Functions — github.com/Mwie1024/Extra-CoT

hub / github.com/Mwie1024/Extra-CoT / functions

Functions1,850 in github.com/Mwie1024/Extra-CoT

⨍ Functions 1,850◇ Types & classes 174↳ Endpoints 4

↓ 2 callersFunction_strip_latex

(s: str)

src/training/qwen1p7b_rl/dataset/rl_dataset/filter_by_ratio_window.py:60

↓ 2 callersMethod_timing

(self, cur_steps: int)

llamafactory/src/llamafactory/train/callbacks.py:205

↓ 2 callersFunction_to_float_maybe

(s: str)

src/training/qwen1p7b_rl/dataset/vllm_inference.py:156

↓ 2 callersFunction_to_num

(s: str)

src/training/qwen1p7b_rl/dataset/rl_dataset/build_diff_by_model_ratio.py:59

↓ 2 callersFunction_to_num

(s: str)

src/training/qwen1p7b_rl/dataset/rl_dataset/build_diff.py:59

↓ 2 callersFunction_to_num

(s: str)

src/training/qwen1p7b_rl/dataset/rl_dataset/filter_by_ratio_window.py:72

↓ 2 callersFunction_to_number

(s: str)

src/training/qwen1p7b_rl/dataset/merge_data.py:64

↓ 2 callersFunction_to_numeric

(s: str)

src/data/compressor_longformer/validate/eval_utils/evaluate_local.py:200

↓ 2 callersFunction_to_numeric

(s: str)

src/data/compressor_longformer/validate/eval_utils/eval_lora.py:243

↓ 2 callersFunction_to_numeric

(s: str)

src/data/compressor_longformer/validate/eval_utils/evaluate.py:101

↓ 2 callersFunction_to_numeric

(s: str)

src/eval/qwen1p7b_eval/vllm_eval.py:95

↓ 2 callersFunction_to_numeric

(s: str)

src/eval/qwen1p7b_eval/special_token_vllm.py:131

↓ 2 callersFunction_to_numeric

(s: str)

src/eval/qwen1p7b_eval/mmlu-stem/eval_vllm.py:134

↓ 2 callersFunction_zpad

(n: int, width: int = 3)

src/data/compressor/dataset_preparation/API/gpt-4o.py:236

↓ 2 callersMethodachat

r"""Asynchronously get a list of responses of the chat model.

llamafactory/src/llamafactory/chat/chat_model.py:81

↓ 2 callersMethodaget_scores

r"""Asynchronously get a list of scores of the reward model.

llamafactory/src/llamafactory/chat/chat_model.py:138

↓ 2 callersFunctionallocate

Generic allocator with: - base on shares (if given) else proportional to availability - apply per-ratio caps (optional) - greed

src/training/qwen1p7b_rl/dataset/rl_4k/convert_data_to_verl.py:243

↓ 2 callersFunctionallocate

Generic allocator with: - base on shares (if given) else proportional to availability - apply per-ratio caps (optional) - greed

src/training/qwen1p7b_rl/dataset/rl_tiktoken_10k/convert_data_to_verl.py:244

↓ 2 callersFunctionallocate

(grouped: Dict[float, List[Dict[str,Any]]], total: int, share: Dict[float,float], caps: Optional[

src/training/qwen1p7b_rl/dataset/rl_model_ratio_v2/convert_data_to_verl.py:269

↓ 2 callersFunctionanalyze_compression_ratio

分析JSONL文件中每个样本的保留索引比例 Args: jsonl_file: JSONL文件路径 Returns: 包含统计信息的字典

src/data/compressor/dataset_preparation/API/analyze_compression_ratio.py:13

↓ 2 callersFunctionas_text

将任意类型安全转为字符串（保证可计数）。

src/data/compressor_longformer/validate/tokenize_analysis.py:77

↓ 2 callersFunctionas_text

将任意类型安全转为字符串（保证可计数）。

src/data/compressor_longformer/validate/longformer_pipeline/llama3.2_3b_instruct/tokenize_analysis.py:77

↓ 2 callersMethodastream_chat

r"""Asynchronously get the response token-by-token of the chat model.

llamafactory/src/llamafactory/chat/chat_model.py:113

↓ 2 callersMethodaudit_chunks

(self, original_text: str, chunk_spans: List[Tuple[int,int]])

src/data/compressor/dataset_preparation/camel/camel_chunk_cot_and_answer/gpt_chunk_camel_data.py:607

