MCPcopy Create free account

hub / github.com/XueFuzhao/OpenMoE / functions

Functions29 in github.com/XueFuzhao/OpenMoE

⨍ Functions 29◇ Types & classes 3
↓ 3 callersFunctionmin_max_normalize
(score, min_score, max_score)
eval/result_retrieval_bigbenchlite.py:153
↓ 2 callersFunctionread_csv_and_convert_to_dataframe
(file_path)
eval/plot_bigbench.py:37
↓ 1 callersFunction_apply_recursive
(module: nn.Module)
analysis/colossalai_replace/layer.py:410
↓ 1 callersMethod_ep_process
Expert Parallel Args: dispatch_data (torch.Tensor): (num_experts, capacity, hidden_size) Returns: t
analysis/colossalai_replace/layer.py:241
↓ 1 callersMethod_local_process
(self, expert_in: torch.Tensor)
analysis/colossalai_replace/layer.py:236
↓ 1 callersMethod_tp_process
without overlap: | C | | A | | R | with overlap: | C1 ||
analysis/colossalai_replace/layer.py:317
↓ 1 callersFunctionapply_llama_chat_template
(tokenizer, input_strs, sys_prompt)
script/inference_on_multi_devices.py:73
↓ 1 callersFunctioncompute_scores_repo
Returns a list of normalized scores whose length is equal to the model number, each score is a string concatenating the scores of n-shot.
eval/result_retrieval_bigbenchlite.py:158
↓ 1 callersFunctioncompute_scores_result
(task: str, models: list, model_jsons: list, number_of_shots: list, task_metric_meta
eval/result_retrieval_bigbenchlite.py:271
↓ 1 callersFunctionfind_repo_task_dir
(repo_dir: str, tasks: list)
eval/result_retrieval_bigbenchlite.py:102
↓ 1 callersFunctionfind_repo_task_model_json
(task_dir: str, models: list)
eval/result_retrieval_bigbenchlite.py:130
↓ 1 callersFunctionfind_result_task_jsonsubsets
(results_dir: str, tasks: list)
eval/result_retrieval_bigbenchlite.py:230
↓ 1 callersFunctionfind_result_taskmodel_json
(task, task_jsons: list, models: list, number_of_shots: list)
eval/result_retrieval_bigbenchlite.py:251
↓ 1 callersFunctioninference
(model, tokenizer, input_strs, gen_kwargs, add_special_tokens=True, split_special_tokens=False,
script/inference_on_multi_devices.py:37
↓ 1 callersMethodreset_parameters
(self)
analysis/colossalai_replace/layer.py:146
↓ 1 callersFunctionresult_retrieval_bigbenchlite
(file_to_save:str, results_dir: str, repo_dir:list, models_repo: list, mode
eval/result_retrieval_bigbenchlite.py:356
↓ 1 callersFunctionretrieve_and_write_repo
(file_to_save: str, repo_dir: list, models: list, tasks: list,
eval/result_retrieval_bigbenchlite.py:206
↓ 1 callersFunctionretrieve_and_write_results
(file_to_save, results_dir, models, tasks, number_of_shots, task_metric_metainfos)
eval/result_retrieval_bigbenchlite.py:331
↓ 1 callersFunctionupdate_task_metainfo
(task_metric_meta, model_json)
eval/result_retrieval_bigbenchlite.py:120
Method__call__
(self, input_ids: torch.LongTensor, scores: torch.FloatTensor)
script/inference_on_multi_devices.py:27
Method__init__
(self, base_len: int, eos_token_id: int)
script/inference_on_multi_devices.py:22
Method__init__
( self, num_experts: int, hidden_size: int, intermediate_size: int, ro
analysis/colossalai_replace/layer.py:49
Functionapply_load_balance
apply load balance to every experts in the model
analysis/colossalai_replace/layer.py:405
Functioncalculate_activated_parameters
(row)
eval/plot_bigbench.py:75
Functioncalculate_num_tokens
(model_name)
eval/plot_bigbench.py:49
Functionconvert_size
(size_str)
eval/plot_bigbench.py:62
Functioncustom_sort_key
(item)
eval/result_retrieval_bigbenchlite.py:217
Methodforward
Args: inputs (torch.Tensor): The input tensor of shape (batch_size, seq_len, hidden_size) Returns: torch.Ten
analysis/colossalai_replace/layer.py:149
Methodget_chunk_slice
(idx: int, chunk_size: int)
analysis/colossalai_replace/layer.py:362