github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-08-17 23:46:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

tested after rebasing, fix importance sampling bug

Browse Source
This commit is contained in:
YeAnbang 2025-08-06 06:15:15 +00:00
parent 3746f73854
commit 32b2148670
4 changed files with 4 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
applications/ColossalChat/examples/requirements.txt
View File

@ -1,4 +1,4 @@
pandas>=1.4.1 pandas>=1.4.1
sentencepiece sentencepiece
colossalai==0.4.7 colossalai>=0.4.7
prompt_toolkit prompt_toolkit

2
applications/ColossalChat/requirements.txt
View File

@ -1,4 +1,4 @@
transformers==4.39.3 transformers>=4.39.3
tqdm tqdm
datasets==2.14.7 datasets==2.14.7
loralib loralib

2
applications/ColossalChat/rl_example.py
View File

@ -152,7 +152,7 @@ if __name__ == "__main__":
"-ei", "-ei",
"--eval-interval", "--eval-interval",
type=int, type=int,
default=100, default=-1,
help="Interval for evaluation. Evaluate every ei training steps.", help="Interval for evaluation. Evaluate every ei training steps.",
) )
parser.add_argument( parser.add_argument(

1
colossalai/shardformer/modeling/qwen3.py
View File

@ -273,6 +273,7 @@ class Qwen3PipelineForwards:
hidden_states: Optional[torch.FloatTensor] = None, hidden_states: Optional[torch.FloatTensor] = None,
stage_index: Optional[List[int]] = None, stage_index: Optional[List[int]] = None,
shard_config: ShardConfig = None, shard_config: ShardConfig = None,
**kwargs
): ):
r""" r"""
Args: Args: