SYMBOL INDEX (157 symbols across 21 files)

FILE: Extra-Chapter/generation-method/llm_generation.py
  function test_decoding_strategies (line 4) | def test_decoding_strategies():
  function test_original_generation (line 117) | def test_original_generation():

FILE: Extra-Chapter/s1-vllm-thinking-budget/s1.py
  function build_input (line 5) | def build_input(prompt, tokenizer):
  function count_thinking_token (line 18) | def count_thinking_token(outputs, tokenizer):
  function count_token (line 24) | def count_token(string, tokenizer):
  function run_thinking_budget_sample (line 28) | def run_thinking_budget_sample(llm_model, tokenizer, user_input, thinkin...
  function run_sample (line 87) | def run_sample(llm_model, tokenizer, user_input):

FILE: docs/chapter2/code/transformer.py
  class ModelArgs (line 9) | class ModelArgs:
  class MultiHeadAttention (line 21) | class MultiHeadAttention(nn.Module):
    method __init__ (line 23) | def __init__(self, args: ModelArgs, is_causal=False):
    method forward (line 55) | def forward(self, q: torch.Tensor, k: torch.Tensor, v: torch.Tensor):
  class LayerNorm (line 100) | class LayerNorm(nn.Module):
    method __init__ (line 102) | def __init__(self, features, eps=1e-6):
    method forward (line 109) | def forward(self, x):
  class MLP (line 116) | class MLP(nn.Module):
    method __init__ (line 118) | def __init__(self, dim: int, hidden_dim: int, dropout: float):
    method forward (line 127) | def forward(self, x):
  class EncoderLayer (line 134) | class EncoderLayer(nn.Module):
    method __init__ (line 135) | def __init__(self, args):
    method forward (line 144) | def forward(self, x):
  class Encoder (line 153) | class Encoder(nn.Module):
    method __init__ (line 155) | def __init__(self, args):
    method forward (line 161) | def forward(self, x):
  class DecoderLayer (line 167) | class DecoderLayer(nn.Module):
    method __init__ (line 169) | def __init__(self, args):
    method forward (line 182) | def forward(self, x, enc_out):
  class Decoder (line 194) | class Decoder(nn.Module):
    method __init__ (line 196) | def __init__(self, args):
    method forward (line 202) | def forward(self, x, enc_out):
  class PositionalEncoding (line 208) | class PositionalEncoding(nn.Module):
    method __init__ (line 211) | def __init__(self, args):
    method forward (line 229) | def forward(self, x):
  class Transformer (line 235) | class Transformer(nn.Module):
    method __init__ (line 238) | def __init__(self, args):
    method get_num_params (line 262) | def get_num_params(self, non_embedding=False):
    method _init_weights (line 272) | def _init_weights(self, module):
    method forward (line 283) | def forward(self, idx, targets=None):
  function main (line 322) | def main():

FILE: docs/chapter5/code/dataset.py
  class PretrainDataset (line 10) | class PretrainDataset(Dataset):
    method __init__ (line 11) | def __init__(self, data_path, tokenizer, max_length=512):
    method __len__ (line 25) | def __len__(self):
    method __getitem__ (line 28) | def __getitem__(self, index: int):
  class SFTDataset (line 48) | class SFTDataset(Dataset):
    method __init__ (line 49) | def __init__(self, data_path, tokenizer, max_length=512):
    method __len__ (line 62) | def __len__(self):
    method generate_loss_mask (line 65) | def generate_loss_mask(self, input_ids):
    method __getitem__ (line 101) | def __getitem__(self, index: int):

FILE: docs/chapter5/code/ddp_pretrain.py
  function Logger (line 25) | def Logger(content):
  function get_lr (line 34) | def get_lr(it, all):
  function train_epoch (line 68) | def train_epoch(epoch):
  function init_model (line 171) | def init_model():

FILE: docs/chapter5/code/ddp_sft_full.py
  function Logger (line 24) | def Logger(content):
  function get_lr (line 28) | def get_lr(it, all):
  function train_epoch (line 51) | def train_epoch(epoch):
  function init_model (line 122) | def init_model():

FILE: docs/chapter5/code/deal_dataset.py
  function split_text (line 14) | def split_text(text, chunk_size=512):
  function convert_message (line 29) | def convert_message(data):

FILE: docs/chapter5/code/export_model.py
  function count_parameters (line 9) | def count_parameters(model):
  function export_model (line 13) | def export_model(tokenizer_path, model_config, model_ckpt_path, save_dir...

