n \"acc_stderr\": 0.012955065963710691,\n "
},
{
"path": "evals/external/facebook-opt-6.7b.json",
"chars": 2185,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.30716723549488056,\n \"acc_stderr\": 0.013481034054980945,\n "
},
{
"path": "evals/external/huggyllama-llama-7b.json",
"chars": 2578,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.4189419795221843,\n \"acc_stderr\": 0.0144181"
},
{
"path": "evals/external/kittn_mistral-7B-v0.1-hf.json",
"chars": 2200,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.5025597269624573,\n \"acc_stderr\": 0.014611199329843777,\n "
},
{
"path": "evals/external/meta-llama-Llama-2-13b-hf.json",
"chars": 2197,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.4863481228668942,\n \"acc_stderr\": 0.014605943429860945,\n "
},
{
"path": "evals/external/meta-llama-Llama-2-7b.json",
"chars": 2831,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.4300341296928328,\n \"acc_stderr\": 0.0144676"
},
{
"path": "evals/external/microsoft-phi-1_5.json",
"chars": 2193,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.4445392491467577,\n \"acc_stderr\": 0.014521226405627079,\n "
},
{
"path": "evals/external/mosaicml-mpt-7b.json",
"chars": 2813,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.4052901023890785,\n \"acc_stderr\": 0.0143468"
},
{
"path": "evals/external/openlm-research-open_llama_13b",
"chars": 780,
"preview": "{\n \"results\": {\n \"lambada_openai\": {\n \"ppl\": 3.569586285393247,\n \"ppl_stderr\": 0.07265272740503036,\n "
},
{
"path": "evals/external/openlm-research-open_llama_3b_v2.json",
"chars": 2833,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.3387372013651877,\n \"acc_stderr\": 0.0138305"
},
{
"path": "evals/external/openlm-research-open_llama_7b_v2.json",
"chars": 2842,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.38822525597269625,\n \"acc_stderr\": 0.014241"
},
{
"path": "evals/external/tiiuae_falcon-7b.json",
"chars": 2190,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.40273037542662116,\n \"acc_stderr\": 0.014332236306790149,\n "
},
{
"path": "evals/external/togethercomputer-RedPajama-INCITE-7B-Base2.json",
"chars": 2393,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.3771331058020478,\n \"acc_stderr\": 0.014163366896192593,\n "
},
{
"path": "evals/open_llm_leaderboard/stablelm-beta-3b-v2-arc-challenge.json",
"chars": 475,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.36689419795221845,\n \"acc_norm\": 0.39675767"
},
{
"path": "evals/open_llm_leaderboard/stablelm-beta-3b-v2-hellaswag.json",
"chars": 466,
"preview": "{\n \"results\": {\n \"hellaswag\": {\n \"acc\": 0.5239992033459471,\n \"acc_norm\": 0.7066321449910"
},
{
"path": "evals/open_llm_leaderboard/stablelm-beta-3b-v2-mmmlu.json",
"chars": 10578,
"preview": "{\n \"results\": {\n \"hendrycksTest-professional_psychology\": {\n \"acc\": 0.2826797385620915,\n "
},
{
"path": "evals/open_llm_leaderboard/stablelm-beta-3b-v2-truthfulqa_mc.json",
"chars": 470,
"preview": "{\n \"results\": {\n \"truthfulqa_mc\": {\n \"mc1\": 0.22399020807833536,\n \"mc2\": 0.3587313729483"
},
{
"path": "evals/open_llm_leaderboard/stablelm-beta-7b-v2-arc-challenge.json",
"chars": 473,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.431740614334471,\n \"acc_norm\": 0.4726962457"
},
{
"path": "evals/open_llm_leaderboard/stablelm-beta-7b-v2-hellaswag.json",
"chars": 466,
"preview": "{\n \"results\": {\n \"hellaswag\": {\n \"acc\": 0.5721967735510854,\n \"acc_norm\": 0.7706632144991"
},
{
"path": "evals/open_llm_leaderboard/stablelm-beta-7b-v2-mmmlu.json",
"chars": 10565,
"preview": "{\n \"results\": {\n \"hendrycksTest-high_school_european_history\": {\n \"acc\": 0.48484848484848486,\n "
},
{
"path": "evals/open_llm_leaderboard/stablelm-beta-7b-v2-truthfulqa_mc.json",
"chars": 468,
"preview": "{\n \"results\": {\n \"truthfulqa_mc\": {\n \"mc1\": 0.2423500611995104,\n \"mc2\": 0.36458564527224"
},
{
"path": "evals/stablelm-3b-4e1t.json",
"chars": 2191,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.3779863481228669,\n \"acc_stderr\": 0.014169664520303103,\n "
},
{
"path": "evals/stablelm-base-alpha-3b-v2.json",
"chars": 2837,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.3242320819112628,\n \"acc_stderr\": 0.0136788"
},
{
"path": "evals/stablelm-base-alpha-3b.json",
"chars": 2827,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.2363481228668942,\n \"acc_stderr\": 0.0124149"
},
{
"path": "evals/stablelm-base-alpha-7b-v2.json",
"chars": 2836,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.3848122866894198,\n \"acc_stderr\": 0.0142183"
},
{
"path": "evals/stablelm-base-alpha-7b.json",
"chars": 2835,
"preview": "{\n \"results\": {\n \"arc_challenge\": {\n \"acc\": 0.24829351535836178,\n \"acc_stderr\": 0.012624"
},
{
"path": "notebooks/stablelm-alpha.ipynb",
"chars": 9973,
"preview": "{\n \"cells\": [\n {\n \"cell_type\": \"markdown\",\n \"metadata\": {\n \"id\": \"4weyZUFfgUlD\"\n },\n \"sou"
}
]
About this extraction
This page contains the full source code of the Stability-AI/StableLM GitHub repository, extracted and formatted as plain text for AI agents and large language models (LLMs). The extraction includes 49 files (178.3 KB), approximately 60.0k tokens. Use this with OpenClaw, Claude, ChatGPT, Cursor, Windsurf, or any other AI tool that accepts text input. You can copy the full output to your clipboard or download it as a .txt file.
Extracted by GitExtract — free GitHub repo to text converter for AI. Built by Nikandr Surkov.