↓ 2 callersFunctionbarrier

()

src/data/compressor_longformer/train/train_longformer_v2.py:49

↓ 2 callersFunctionbarrier

()

src/data/compressor_longformer/train/train_longformer.py:65

↓ 2 callersFunctionbatch_confmat

Compute confusion-like counts for binary classes at valid positions: valid = (mask == 1) & (gold != -100) Returns dict of counts we can

src/data/compressor_longformer/train/train_longformer.py:191

↓ 2 callersFunctionbin_of

(actual: float, bins: List[Tuple[float,float,float,str]])

src/training/qwen1p7b_rl/dataset/rl_dataset/build_diff.py:112

↓ 2 callersFunctionbin_of

(actual: float, bins: List[Tuple[float,float,float,str]])

src/training/qwen1p7b_rl/dataset/rl_dataset/filter_by_ratio_window.py:125

↓ 2 callersFunctionbucket_index

(val: int, bins: List[Tuple[int,int]])

src/data/sft/metamath_145k_query/select_sft_subset.py:86

↓ 2 callersFunctionbuild_id_map

(items: List[Dict[str, Any]])

src/data/compressor_longformer/validate/build_llamafactory_input_special_token.py:82

↓ 2 callersFunctionbuild_id_map

(items: List[Dict[str, Any]])

src/data/compressor_longformer/validate/build_llamafactory_input.py:81

↓ 2 callersFunctionbuild_id_map

(items: List[Dict[str, Any]])

src/data/compressor_longformer/validate/build_mixture_llamafactory.py:80

↓ 2 callersFunctionbuild_messages_qwen

Qwen 风格的 system + user 消息

src/training/qwen1p7b_rl/dataset/rl_tiktoken_10k/validata_verl_mean_response_length.py:33

↓ 2 callersFunctionbuild_user_text_special

你的控制头拼接函数：在 user 文本末尾加 <COMP_xx> 或 <COMP_AUTO> ratio <= 1.0 -> <COMP_xx>；ratio == 2.0 -> <COMP_AUTO>

src/training/qwen1p7b_rl/dataset/rl_tiktoken_10k/validata_verl_mean_response_length.py:17

↓ 2 callersFunctioncalculate_tps

r"""Calculate effective tokens per second.

llamafactory/src/llamafactory/extras/misc.py:104

↓ 2 callersFunctioncoerce_bool

(x)

src/data/compressor/dataset_preparation/API/api_result_completiness/merge_results.py:27

↓ 2 callersFunctioncollect_anchors

(s: str)

src/data/sft/longformer_pipeline/query_result/Compression/RL_actual_ratio/filter_bad_sample.py:169

↓ 2 callersFunctioncompute_actual_min_for_id

(idx: Dict[float, Dict[str, dict]], sid: str)

src/training/qwen1p7b_rl/dataset/merge_data.py:127

↓ 2 callersFunctioncompute_quotas

Compute integer quotas by rounding and fix rounding drift to match total exactly.

src/data/sft/longformer_pipeline/query_result/Compression/RL_actual_ratio/sample_recat_10k_auto_data.py:100

↓ 2 callersMethodconcatenated_forward

( self, model: "PreTrainedModel", batch: dict[str, "torch.Tensor"] )

llamafactory/src/llamafactory/train/kto/trainer.py:169

↓ 2 callersMethodconcatenated_forward

r"""Compute the sum log probabilities of the labels under given logits if loss_type is not IPO, ORPO or SimPO. Otherwise the average log prob

llamafactory/src/llamafactory/train/dpo/trainer.py:211

↓ 2 callersFunctioncount_parameters

r"""Return the number of trainable parameters and number of all parameters in the model.

llamafactory/src/llamafactory/extras/misc.py:117

↓ 2 callersFunctioncount_tokens

(encoder, text: str)

src/data/compressor_longformer/validate/longformer_pipeline/filter_by_token_ratio.py:16

↓ 2 callersFunctioncount_tokens

(encoder, text: str)

src/data/sft/metamath_145k_query/select_sft_subset.py:27

↓ 2 callersFunctioncount_tokens

(encoder, text: str)

src/data/sft/longformer_pipeline/filter_by_token_ratio.py:16

↓ 2 callersFunctioncount_tokens

(s: str)

src/training/qwen1p7b_rl/dataset/merge_data.py:13

↓ 2 callersFunctioncounts_to_metrics

(c: Dict[str, int])

src/data/compressor_longformer/train/train_longformer.py:222

↓ 2 callersFunctioncreate_app

(chat_model: "ChatModel")

llamafactory/src/llamafactory/api/app.py:69

↓ 2 callersFunctioncreate_chat_box

( engine: "Engine", visible: bool = False )

