Functions — github.com/JIA-Lab-research/LongLoRA

hub / github.com/JIA-Lab-research/LongLoRA / functions

Functions157 in github.com/JIA-Lab-research/LongLoRA

⨍ Functions 157◇ Types & classes 28

↓ 7 callersMethod__init__

(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.)

pdf2txt/beit.py:59

↓ 7 callersFunctionapply_rotary_pos_emb

(q, k, cos, sin, position_ids)

gptneox_attn_replace.py:21

↓ 7 callersFunctionreplace_llama_attn

(use_flash_attn=True, use_full=False, inference=False)

llama_attn_replace.py:460

↓ 6 callersFunctionrotate_half

Rotates half the hidden dims of the input.

gptneox_attn_replace.py:15

↓ 4 callersFunction_cfg

(url='', **kwargs)

pdf2txt/beit.py:33

↓ 4 callersFunctionapply_rotary_pos_emb_single

(x, cos, sin, position_ids)

streaming_llm/pos_shift/modify_llama.py:26

↓ 4 callersMethodis_first_device

(self)

eval_distributed.py:115

↓ 3 callersMethodcompute

(self)

eval_distributed.py:82

↓ 3 callersFunctionreplace_gpt_neox_attn

(use_flash_attn=True, use_full=False)

gptneox_attn_replace.py:191

↓ 3 callersFunctionshift

(qkv, bsz, q_len, group_size, num_heads, head_dim)

llama_attn_replace.py:287

↓ 3 callersFunctionshift

(qkv, num_heads, head_dim)

gptneox_attn_replace.py:145

↓ 3 callersFunctionshift

(qkv, bsz, q_len, group_size, num_heads, head_dim)

llama_attn_replace_sft.py:293

↓ 2 callersFunctionapply_rotary_pos_emb_single

(x, cos, sin, position_ids)

streaming_llm/pos_shift/modify_gpt_neox.py:20

↓ 2 callersMethodforward

(self, x)

pdf2txt/beit.py:68

↓ 2 callersMethodforward_features

(self, x)

pdf2txt/beit.py:564

↓ 2 callersFunctioniceildiv

(x, y)

eval.py:53

↓ 2 callersFunctionreplace_llama_attn

(use_flash_attn=True, use_full=False, inference=False)

llama_attn_replace_sft.py:466

↓ 1 callersFunction_make_r_io_base

(f, mode: str)

supervised-fine-tune-qlora.py:41

↓ 1 callersFunction_make_r_io_base

(f, mode: str)

supervised-fine-tune.py:40

↓ 1 callersMethod_prepare_dataloader

(self, dataset: Dataset)

eval_distributed.py:118

↓ 1 callersFunction_tokenize_fn

Tokenize a list of strings.

supervised-fine-tune-qlora.py:139

↓ 1 callersFunction_tokenize_fn

Tokenize a list of strings.

supervised-fine-tune.py:138

↓ 1 callersMethodadd

(self, logits: torch.FloatTensor, labels: torch.LongTensor, model_output: object)

eval_distributed.py:78

↓ 1 callersFunctionadd_vit_config

Add config for VIT.

pdf2txt/config.py:4

↓ 1 callersFunctionapply_rotary_pos_emb_inference

(q, k, cos_sin, position_ids)

llama_attn_replace.py:350

↓ 1 callersFunctionapply_rotary_pos_emb_inference

(q, k, cos_sin, position_ids)

llama_attn_replace_sft.py:356

↓ 1 callersFunctionbuild_VIT_backbone

Create a VIT instance from config. Args: cfg: a detectron2 CfgNode Returns: A VIT backbone instance.

