SYMBOL INDEX (218 symbols across 33 files)

FILE: cli_demo.py
  function _load_model_tokenizer (line 44) | def _load_model_tokenizer(args):
  function _gc (line 68) | def _gc():
  function _clear_screen (line 75) | def _clear_screen():
  function _print_history (line 82) | def _print_history(history):
  function _get_input (line 91) | def _get_input() -> str:
  function main (line 105) | def main():

FILE: dcu-support/cli_demo.py
  function args_parser (line 5) | def args_parser():

FILE: dcu-support/cli_demo_batch.py
  function args_parser (line 5) | def args_parser():

FILE: dcu-support/package/fastllm_pytools/hf_model.py
  function create (line 17) | def create(model,

FILE: dcu-support/package/fastllm_pytools/llm.py
  function set_cpu_threads (line 70) | def set_cpu_threads(threads: int):
  function get_cpu_threads (line 73) | def get_cpu_threads() -> int:
  function print_ins_info (line 76) | def print_ins_info():
  function set_cpu_kvcache (line 79) | def set_cpu_kvcache(cpu_kvcache):
  function get_cpu_kvcache (line 82) | def get_cpu_kvcache():
  function set_cpu_low_mem (line 85) | def set_cpu_low_mem(low_mem):
  function get_cpu_low_mem (line 88) | def get_cpu_low_mem():
  function set_device_map (line 91) | def set_device_map(device_map):
  function from_hf (line 112) | def from_hf(model,
  class model (line 118) | class model:
    method __init__ (line 119) | def __init__ (self, path : str,
    method get_prompt (line 140) | def get_prompt(self,
    method save (line 151) | def save(self, path : str):
    method eval (line 154) | def eval(self):
    method build_tokenizer_decode_token_cache (line 157) | def build_tokenizer_decode_token_cache(self):
    method tokenizer_encode_string (line 168) | def tokenizer_encode_string(self, content: str) -> List[int]:
    method tokenizer_decode_token (line 191) | def tokenizer_decode_token(self, token_id: int) -> bytes:
    method response_logits (line 219) | def response_logits(self,
    method response (line 241) | def response(self,
    method stream_response (line 257) | def stream_response(self,
    method stream_response_raw (line 290) | def stream_response_raw(self,
    method chat (line 317) | def chat(self, tokenizer, query: str, history: List[Tuple[str, str]] =...
    method stream_chat (line 358) | def stream_chat(self, tokenizer, query: str, history: List[Tuple[str, ...
    method set_adapter (line 406) | def set_adapter(self, name: str):
    method disable_adapter (line 409) | def disable_adapter(self):
    method process_chatglm3_response (line 412) | def process_chatglm3_response(self, output, history):
    method build_chatglm3_input (line 433) | def build_chatglm3_input(self, tokenizer, query, history=None, role="u...
    method response_batch (line 446) | def response_batch(self, querys: List[str],
    method chat_batch (line 468) | def chat_batch(self, tokenizer, querys: List[str], historys: List[List...

FILE: dcu-support/package/fastllm_pytools/torch2flm.py
  function writeString (line 5) | def writeString(fo, s):
  function writeKeyValue (line 9) | def writeKeyValue(fo, key, value):
  function write_int8 (line 30) | def write_int8(fo, v):
  function write_int4 (line 41) | def write_int4(fo, v):
  function tofile (line 64) | def tofile(exportPath,

FILE: dcu-support/web_demo.py
  function get_model (line 12) | def get_model():

FILE: eval/evaluate_ceval.py
  function load_models_tokenizer (line 21) | def load_models_tokenizer(args):
  function format_example (line 43) | def format_example(line, include_answer=True):
  function generate_few_shot_prompt (line 55) | def generate_few_shot_prompt(k, subject, dev_df):
  function get_logits (line 67) | def get_logits(tokenizer, model, inputs: List[str]):
  function eval_subject (line 80) | def eval_subject(
  function cal_ceval (line 158) | def cal_ceval(res):
  function main (line 384) | def main(args):

FILE: eval/evaluate_chat_ceval.py
  function load_models_tokenizer (line 23) | def load_models_tokenizer(args):
  function process_before_extraction (line 37) | def process_before_extraction(gen, question, choice_dict):
  function count_substr (line 64) | def count_substr(gen, pattern):
  function extract_choice (line 68) | def extract_choice(gen, prompt, choice_list):
  function format_example (line 95) | def format_example(line):
  function extract_answer (line 102) | def extract_answer(response, row):
  function eval_subject (line 114) | def eval_subject(
  function cal_ceval (line 179) | def cal_ceval(res):
  function main (line 405) | def main(args):

FILE: eval/evaluate_chat_gsm8k.py
  function doc_to_text (line 20) | def doc_to_text(doc, use_fewshot):
  function generate_sample (line 37) | def generate_sample(model, tokenizer, question):
  function extract_answer (line 49) | def extract_answer(s):
  function is_correct (line 62) | def is_correct(completion, answer):

FILE: eval/evaluate_chat_humaneval.py
  function extract_code (line 22) | def extract_code(text, entry_point):
  function generate_sample (line 46) | def generate_sample(model, tokenizer, question, entry_point):

FILE: eval/evaluate_chat_mmlu.py
  function load_models_tokenizer (line 23) | def load_models_tokenizer(args):
  function format_example (line 42) | def format_example(line):
  function process_before_extraction (line 53) | def process_before_extraction(gen, choice_dict):
  function extract_choice (line 62) | def extract_choice(gen, choice_list):
  function extract_answer (line 89) | def extract_answer(response, row):
  function eval_subject (line 98) | def eval_subject(
  function cal_mmlu (line 159) | def cal_mmlu(res):
  function main (line 185) | def main(args):

FILE: eval/evaluate_cmmlu.py
  function load_models_tokenizer (line 24) | def load_models_tokenizer(args):
  function format_example (line 49) | def format_example(line, include_answer=True):
  function generate_few_shot_prompt (line 61) | def generate_few_shot_prompt(k, subject, dev_df):
  function get_logits (line 73) | def get_logits(tokenizer, model, inputs: List[str]):
  function eval_subject (line 86) | def eval_subject(
  function cal_cmmlu (line 164) | def cal_cmmlu(res):
  function main (line 281) | def main(args):

FILE: eval/evaluate_gsm8k.py
  function doc_to_text (line 16) | def doc_to_text(doc):
  function decode (line 25) | def decode(tokens_list, tokenizer, raw_text_len):
  function generate_sample (line 39) | def generate_sample(model, tokenizer, input_txt):
  function extract_answer_hf (line 50) | def extract_answer_hf(completion):
  function extract_answer (line 60) | def extract_answer(completion):
  function is_correct (line 68) | def is_correct(completion, answer):

FILE: eval/evaluate_humaneval.py
  function decode (line 15) | def decode(tokens_list, tokenizer, raw_text_len):
  function generate_sample (line 29) | def generate_sample(model, tokenizer, input_txt):

FILE: eval/evaluate_mmlu.py
  function load_models_tokenizer (line 22) | def load_models_tokenizer(args):
  function format_example (line 44) | def format_example(line, include_answer=True):
  function generate_few_shot_prompt (line 56) | def generate_few_shot_prompt(k, subject, dev_df):
  function get_logits (line 78) | def get_logits(tokenizer, model, inputs: List[str]):
  function eval_subject (line 94) | def eval_subject(
  function cal_mmlu (line 166) | def cal_mmlu(res):
  function main (line 194) | def main(args):

FILE: eval/evaluate_plugin.py
  function is_callable (line 18) | def is_callable(response, golden):
  function process_res (line 22) | def process_res(response):
  class _DummyTokenizer (line 61) | class _DummyTokenizer:
    method tokenize (line 62) | def tokenize(self, text: str):
  function _get_tokenized_string (line 66) | def _get_tokenized_string(tokenizer, text_list):
  function eval_action (line 79) | def eval_action(job):
  function eval_action_input (line 90) | def eval_action_input(job, tokenizer):
  class QWenAgent (line 118) | class QWenAgent(Agent):
    method __init__ (line 130) | def __init__(
    method generate_one (line 164) | def generate_one(self, prompt, stop):
  function load_models_tokenizer (line 185) | def load_models_tokenizer(args):
  function load_jobs (line 203) | def load_jobs(filename):
  function react_inference (line 211) | def react_inference(filename, model, tokenizer):
  function main (line 228) | def main(args):

