[Bug] 测试训练的baichuan2遇到的问题

Question

[Bug] 测试训练的baichuan2遇到的问题

tuqingwen opened this issue 3 months ago · comments

tuqingwen commented 3 months ago

Prerequisite

I have searched Issues and Discussions but cannot get the expected help.
The bug has not been fixed in the latest version.

Type

I'm evaluating with the officially supported tasks/models/datasets.

Environment

4090
torch 2.0.0+cu118
系统：windows

Reproduces the problem - code/configuration sample

Evaluating my own model.

Reproduces the problem - command or script

python run.py --datasets ceval_gen_5f30c7 --models baichuan2-cahds-2k-new-12 --debug

Reproduces the problem - error message

bin D:\Anaconda\envs\Belle\lib\site-packages\bitsandbytes\libbitsandbytes_cuda118.dll
02/28 22:20:15 - OpenCompass - INFO - Loading ceval_gen_5f30c7: configs\datasets\ceval\ceval_gen_5f30c7.py
02/28 22:20:15 - OpenCompass - INFO - Loading baichuan2-cahds-2k-new-12: configs\models\baichuan\baichuan2-cahds-2k-new-12.py
02/28 22:20:15 - OpenCompass - INFO - Loading example: configs\summarizers\example.py
02/28 22:20:15 - OpenCompass - WARNING - SlurmRunner is not used, so the partition argument is ignored.
02/28 22:20:15 - OpenCompass - DEBUG - Modules of opencompass's partitioner registry have been automatically imported from opencompass.partitioners
02/28 22:20:15 - OpenCompass - DEBUG - Get class SizePartitioner from "partitioner" registry in "opencompass"
02/28 22:20:15 - OpenCompass - DEBUG - An SizePartitioner instance is built from registry, and its implementation can be found in opencompass.partitioners.size
02/28 22:20:15 - OpenCompass - DEBUG - Key eval.runner.task.judge_cfg not found in config, ignored.
02/28 22:20:15 - OpenCompass - DEBUG - Key eval.runner.task.dump_details not found in config, ignored.
02/28 22:20:15 - OpenCompass - DEBUG - Additional config: {}
02/28 22:20:15 - OpenCompass - INFO - Partitioned into 1 tasks.
02/28 22:20:15 - OpenCompass - DEBUG - Task 0: [baichuan2-cahds-2k-new-12/ceval-computer_network]
02/28 22:20:15 - OpenCompass - DEBUG - Modules of opencompass's runner registry have been automatically imported from opencompass.runners
02/28 22:20:15 - OpenCompass - DEBUG - Get class LocalRunner from "runner" registry in "opencompass"
02/28 22:20:15 - OpenCompass - DEBUG - An LocalRunner instance is built from registry, and its implementation can be found in opencompass.runners.local
02/28 22:20:15 - OpenCompass - DEBUG - Modules of opencompass's task registry have been automatically imported from opencompass.tasks
02/28 22:20:15 - OpenCompass - DEBUG - Get class OpenICLInferTask from "task" registry in "opencompass"
02/28 22:20:15 - OpenCompass - DEBUG - An OpenICLInferTask instance is built from registry, and its implementation can be found in opencompass.tasks.openicl_infer
NOTE: Redirects are currently not supported in Windows or MacOs.
[W C:\actions-runner_work\pytorch\pytorch\builder\windows\pytorch\torch\csrc\distributed\c10d\socket.cpp:601] [c10d] The client socket has failed to connect to [DESKTOP-LURV0G3]:27665 (system error: 10049 - 在其上下文中，该请求的地无效。).
[W C:\actions-runner_work\pytorch\pytorch\builder\windows\pytorch\torch\csrc\distributed\c10d\socket.cpp:601] [c10d] The client socket has failed to connect to [DESKTOP-LURV0G3]:27665 (system error: 10049 - 在其上下文中，该请求的地无效。).
bin D:\Anaconda\envs\Belle\lib\site-packages\bitsandbytes\libbitsandbytes_cuda118.dll
02/28 22:20:20 - OpenCompass - INFO - Task [baichuan2-cahds-2k-new-12/ceval-computer_network]