pdf2txt/backbone.py:111

↓ 1 callersFunctionbuild_generator

( model, tokenizer, temperature=0.6, top_p=0.9, max_gen_len=4096, use_cache=True )

inference-qlora.py:50

↓ 1 callersFunctionbuild_generator

( model, tokenizer, temperature=0.6, top_p=0.9, max_gen_len=4096, use_cache=True )

inference.py:50

↓ 1 callersFunctionbuild_generator

( model, tokenizer, temperature=0.6, top_p=0.9, max_gen_len=4096, use_cache=True )

demo.py:84

↓ 1 callersFunctionddp_setup

()

eval_distributed.py:262

↓ 1 callersFunctiondetect_objects

(image_path, predictor, cfg)

pdf2txt/pdf2txt.py:48

↓ 1 callersFunctionenable_falcon_pos_shift_attention

(model)

streaming_llm/pos_shift/modify_falcon.py:152

↓ 1 callersFunctionenable_gpt_neox_pos_shift_attention

(model)

streaming_llm/pos_shift/modify_gpt_neox.py:99

↓ 1 callersFunctionenable_llama_pos_shift_attention

(model, use_flash_attn=True)

streaming_llm/pos_shift/modify_llama.py:301

↓ 1 callersFunctionenable_streaming_llm

(model, start_size, recent_size, use_flash_attn=True)

streaming_llm/enable_streaming_llm.py:4

↓ 1 callersFunctionevaluate

(model, data, batch_size, device, seq_length, sliding_window=256, use_cache=False)

eval.py:56

↓ 1 callersMethodevaluate

(self, dataset: Dataset, metric: EvalMetric)

eval_distributed.py:98

↓ 1 callersMethodevict_for_space

(self, past_key_values, num_coming)

streaming_llm/kv_cache.py:66

↓ 1 callersMethodfix_init_weight

(self)

pdf2txt/beit.py:506

↓ 1 callersFunctiongenerate_prompt_landmark

Generates a text file and inserts an passkey at a random position.

passkey_retrivial.py:43

↓ 1 callersFunctionget_as_batch

(data, seq_length, batch_size, device='cpu', sliding_window=256)

eval.py:40

↓ 1 callersFunctionget_forward_function

(use_flash_attn=True, use_full=False)

gptneox_attn_replace.py:83

↓ 1 callersFunctiongreedy_generate

(model, tokenizer, input_ids, past_key_values, max_gen_len)

run_streaming_llama_longalpaca.py:19

↓ 1 callersFunctionjload

Load a .json file into a dictionary.

supervised-fine-tune-qlora.py:46

↓ 1 callersFunctionjload

Load a .json file into a dictionary.

supervised-fine-tune.py:45

↓ 1 callersFunctionload

(model_name_or_path)

streaming_llm/utils.py:51

↓ 1 callersFunctionmain

(args)

eval.py:113

↓ 1 callersFunctionmain

(args)

passkey_retrivial.py:89

↓ 1 callersFunctionmain

(cmd_args: list[str] = None)

eval_distributed.py:266

↓ 1 callersFunctionmain

(args)

run_streaming_llama_longalpaca.py:78

↓ 1 callersFunctionmain

(args)

get_trainable_weights.py:27

↓ 1 callersFunctionmain

(args)

inference-qlora.py:74

↓ 1 callersFunctionmain

(args)

inference.py:74

↓ 1 callersFunctionmain

(args)

merge_lora_weights_and_save_hf_model.py:60

↓ 1 callersFunctionmain

(args)

demo.py:124

↓ 1 callersFunctionmake_supervised_data_module

Make dataset and collator for supervised fine-tuning.

supervised-fine-tune-qlora.py:228

↓ 1 callersFunctionmake_supervised_data_module

Make dataset and collator for supervised fine-tuning.

supervised-fine-tune.py:227

↓ 1 callersMethodnum_tokens

(self)

eval_distributed.py:72

↓ 1 callersFunctionparse_config

()

eval.py:27

↓ 1 callersFunctionparse_config

()

passkey_retrivial.py:29

↓ 1 callersFunctionparse_config

()

get_trainable_weights.py:19

↓ 1 callersFunctionparse_config

()

inference-qlora.py:27

↓ 1 callersFunctionparse_config

()

inference.py:27

↓ 1 callersFunctionparse_config

()

merge_lora_weights_and_save_hf_model.py:28

↓ 1 callersFunctionparse_config

()

demo.py:15

↓ 1 callersFunctionpasskey_retrieval_test

(model, tokenizer, device, use_cache=False, n_garbage=60000, seed=666)

passkey_retrivial.py:70

↓ 1 callersFunctionpreprocess

Preprocess the data by tokenizing.

