ros/LLaMA-Factory
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[misc] fix packing and eval plot (#7623)

Browse Source
This commit is contained in:
hoshi-hiyouga
2025-04-07 18:20:57 +08:00
committed by GitHub
parent 5115dc8c7f
commit c3c0efbaa0
70 changed files with 288 additions and 194 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
scripts/vllm_infer.py
View File

@@ -20,7 +20,7 @@ from transformers import Seq2SeqTrainingArguments
from llamafactory.data import get_dataset, get_template_and_fix_tokenizer
from llamafactory.extras.constants import IGNORE_INDEX
from llamafactory.extras.misc import check_version, get_device_count
from llamafactory.extras.misc import get_device_count
from llamafactory.extras.packages import is_vllm_available
from llamafactory.hparams import get_infer_args
from llamafactory.model import load_tokenizer
@@ -56,7 +56,6 @@ def vllm_infer(
Usage: python vllm_infer.py --model_name_or_path meta-llama/Llama-2-7b-hf --template llama --dataset alpaca_en_demo
"""
check_version("vllm>=0.4.3,<=0.8.2")
if pipeline_parallel_size > get_device_count():
raise ValueError("Pipeline parallel size should be smaller than the number of gpus.")