FILE: examples/add_merges.py
  function load_tiktoken_bpe (line 21) | def load_tiktoken_bpe(tiktoken_bpe_file: str) -> "dict[bytes, int]":
  function dump_tiktoken_bpe (line 29) | def dump_tiktoken_bpe(bpe_ranks: "dict[bytes, int]", tiktoken_bpe_file: ...
  function bytes_to_pieces (line 35) | def bytes_to_pieces(the_bytes: bytes) -> "tuple[bytes]":
  function get_pairs (line 39) | def get_pairs(pieces: "tuple[bytes]") -> "set[tuple[bytes, bytes]]":
  function get_stats (line 43) | def get_stats(
  function merge_vocab (line 53) | def merge_vocab(
  function apply_bp (line 59) | def apply_bp(
  function bpe (line 84) | def bpe(word: bytes, merges: "dict[bytes,int]") -> "tuple[bytes, ...]":
  function best_pair_sort_key (line 97) | def best_pair_sort_key(
  function learn_bpe (line 109) | def learn_bpe(
  function load_expand_vocab (line 133) | def load_expand_vocab(path: Path) -> "dict[str, int]":
  function make_new_merges_by_bpe (line 161) | def make_new_merges_by_bpe(
  function main (line 194) | def main():

FILE: examples/auto_comments.py
  function parse_args (line 14) | def parse_args():
  class QWenChat (line 21) | class QWenChat():
    method __init__ (line 22) | def __init__(self):
    method chat (line 38) | def chat(self, query, system = ""):
  function gen_code_comments (line 49) | def gen_code_comments(context, model = None, **kwargs):
  function read_file (line 53) | def read_file(path):
  function write_file (line 58) | def write_file(path, context):
  function split_context_by_maxline (line 63) | def split_context_by_maxline(text):
  function split_context_by_splitkey (line 75) | def split_context_by_splitkey(text):
  function merge_code_and_comments (line 80) | def merge_code_and_comments(original_file, comments_path):
  function deal_one_file (line 138) | def deal_one_file(model, path, args):
  function deal_folder (line 164) | def deal_folder(model, path, args):
  function transfer (line 175) | def transfer(args):

FILE: examples/function_call_examples.py
  function call_qwen (line 18) | def call_qwen(messages, functions=None):
  function test_1 (line 37) | def test_1():
  function test_2 (line 58) | def test_2():
  function test_3 (line 171) | def test_3():
  function test_4 (line 220) | def test_4():

FILE: examples/function_call_finetune_examples.py
  function format_train_sample (line 19) | def format_train_sample(messages):
  function build_react_instruction (line 60) | def build_react_instruction(functions):
  function main (line 86) | def main():

FILE: examples/react_demo.py
  function llm_with_plugin (line 72) | def llm_with_plugin(prompt: str, history, list_of_plugin_info=()):
  function build_input_text (line 99) | def build_input_text(chat_history, list_of_plugin_info) -> str:
  function text_completion (line 144) | def text_completion(input_text: str, stop_words) -> str:  # 作为一个文本续写模型来使用
  function parse_latest_plugin_call (line 165) | def parse_latest_plugin_call(text):
  function call_plugin (line 190) | def call_plugin(plugin_name: str, plugin_args: str) -> str:
  function test (line 210) | def test():

FILE: examples/vllm_wrapper.py
  function get_stop_words_ids (line 23) | def get_stop_words_ids(chat_format, tokenizer):
  function make_context (line 32) | def make_context(
  class vLLMWrapper (line 104) | class vLLMWrapper:
    method __init__ (line 105) | def __init__(self,
    method chat (line 147) | def chat(self,

FILE: finetune.py
  class ModelArguments (line 25) | class ModelArguments:
  class DataArguments (line 30) | class DataArguments:
  class TrainingArguments (line 41) | class TrainingArguments(transformers.TrainingArguments):
  class LoraArguments (line 54) | class LoraArguments:
  function maybe_zero_3 (line 66) | def maybe_zero_3(param):
  function get_peft_state_maybe_zero_3 (line 77) | def get_peft_state_maybe_zero_3(named_params, bias):
  function rank0_print (line 104) | def rank0_print(*args):
  function safe_save_model_for_hf_trainer (line 109) | def safe_save_model_for_hf_trainer(trainer: transformers.Trainer, output...
  function preprocess (line 125) | def preprocess(
  class SupervisedDataset (line 179) | class SupervisedDataset(Dataset):
    method __init__ (line 182) | def __init__(self, raw_data, tokenizer: transformers.PreTrainedTokeniz...
    method __len__ (line 193) | def __len__(self):
    method __getitem__ (line 196) | def __getitem__(self, i) -> Dict[str, torch.Tensor]:
  class LazySupervisedDataset (line 204) | class LazySupervisedDataset(Dataset):
    method __init__ (line 207) | def __init__(self, raw_data, tokenizer: transformers.PreTrainedTokeniz...
    method __len__ (line 217) | def __len__(self):
    method __getitem__ (line 220) | def __getitem__(self, i) -> Dict[str, torch.Tensor]:
  function make_supervised_data_module (line 235) | def make_supervised_data_module(
  function train (line 256) | def train():

FILE: openai_api.py
  class BasicAuthMiddleware (line 29) | class BasicAuthMiddleware(BaseHTTPMiddleware):
    method __init__ (line 31) | def __init__(self, app, username: str, password: str):
    method dispatch (line 36) | async def dispatch(self, request: Request, call_next):
  function _gc (line 50) | def _gc(forced: bool = False):
  function lifespan (line 63) | async def lifespan(app: FastAPI):  # collects GPU memory
  class ModelCard (line 79) | class ModelCard(BaseModel):
  class ModelList (line 89) | class ModelList(BaseModel):
  class ChatMessage (line 94) | class ChatMessage(BaseModel):
  class DeltaMessage (line 100) | class DeltaMessage(BaseModel):
  class ChatCompletionRequest (line 105) | class ChatCompletionRequest(BaseModel):
  class ChatCompletionResponseChoice (line 117) | class ChatCompletionResponseChoice(BaseModel):
  class ChatCompletionResponseStreamChoice (line 123) | class ChatCompletionResponseStreamChoice(BaseModel):
  class ChatCompletionResponse (line 129) | class ChatCompletionResponse(BaseModel):
  function list_models (line 138) | async def list_models():
  function add_extra_stop_words (line 145) | def add_extra_stop_words(stop_words):
  function trim_stop_words (line 157) | def trim_stop_words(response, stop_words):
  function parse_messages (line 191) | def parse_messages(messages, functions):
  function parse_response (line 310) | def parse_response(response):
  function text_complete_last_message (line 356) | def text_complete_last_message(history, stop_words_ids, gen_kwargs, syst...
  function create_chat_completion (line 386) | async def create_chat_completion(request: ChatCompletionRequest):
  function _dump_json (line 460) | def _dump_json(data: BaseModel, *args, **kwargs) -> str:
  function predict (line 467) | async def predict(
  function _get_args (line 536) | def _get_args():

FILE: recipes/inference/vllm/vllm_wrapper.py
  function get_stop_words_ids (line 23) | def get_stop_words_ids(chat_format, tokenizer):
  function make_context (line 32) | def make_context(
  class vLLMWrapper (line 104) | class vLLMWrapper:
    method __init__ (line 105) | def __init__(self,
    method chat (line 147) | def chat(self,

FILE: recipes/tests/test_finetune/test_finetune_ds.py
  function test_finetune (line 50) | def test_finetune(num_gpus, train_type, is_chat, docker_version, deepspe...

FILE: recipes/tests/test_inference/test_inference_api.py
  function test_inference_api (line 34) | def test_inference_api(docker_version, use_cpu, use_int4):

FILE: recipes/tests/test_inference/test_inference_vllm_fschat.py
  function test_inference_vllm_fschat (line 29) | def test_inference_vllm_fschat(num_gpus, use_int4):

FILE: recipes/tests/utils.py
  function run_in_subprocess (line 7) | def run_in_subprocess(cmd):
  function simple_openai_api (line 37) | def simple_openai_api(model):
  function TelnetPort (line 51) | def TelnetPort(server_ip, port):

FILE: run_gptq.py
  function preprocess (line 13) | def preprocess(

FILE: utils.py
  function _device_map (line 6) | def _device_map(num_gpus, num_layers):
  function load_model_on_gpus (line 28) | def load_model_on_gpus(model_name_or_path, num_gpus: int = 2):

FILE: web_demo.py
  function _get_args (line 21) | def _get_args():
  function _load_model_tokenizer (line 40) | def _load_model_tokenizer(args):
  function postprocess (line 64) | def postprocess(self, y):
  function _parse_text (line 78) | def _parse_text(text):
  function _gc (line 110) | def _gc():
  function _launch_demo (line 117) | def _launch_demo(args, model, tokenizer, config):
  function main (line 200) | def main():