Open diaojunxian opened 1 year ago
试试:先删掉该依赖库tqdm,单独安装好 macropodus后再装高版本的tqdm
试试:先删掉该依赖库tqdm,单独安装好 macropodus后再装高版本的tqdm
已经尝试过,虽然可以安装上,但是运行的时候,还是会提示 macropodus 需要低版本的 tqdm
不影响运行就行,其实该项目对哪个tqdm版本并无特殊需求。报错是包管理的要求,并不影响代码正常跑
不影响运行就行,其实该项目对哪个tqdm版本并无特殊需求。报错是包管理的要求,并不影响代码正常跑
会强制抛出异常,代码无法运行,😔
方便贴出来吗,python版本, 以及运行报错信息
@yongzhuo
这里的报错是 tqdm 版本过高:tqdm == 4.65.0
CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.so.11.0
CUDA SETUP: Highest compute capability among GPUs detected: 8.0
CUDA SETUP: Detected CUDA version 113
CUDA SETUP: Loading binary /opt/conda/lib/python3.10/site-packages/bitsandbytes/libbitsandbytes_cuda113.so...
2023-07-27 05:47:54,890 - seg_basic.py[line:19] - INFO: path of dict cache is /opt/conda/lib/python3.10/site-packages/macropodus/data/cache/macropodus.cache!
Traceback (most recent call last):
File "/home/cloudadmin/test-sample/test_chatglm/chatglm-maths/chatglm_maths/t10_toy_trl_train_ppo.py", line 31, in <module>
import macropodus
File "/opt/conda/lib/python3.10/site-packages/macropodus/__init__.py", line 12, in <module>
from macropodus.summarize import keyword, textrank, summarization
File "/opt/conda/lib/python3.10/site-packages/macropodus/summarize/__init__.py", line 11, in <module>
from macropodus.summarize.graph_base.textrank import TextRankSum, TextRankKey
File "/opt/conda/lib/python3.10/site-packages/macropodus/summarize/graph_base/textrank.py", line 8, in <module>
from macropodus.summarize.graph_base.textrank_word2vec import TextrankWord2vec
File "/opt/conda/lib/python3.10/site-packages/macropodus/summarize/graph_base/textrank_word2vec.py", line 8, in <module>
from macropodus.similarity.similarity_word2vec_char import SimW2vChar
File "/opt/conda/lib/python3.10/site-packages/macropodus/similarity/__init__.py", line 8, in <module>
from macropodus.similarity.similarity_word2vec_char import SimW2vChar
File "/opt/conda/lib/python3.10/site-packages/macropodus/similarity/similarity_word2vec_char.py", line 8, in <module>
from macropodus.base.word2vec import W2v
File "/opt/conda/lib/python3.10/site-packages/macropodus/base/word2vec.py", line 11, in <module>
import gensim
File "/opt/conda/lib/python3.10/site-packages/gensim/__init__.py", line 5, in <module>
from gensim import parsing, corpora, matutils, interfaces, models, similarities, summarization, utils # noqa:F401
File "/opt/conda/lib/python3.10/site-packages/gensim/corpora/__init__.py", line 12, in <module>
from .dictionary import Dictionary # noqa:F401
File "/opt/conda/lib/python3.10/site-packages/gensim/corpora/dictionary.py", line 11, in <module>
from collections import Mapping, defaultdict
ImportError: cannot import name 'Mapping' from 'collections' (/opt/conda/lib/python3.10/collections/__init__.py)
我降低 tqdm == 4.31.0
/home/cloudadmin/test-sample/test_chatglm/chatglm-maths
Traceback (most recent call last):
File "/home/cloudadmin/test-sample/test_chatglm/chatglm-maths/chatglm_maths/t10_toy_trl_train_ppo.py", line 25, in <module>
from trl import PPOConfig, AutoModelForCausalLMWithValueHead, create_reference_model
File "/opt/conda/lib/python3.10/site-packages/trl/__init__.py", line 5, in <module>
from .core import set_seed
File "/opt/conda/lib/python3.10/site-packages/trl/core.py", line 23, in <module>
from transformers import top_k_top_p_filtering
File "/opt/conda/lib/python3.10/site-packages/transformers/__init__.py", line 26, in <module>
from . import dependency_versions_check
File "/opt/conda/lib/python3.10/site-packages/transformers/dependency_versions_check.py", line 16, in <module>
from .utils.versions import require_version, require_version_core
File "/opt/conda/lib/python3.10/site-packages/transformers/utils/__init__.py", line 60, in <module>
from .hub import (
File "/opt/conda/lib/python3.10/site-packages/transformers/utils/hub.py", line 32, in <module>
from huggingface_hub import (
File "/opt/conda/lib/python3.10/site-packages/huggingface_hub/__init__.py", line 311, in __getattr__
submod = importlib.import_module(submod_path)
File "/opt/conda/lib/python3.10/importlib/__init__.py", line 126, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File "/opt/conda/lib/python3.10/site-packages/huggingface_hub/hf_api.py", line 44, in <module>
from ._commit_api import (
File "/opt/conda/lib/python3.10/site-packages/huggingface_hub/_commit_api.py", line 14, in <module>
from tqdm.contrib.concurrent import thread_map
ModuleNotFoundError: No module named 'tqdm.contrib'
很奇怪的bug,python10么。冲突还比较多,gensim==3.7.1是不能换的,你试试tqdm==4.49.0呢
4.49.0
是 python 3.10
CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.so.11.0
CUDA SETUP: Highest compute capability among GPUs detected: 8.0
CUDA SETUP: Detected CUDA version 113
CUDA SETUP: Loading binary /opt/conda/lib/python3.10/site-packages/bitsandbytes/libbitsandbytes_cuda113.so...
