github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-06-26 23:38:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix typ and parameter description

Browse Source
This commit is contained in:
YeAnbang 2025-06-05 15:41:14 +08:00
parent 0d008110e7
commit 96faf54542
1 changed files with 6 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
applications/ColossalChat/rl_example.py
View File

@ -126,25 +126,25 @@ if __name__ == "__main__":
"--tensor-parallel-size",
type=int,
default=1,
help="Tensor parallel size for the inference backend. Please check the generation arguments documentation for your backend.",
help="Tensor parallel size for the trainer (consumer). Please check the generation arguments documentation for your backend.",
)
parser.add_argument(
"-pp",
"--pipeline-parallel-size",
type=int,
default=1,
help="Pipeline parallel size for the inference backend. Please check the generation arguments documentation for your backend.",
help="Pipeline parallel size for the trainer (consumer). Please check the generation arguments documentation for your backend.",
)
parser.add_argument(
"-zero",
"--zero-stage",
type=int,
default=0,
help="Zero stage for the inference backend. Please check the generation arguments documentation for your backend.",
help="Zero stage for the trainer (consumer). Please check the generation arguments documentation for your backend.",
)
parser.add_argument(
"-ptp",
"--produce-tensor-parallel-size",
"--producer-tensor-parallel-size",
type=int,
default=1,
help="Tensor parallel size for the producer. Please check the generation arguments documentation for your backend.",
@ -206,7 +206,7 @@ if __name__ == "__main__":
enforce_eager=True,
enable_chunked_prefill=True,
max_model_len=args.max_new_tokens + args.max_prompt_tokens,
tensor_parallel_size=args.produce_tensor_parallel_size,
tensor_parallel_size=args.producer_tensor_parallel_size,
)
)
generate_config.update(
@ -276,7 +276,7 @@ if __name__ == "__main__":
launch_distributed(
num_producers=args.num_inferencer,
num_proc_per_producer=inference_model_config.get("tensor_parallel_size", args.produce_tensor_parallel_size),
num_proc_per_producer=inference_model_config.get("tensor_parallel_size", args.producer_tensor_parallel_size),
num_consumer_procs=args.num_trainers,
num_episodes=args.num_episodes,
inference_batch_size=args.inference_batch_size,