FILE: docs/chapter5/code/k_model.py
  class ModelConfig (line 14) | class ModelConfig(PretrainedConfig):
    method __init__ (line 16) | def __init__(
  class RMSNorm (line 46) | class RMSNorm(nn.Module):
    method __init__ (line 47) | def __init__(self, dim: int, eps: float):
    method _norm (line 54) | def _norm(self, x):
    method forward (line 61) | def forward(self, x):
  function precompute_freqs_cis (line 70) | def precompute_freqs_cis(dim: int, end: int, theta: float = 10000.0):
  function reshape_for_broadcast (line 85) | def reshape_for_broadcast(freqs_cis: torch.Tensor, x: torch.Tensor):
  function apply_rotary_emb (line 97) | def apply_rotary_emb(
  function repeat_kv (line 124) | def repeat_kv(x: torch.Tensor, n_rep: int) -> torch.Tensor:
  class Attention (line 139) | class Attention(nn.Module):
    method __init__ (line 140) | def __init__(self, args: ModelConfig):
    method forward (line 182) | def forward(self, x: torch.Tensor, freqs_cos: torch.Tensor, freqs_sin:...
  class MLP (line 250) | class MLP(nn.Module):
    method __init__ (line 251) | def __init__(self, dim: int, hidden_dim: int, multiple_of: int, dropou...
    method forward (line 268) | def forward(self, x):
  class DecoderLayer (line 276) | class DecoderLayer(nn.Module):
    method __init__ (line 277) | def __init__(self, layer_id: int, args: ModelConfig):
    method forward (line 301) | def forward(self, x, freqs_cos, freqs_sin, attention_mask: Optional[to...
  class Transformer (line 309) | class Transformer(PreTrainedModel):
    method __init__ (line 313) | def __init__(self, args: ModelConfig = None):
    method _init_weights (line 355) | def _init_weights(self, module):
    method _prepare_attention_mask (line 364) | def _prepare_attention_mask(self, attention_mask: Optional[torch.Tenso...
    method _left_pad_by_attention_mask (line 378) | def _left_pad_by_attention_mask(
    method forward (line 403) | def forward(self, tokens: torch.Tensor, targets: Optional[torch.Tensor...
    method generate (line 463) | def generate(
    method _greedy_decode (line 527) | def _greedy_decode(self, logits: torch.Tensor) -> torch.Tensor:
    method _random_sample (line 540) | def _random_sample(self, logits: torch.Tensor, temperature: float = 1....
    method _beam_search (line 566) | def _beam_search(self, idx: torch.Tensor, max_new_tokens: int, num_bea...
    method generate_super (line 677) | def generate_super(self,

FILE: docs/chapter5/code/model_sample.py
  class TextGenerator (line 9) | class TextGenerator:
    method __init__ (line 10) | def __init__(self,
    method chat_template (line 62) | def chat_template(self, prompt):
    method sft_sample (line 69) | def sft_sample(self,
    method pretrain_sample (line 99) | def pretrain_sample(self,

FILE: docs/chapter5/code/train_tokenizer.py
  function read_texts_from_jsonl (line 17) | def read_texts_from_jsonl(file_path: str) -> Generator[str, None, None]:
  function create_tokenizer_config (line 33) | def create_tokenizer_config(save_dir: str) -> None:
  function train_tokenizer (line 77) | def train_tokenizer(data_path: str, save_dir: str, vocab_size: int = 819...
  function eval_tokenizer (line 128) | def eval_tokenizer(tokenizer_path: str) -> None:
  function main (line 174) | def main():

FILE: docs/chapter6/code/finetune.py
  class ModelArguments (line 45) | class ModelArguments:
  class DataTrainingArguments (line 70) | class DataTrainingArguments:
  function preprocess (line 87) | def preprocess(sources, tokenizer, max_len, system_message: str = "You a...
  class SupervisedDataset (line 157) | class SupervisedDataset(Dataset):
    method __init__ (line 159) | def __init__(self, raw_data, tokenizer, max_len: int):
    method __len__ (line 169) | def __len__(self):
    method __getitem__ (line 172) | def __getitem__(self, i) -> Dict[str, torch.Tensor]:
  function main (line 180) | def main():