2023-07-27 06:23:12,857 - seg_basic.py[line:19] - INFO: path of dict cache is /opt/conda/lib/python3.10/site-packages/macropodus/data/cache/macropodus.cache!
Traceback (most recent call last):
File "/home/cloudadmin/test-sample/test_chatglm/chatglm-maths/chatglm_maths/t10_toy_trl_train_ppo.py", line 31, in
python10版本太高了,该项目只测过3.7-3.9。 把/opt/conda/lib/python3.10/site-packages/gensim/corpora/dictionary.py中 from collections import Mapping, defaultdict 改为 from collections.abc import Mapping from collections import defaultdict 试试呢
python10版本太高了,该项目只测过3.7-3.9。 把/opt/conda/lib/python3.10/site-packages/gensim/corpora/dictionary.py中 from collections import Mapping, defaultdict 改为 from collections.abc import Mapping from collections import defaultdict 试试呢
我降低 python 版本到 3.9 可以执行了,但是还是有奇怪报错。
已经降低 peft == 0.2.0 transformers==4.28.0
ake sure to set config.batch_size to the correct value before training.
warnings.warn(
tqdm: 0%| | 0/37 [00:07<?, ?it/s]
Traceback (most recent call last):
File "/home/cloudadmin/test-sample/test_chatglm/chatglm-maths/chatglm_maths/t10_toy_trl_train_ppo.py", line 210, in <module>
response_text = tokenizer.decode(response_ids)
File "/opt/conda/envs/3.9/lib/python3.9/site-packages/transformers/tokenization_utils_base.py", line 3486, in decode
return self._decode(
File "/home/cloudadmin/test-sample/test_chatglm/chatglm-maths/chatglm_maths/models/tokenization_chatglm.py", line 284, in _decode
return super()._decode(token_ids, **kwargs)
File "/opt/conda/envs/3.9/lib/python3.9/site-packages/transformers/tokenization_utils.py", line 931, in _decode
filtered_tokens = self.convert_ids_to_tokens(token_ids, skip_special_tokens=skip_special_tokens)
File "/opt/conda/envs/3.9/lib/python3.9/site-packages/transformers/tokenization_utils.py", line 906, in convert_ids_to_tokens
index = int(index)
TypeError: int() argument must be a string, a bytes-like object or a number, not 'list'
那就是Bug, response_ids = response_tensor.detach().cpu().numpy().tolist()改为 response_ids = response_tensor.detach().cpu().numpy().tolist()[0]
model.cuda() 改为 model.half().cuda()
outputs = model(torch.cat([start_ids, end_ids], dim=-1))改为 outputs = model(torch.cat([start_ids, end_ids], dim=-1).cuda())
几个样例是能跑完的,日志: [tensor(0.1491, dtype=torch.float64)] tqdm: 97%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████▉ | 36/37 [05:17<00:08, 8.76s/it][115, 123, 119860, 5, 19052, 5, 123, 123, 119860, 194, 5, 123, 123, 5, 120, 115, 5, 115, 123, 115, 123, 5, 123, 115, 115, 115, 115, 123, 5, 123, 115, 355, 410, 410, 115, 5, 410, 5, 5, 115, 5, 115, 5327, 5327, 115, 5327, 5327, 5327, 5327, 5327, 5327, 5327, 115, 123, 115, 115, 5327, 5327, 5327, 115, 5327, 355, 64035, 5327, 65312, 5, 5327, 5327, 5327, 5327, 5327, 65312, 63917, 5327, 5, 5327, 5, 5327, 5, 65312, 5, 65312, 65312, 65312, 5, 63917, 5327, 5, 65312, 5, 5, 5, 5327, 5, 65312, 5, 65312, 5, 5, 115, 65312, 5, 5, 5, 174, 65312, 5, 5, 5, 5, 5327, 5, 5, 5, 5, 5327, 5, 5, 65312, 5, 5, 65312, 5, 5, 65312, 5, 5, 5] [tensor(0.1409, dtype=torch.float64)] tqdm: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 37/37 [05:26<00:00, 8.82s/it] 2023-07-27 16:09:00,568 - t10_toy_trl_train_ppo.py[line:127] - INFO: **model_save_path is ./fine_tuning_t10/ppo/pytorch_model.bin**
安装的时候有警告
虽然安装好了,执行代码的时候,会提示 tqdm 被其他库依赖的时候,需要更高版本,导致代码无法正常编译执行。