supervised-fine-tune-qlora.py:163

↓ 1 callersFunctionpreprocess

Preprocess the data by tokenizing.

supervised-fine-tune.py:162

↓ 1 callersFunctionprocess_pdf

(pdf_file, outputs_dir, config_file)

pdf2txt/pdf2txt.py:77

↓ 1 callersFunctionread_txt_file

(material_txt)

inference-qlora.py:41

↓ 1 callersFunctionread_txt_file

(material_txt)

inference.py:41

↓ 1 callersFunctionread_txt_file

(material_txt)

demo.py:77

↓ 1 callersFunctionrun_eval

(args: EvalArguments)

eval_distributed.py:191

↓ 1 callersFunctionsmart_tokenizer_and_embedding_resize

Resize tokenizer and embedding. Note: This is the unoptimized version that may make your embedding size not be divisible by 64.

fine-tune.py:71

↓ 1 callersFunctionsmart_tokenizer_and_embedding_resize

Resize tokenizer and embedding. Note: This is the unoptimized version that may make your embedding size not be divisible by 64.

supervised-fine-tune-qlora.py:116

↓ 1 callersFunctionsmart_tokenizer_and_embedding_resize

Resize tokenizer and embedding. Note: This is the unoptimized version that may make your embedding size not be divisible by 64.

merge_lora_weights_and_save_hf_model.py:38

↓ 1 callersFunctionsmart_tokenizer_and_embedding_resize

Resize tokenizer and embedding. Note: This is the unoptimized version that may make your embedding size not be divisible by 64.

supervised-fine-tune.py:115

↓ 1 callersFunctionstreaming_inference

(model, tokenizer, prompts, kv_cache=None, max_gen_len=1000)

run_streaming_llama_longalpaca.py:61

↓ 1 callersFunctiontrain

()

fine-tune.py:104

↓ 1 callersFunctiontrain

()

supervised-fine-tune-qlora.py:235

↓ 1 callersFunctiontrain

()

supervised-fine-tune.py:234

Method__call__

(self, instances: Sequence[Dict])

supervised-fine-tune-qlora.py:215

Method__call__

(self, instances: Sequence[Dict])

supervised-fine-tune.py:214

Method__call__

(self, past_key_values)

streaming_llm/kv_cache.py:40

Method__getitem__

(self, index)

eval_distributed.py:60

Method__getitem__

(self, i)

supervised-fine-tune-qlora.py:205

Method__getitem__

(self, i)

supervised-fine-tune.py:204

Method__init__

(self, data_path: str, seq_length: int, sliding_window: int = 256)

eval_distributed.py:47

Method__init__

(self, model: Union[PreTrainedModel, nn.Module], batch_size: int,

eval_distributed.py:87

Method__init__

(self, vocab_size: int, gpu_id: int)

eval_distributed.py:129

Method__init__

(self, data_path: str, tokenizer: transformers.PreTrainedTokenizer)

supervised-fine-tune-qlora.py:181

Method__init__

(self, data_path: str, tokenizer: transformers.PreTrainedTokenizer)

supervised-fine-tune.py:180

Method__init__

(self, drop_prob=None)

pdf2txt/beit.py:47

Method__init__

( self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0., w

pdf2txt/beit.py:79

Method__init__

(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop=0., attn_drop=0., dr

pdf2txt/beit.py:211

Method__init__

(self, img_size=[224, 224], patch_size=16, in_chans=3, embed_dim=768)

pdf2txt/beit.py:247

Method__init__

(self, backbone, img_size=[224, 224], feature_size=None, in_chans=3, embed_dim=768)

pdf2txt/beit.py:285

Method__init__

(self, window_size, num_heads)

pdf2txt/beit.py:318

Method__init__

(self, img_size=[224, 224], patch_size=16, in_chans=3,

pdf2txt/beit.py:402

next →1–100 of 157, ranked by callers