llamafactory/src/llamafactory/webui/components/chatbot.py:49

↓ 2 callersFunctioncreate_preview_box

(dataset_dir: "gr.Textbox", dataset: "gr.Dropdown")

llamafactory/src/llamafactory/webui/components/data.py:86

↓ 2 callersFunctioncreate_ui

(demo_mode: bool = False)

llamafactory/src/llamafactory/webui/interface.py:38

↓ 2 callersFunctiondictify

(data: "BaseModel")

llamafactory/src/llamafactory/api/common.py:23

↓ 2 callersFunctionensure_dir

(p: str)

src/data/compressor_longformer/eval/compare_longformer_llmlingua2/compare.py:39

↓ 2 callersFunctionexport_model

(args: Optional[dict[str, Any]] = None)

llamafactory/src/llamafactory/train/tuner.py:113

↓ 2 callersFunctionextract_boxed_answer

(model_output: str)

src/training/qwen1p7b_rl/dataset/vllm_inference.py:135

↓ 2 callersFunctionextract_id

从对象里取 id；若 obj 不是 dict，则把整行当作 id 值（兼容B只有id的情形）

src/data/compressor_longformer/validate/Tokenskip_pipeline/compression_llmlingua2/sample_original_data.py:16

↓ 2 callersFunctionextract_key

(obj: dict, keys: list[str])

src/data/compressor_longformer/validate/longformer_pipeline/llama3.2_3b_instruct/real_ratio/intersect_by_question.py:56

↓ 2 callersFunctionextract_key

(obj: dict, keys: list[str])

src/data/sft/longformer_pipeline/intersect_by_question.py:56

↓ 2 callersFunctionextract_question

(rec: dict, model_type: str)

src/data/compressor_longformer/validate/longformer_pipeline/longformer_tokenizer_compression.py:73

↓ 2 callersFunctionextract_question

(rec: dict, model_type: str)

src/data/sft/longformer_pipeline/longformer_compressor.py:43

↓ 2 callersFunctionextract_think_inner

(text: str)

src/data/sft/longformer_pipeline/query_result/Compression/RL_actual_ratio/filter_bad_sample.py:95

↓ 2 callersFunctionextract_think_inner

(text: str)

src/training/qwen1p7b_rl/dataset/merge_data.py:40

↓ 2 callersMethodextract_tool

r"""Extract tool message.

llamafactory/src/llamafactory/data/template.py:85

↓ 2 callersFunctionfind_expanded_modules

r"""Find the modules in the expanded blocks to apply lora.

llamafactory/src/llamafactory/model/model_utils/misc.py:55

↓ 2 callersFunctionfirst_nonempty

(d: Dict[str, Any], keys: List[str])

src/data/sft/longformer_pipeline/query_result/autofollow_sft_dataset/rewrite_follow_to_unified_json_v3.py:40

↓ 2 callersMethodforward

r"""Run forward pass and computes the log probabilities.

llamafactory/src/llamafactory/train/kto/trainer.py:134

↓ 2 callersMethodget_all_math_spans

(self, text: str, sanitize_unbalanced: bool = True)

src/data/compressor/dataset_preparation/camel/camel_chunk_cot_and_answer/gpt_chunk_camel_data.py:239

↓ 2 callersMethodget_all_math_spans

获取所有数学公式的位置区间

src/data/compressor/dataset_preparation/camel/camel_chunk_cot_and_answer/chunk_camel_data.py:38

↓ 2 callersMethodget_all_math_spans

获取所有数学公式的位置区间

src/data/compressor/dataset_preparation/camel/camel_chunk_cot_and_answer/claude_chunk_camel_data.py:36

↓ 2 callersMethodget_all_math_spans

(self, text: str, sanitize_unbalanced: bool = True)

src/data/compressor/dataset_preparation/camel/camel_chunk_only_cot/gpt_chunk_camel_data.py:239

↓ 2 callersFunctionget_batch_logps

r"""Compute the log probabilities of the given labels under the given logits. Returns: logps: A tensor of shape (batch_size,) containing

llamafactory/src/llamafactory/train/trainer_utils.py:587

↓ 2 callersFunctionget_by_dotted

(obj, path: str)

src/training/qwen1p7b_rl/dataset/filter_jsonl_by_query.py:6

↓ 2 callersFunctionget_dataset_module

r"""Convert dataset or dataset dict to dataset module.

llamafactory/src/llamafactory/data/data_utils.py:121

↓ 2 callersFunctionget_llamafactory_input

()

src/data/compressor_longformer/validate/Tokenskip_pipeline/build_llamafactory_input_numeric.py:34

