Skip to content

Commit

Permalink
Added missing nvidia-ml-py package (#558)
Browse files Browse the repository at this point in the history
  • Loading branch information
tgaddair authored Jul 26, 2024
1 parent 15a38d5 commit d1a4d09
Show file tree
Hide file tree
Showing 5 changed files with 1,570 additions and 1,096 deletions.
4 changes: 2 additions & 2 deletions server/Makefile
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@ gen-server:

install: gen-server
pip install pip --upgrade
pip install torch==2.2.0
pip install torch==2.4.0
pip install -r requirements.txt
pip install -e ".[bnb, accelerate, quantize, peft, outlines]"

Expand All @@ -31,7 +31,7 @@ run-dev:
# SAFETENSORS_FAST_GPU=1 python -m torch.distributed.run --nproc_per_node=1 lorax_server/cli.py serve flozi00/Mistral-7B-german-assistant-v5-4bit-autogptq --quantize gptq

export-requirements:
poetry export -o requirements.txt -E bnb --without-hashes
poetry export -o requirements.txt --without-hashes

format:
pip install ruff
Expand Down
4 changes: 2 additions & 2 deletions server/lorax_server/utils/paged_attention.py
Original file line number Diff line number Diff line change
Expand Up @@ -12,8 +12,8 @@
import vllm._custom_ops as ops
except Exception as e:
raise ImportError(
f"Could not import vllm paged attention. Make sure your installation is correct. Complete error: {e}"
)
f"Could not import vllm paged attention. Make sure your installation is correct. Error: {e}"
) from e

# TODO(travis): fix for CUDA 8.9 (Lovelace) and 9.0 (Hopper)
# if torch.cuda.is_available():
Expand Down
Loading

0 comments on commit d1a4d09

Please sign in to comment.