Skip to content

Update num_proc handling for vllm and Ray mode #973

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ArdalanM wants to merge 1 commit into
base: main
Choose a base branch
from
+5 −4
Open

Update num_proc handling for vllm and Ray mode #973

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 5 additions & 4 deletions data_juicer/ops/mapper/text_tagging_by_prompt_mapper.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -114,7 +114,11 @@ def __init__(
""" # noqa

super().__init__(*args, **kwargs)
self.num_proc = 1
if not enable_vllm or not is_ray_mode():
# In Ray mode with vllm, num_proc is left to the caller so Ray can
# schedule one actor per GPU for data parallelism.
self.num_proc = 1

Comment thread
ArdalanM marked this conversation as resolved.

self.prompt = prompt
self.tag_list = tag_list
Expand All @@ -129,9 +133,6 @@ def __init__(
sampling_params = update_sampling_params(sampling_params, hf_model, self.enable_vllm)

if enable_vllm:
if not is_ray_mode():
# cannot initialize vllm replicas on different GPUs
self.num_proc = 1
self.model_key = prepare_model(
model_type="vllm",
pretrained_model_name_or_path=hf_model,
Expand Down
Loading