↓ 2 callersMethodget_ollama_modelfile

r"""Return the ollama modelfile. TODO: support function calling.

llamafactory/src/llamafactory/data/template.py:310

↓ 2 callersFunctionget_question_from_original

(rec: Dict[str, Any])

src/data/compressor_longformer/validate/build_llamafactory_input.py:89

↓ 2 callersFunctionget_ratio

(obj: Dict[str,Any])

src/training/qwen1p7b_rl/dataset/rl_4k/convert_data_to_verl.py:102

↓ 2 callersFunctionget_ratio

(obj: Dict[str,Any])

src/training/qwen1p7b_rl/dataset/rl_tiktoken_10k/convert_data_to_verl.py:102

↓ 2 callersFunctionget_seqlens_in_batch

r"""Get the sequnce lengths in the current batch. e.g. ```python # input [ [1, 1, 2, 2, 2, 0], [1, 2, 2, 3, 3, 3],

llamafactory/src/llamafactory/model/model_utils/packing.py:55

↓ 2 callersFunctionget_tool_utils

(name: str)

llamafactory/src/llamafactory/data/tool_utils.py:443

↓ 2 callersFunctiongroup_by_ratio

(lst: List[Dict[str,Any]])

src/training/qwen1p7b_rl/dataset/rl_4k/convert_data_to_verl.py:181

↓ 2 callersFunctiongroup_by_ratio

(lst: List[Dict[str,Any]])

src/training/qwen1p7b_rl/dataset/rl_tiktoken_10k/convert_data_to_verl.py:182

↓ 2 callersFunctiongroup_by_ratio

(lst: List[Dict[str,Any]])

src/training/qwen1p7b_rl/dataset/rl_model_ratio_v2/convert_data_to_verl.py:205

↓ 2 callersFunctionhas_neighbor_mathy

(toks: List[Token], i: int)

src/data/compressor/dataset_preparation/camel/camel_chunk_cot_and_answer/math_tokenizer_no_space.py:186

↓ 2 callersFunctionhas_top_level_comma_between

(t_left: Token, t_right: Token)

src/data/compressor/dataset_preparation/camel/camel_chunk_cot_and_answer/math_tokenizer_no_space.py:253

↓ 2 callersFunctionhist_by_ratio

(lst: List[Dict[str,Any]])

src/training/qwen1p7b_rl/dataset/rl_4k/convert_data_to_verl.py:306

↓ 2 callersFunctionhist_by_ratio

(lst: List[Dict[str,Any]])

src/training/qwen1p7b_rl/dataset/rl_tiktoken_10k/convert_data_to_verl.py:307

↓ 2 callersFunctionhist_by_ratio

(lst: List[Dict[str,Any]])

src/training/qwen1p7b_rl/dataset/rl_model_ratio_v2/convert_data_to_verl.py:327

↓ 2 callersFunctioninfer_optim_dtype

r"""Infer the optimal dtype according to the model_dtype and device compatibility.

llamafactory/src/llamafactory/extras/misc.py:214

↓ 2 callersFunctionis_cmp_op

(lex: str)

src/data/compressor/dataset_preparation/camel/camel_chunk_cot_and_answer/math_tokenizer_no_space.py:192

↓ 2 callersFunctionis_connector_ident

(t: Token)

src/data/compressor/dataset_preparation/camel/camel_chunk_cot_and_answer/math_tokenizer_no_space.py:183

↓ 2 callersFunctionis_fastapi_available

()

llamafactory/src/llamafactory/extras/packages.py:49

↓ 2 callersFunctionis_matplotlib_available

()

llamafactory/src/llamafactory/extras/packages.py:69

↓ 2 callersFunctionis_requests_available

()

llamafactory/src/llamafactory/extras/packages.py:81

↓ 2 callersFunctionis_sglang_available

()

llamafactory/src/llamafactory/extras/packages.py:106

↓ 2 callersFunctionis_vllm_available

()

llamafactory/src/llamafactory/extras/packages.py:102

↓ 2 callersFunctioniter_jsonl

(path: str)

src/data/compressor_longformer/validate/eval_utils/tmp.py:9

↓ 2 callersFunctioniter_jsonl

Yield (lineno, raw_line, obj) per non-empty line; skip malformed JSON.

src/data/sft/metamath_145k_query/filter_jsonl_by_query.py:8

↓ 2 callersFunctioniter_jsonl

(path: str)

src/training/qwen1p7b_rl/dataset/filter_jsonl_by_query.py:15

← previous next →301–400 of 1,850, ranked by callers