xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-04-12 21:17:12 +08:00
Code Issues Packages Projects Releases Wiki Activity

Revert model change

Browse Source 
Signed-off-by: knlnguyen1802 <knlnguyen1802@gmail.com>
This commit is contained in:
knlnguyen1802 2025-11-13 10:38:53 +08:00
parent bdf34c1265
commit c594843112
1 changed files with 2 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
examples/offline_inference/rlhf_colocate.py
View File

@ -79,9 +79,7 @@ class RayTrainingActor:
# Ray sets CUDA_VISIBLE_DEVICES to the GPUs assigned to this actor.
from transformers import AutoModelForCausalLM
self.model = AutoModelForCausalLM.from_pretrained(
"/mnt/nvme3n1/models/qwen2.5_7B"
)
self.model = AutoModelForCausalLM.from_pretrained("facebook/opt-125m")
self.model.to("cuda:0")
# Zero out all the parameters.
for name, p in self.model.named_parameters():
@ -306,7 +304,7 @@ def setup_train_cluster():
placement_group_capture_child_tasks=True,
),
)(MyLLM).remote(
model="/mnt/nvme3n1/models/qwen2.5_7B",
model="facebook/opt-125m",
enforce_eager=True,
worker_extension_cls="rlhf_utils.ColocateWorkerExtension",
tensor_parallel_size=2,