FILE: docs/chapter6/code/pretrain.py
  class ModelArguments (line 41) | class ModelArguments:
  class DataTrainingArguments (line 72) | class DataTrainingArguments:
  function main (line 92) | def main():

FILE: docs/chapter7/Agent/src/core.py
  class Agent (line 14) | class Agent:
    method __init__ (line 15) | def __init__(self, client: OpenAI, model: str = "Qwen/Qwen2.5-32B-Inst...
    method get_tool_schema (line 24) | def get_tool_schema(self) -> List[Dict[str, Any]]:
    method handle_tool_call (line 28) | def handle_tool_call(self, tool_call):
    method get_completion (line 42) | def get_completion(self, prompt) -> str:

FILE: docs/chapter7/Agent/src/tools.py
  function get_current_datetime (line 6) | def get_current_datetime() -> str:
  function add (line 15) | def add(a: float, b: float):
  function mul (line 24) | def mul(a: float, b: float):
  function compare (line 33) | def compare(a: float, b: float):
  function count_letter_in_string (line 47) | def count_letter_in_string(a: str, b: str):
  function search_wikipedia (line 60) | def search_wikipedia(query: str) -> str:
  function get_current_temperature (line 84) | def get_current_temperature(latitude: float, longitude: float) -> str:

FILE: docs/chapter7/Agent/src/utils.py
  function function_to_json (line 5) | def function_to_json(func) -> dict:

FILE: docs/chapter7/Agent/web_demo.py
  function load_agent (line 22) | def load_agent():

FILE: docs/chapter7/RAG/Embeddings.py
  class BaseEmbeddings (line 21) | class BaseEmbeddings:
    method __init__ (line 25) | def __init__(self, path: str, is_api: bool) -> None:
    method get_embedding (line 35) | def get_embedding(self, text: str, model: str) -> List[float]:
    method cosine_similarity (line 49) | def cosine_similarity(cls, vector1: List[float], vector2: List[float])...
  class OpenAIEmbedding (line 82) | class OpenAIEmbedding(BaseEmbeddings):
    method __init__ (line 86) | def __init__(self, path: str = '', is_api: bool = True) -> None:
    method get_embedding (line 95) | def get_embedding(self, text: str, model: str = "BAAI/bge-m3") -> List...

FILE: docs/chapter7/RAG/LLM.py
  class BaseModel (line 29) | class BaseModel:
    method __init__ (line 30) | def __init__(self, model) -> None:
    method chat (line 33) | def chat(self, prompt: str, history: List[dict], content: str) -> str:
    method load_model (line 36) | def load_model(self):
  class OpenAIChat (line 39) | class OpenAIChat(BaseModel):
    method __init__ (line 40) | def __init__(self, model: str = "Qwen/Qwen2.5-32B-Instruct") -> None:
    method chat (line 43) | def chat(self, prompt: str, history: List[dict], content: str) -> str:

FILE: docs/chapter7/RAG/VectorBase.py
  class VectorStore (line 19) | class VectorStore:
    method __init__ (line 20) | def __init__(self, document: List[str] = ['']) -> None:
    method get_vector (line 23) | def get_vector(self, EmbeddingModel: BaseEmbeddings) -> List[List[floa...
    method persist (line 30) | def persist(self, path: str = 'storage'):
    method load_vector (line 39) | def load_vector(self, path: str = 'storage'):
    method get_similarity (line 45) | def get_similarity(self, vector1: List[float], vector2: List[float]) -...
    method query (line 48) | def query(self, query: str, EmbeddingModel: BaseEmbeddings, k: int = 1...

FILE: docs/chapter7/RAG/utils.py
  class ReadFiles (line 25) | class ReadFiles:
    method __init__ (line 30) | def __init__(self, path: str) -> None:
    method get_files (line 34) | def get_files(self):
    method get_content (line 50) | def get_content(self, max_token_len: int = 600, cover_content: int = 1...
    method get_chunk (line 61) | def get_chunk(cls, text: str, max_token_len: int = 600, cover_content:...
    method read_file_content (line 136) | def read_file_content(cls, file_path: str):
    method read_pdf (line 148) | def read_pdf(cls, file_path: str):
    method read_markdown (line 158) | def read_markdown(cls, file_path: str):
    method read_text (line 171) | def read_text(cls, file_path: str):
  class Documents (line 177) | class Documents:
    method __init__ (line 181) | def __init__(self, path: str = '') -> None:
    method get_content (line 184) | def get_content(self):