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 4/4 [00:05<00:00, 1.33s/it]
02/28 22:20:28 - OpenCompass - INFO - Start inferencing [baichuan2-cahds-2k-new-12/ceval-computer_network]
100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 19/19 [00:00<?, ?it/s]
[2024-02-28 22:20:28,776] [opencompass.openicl.icl_inferencer.icl_gen_inferencer] [INFO] Starting inference process...
0%| | 0/3 [00:00<?, ?it/s]D
:\Anaconda\envs\Belle\lib\site-packages\transformers\generation\configuration_utils.py:389: UserWarning: do_sample is set to False. However, temperature is set to 0.3 -- this flag is only used in sample-based generation modes. You should set do_sample=True or unset temperature.
warnings.warn(
D:\Anaconda\envs\Belle\lib\site-packages\transformers\generation\configuration_utils.py:394: UserWarning: do_sample is set to False. However, top_p is set to 0.85 -- this flag is only used in sample-based generation modes. You should set do_sample=True or unset top_p.
warnings.warn(
D:\Anaconda\envs\Belle\lib\site-packages\transformers\generation\configuration_utils.py:404: UserWarning: do_sample is set to False. However, top_k is set to 5 -- this flag is only used in sample-based generation modes. You should set do_sample=True or unset top_k.
warnings.warn(
100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:02<00:00, 1.46it/s]
02/28 22:20:30 - OpenCompass - INFO - time elapsed: 10.31s
02/28 22:20:35 - OpenCompass - DEBUG - Get class NaivePartitioner from "partitioner" registry in "opencompass"
02/28 22:20:35 - OpenCompass - DEBUG - An NaivePartitioner instance is built from registry, and its implementation can be found in opencompass.partitioners.naive
02/28 22:20:35 - OpenCompass - DEBUG - Key eval.runner.task.judge_cfg not found in config, ignored.
02/28 22:20:35 - OpenCompass - DEBUG - Key eval.runner.task.dump_details not found in config, ignored.
02/28 22:20:35 - OpenCompass - DEBUG - Additional config: {'eval': {'runner': {'task': {}}}}
02/28 22:20:35 - OpenCompass - INFO - Partitioned into 1 tasks.
02/28 22:20:35 - OpenCompass - DEBUG - Task 0: [baichuan2-cahds-2k-new-12/ceval-computer_network]
02/28 22:20:35 - OpenCompass - DEBUG - Get class LocalRunner from "runner" registry in "opencompass"
02/28 22:20:35 - OpenCompass - DEBUG - An LocalRunner instance is built from registry, and its implementation can be found in opencompass.runners.local
02/28 22:20:35 - OpenCompass - DEBUG - Get class OpenICLEvalTask from "task" registry in "opencompass"
02/28 22:20:35 - OpenCompass - DEBUG - An OpenICLEvalTask instance is built from registry, and its implementation can be found in opencompass.tasks.openicl_eval
系统找不到文件 C:\Users\CUMT\AppData\Local\Microsoft\WindowsApps\python3.exe。
02/28 22:20:37 - OpenCompass - DEBUG - An DefaultSummarizer instance is built from registry, and its implementation can be found in opencompass.summarizers.default
dataset version metric mode baichuan2-cahds-2k-new-12

ceval-computer_network - - - -
02/28 22:20:37 - OpenCompass - INFO - write summary to F:\opencompass\outputs\default\20240228_222015\summary\summary_20240228_222015.txt
02/28 22:20:37 - OpenCompass - INFO - write csv to F:\opencompass\outputs\default\20240228_222015\summary\summary_20240228_222015.csv

系统是windows

Other information

No response

Songyang Zhang · Answer 1 · Wed Feb 28 2024 22:26:52 GMT+0800 (China Standard Time)

OpenCompass is not compatible with Windows.