Skip to content

Update README.md #5

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

Merged
jeff-da merged 2 commits into from
Nov 22, 2024
Merged

Update README.md #5

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1b168e1
Update README.md
hughbzhang Nov 22, 2024
2ce27cc
Update README.md
hughbzhang Nov 22, 2024
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
6 changes: 4 additions & 2 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ pip3 install -e .
pip3 install -r requirements.txt
pip3 install transformers # likely you want the latest version of transformers
pip3 install flash-attn==2.5.8 --no-build-isolation
'''
```


## Running commands
Expand All @@ -31,4 +31,6 @@ Example for Llama-3-8B-Instruct
```
lm_eval --model vllm --model_args pretrained=meta-llama/Meta-Llama-3-8B-Instruct,tensor_parallel_size=8,dtype=auto,gpu_memory_utilization=0.8,data_parallel_size=1,trust_remote_code=True --batch_size auto --tasks gsm8k --seed 0 --gen_kwargs temperature=0.0 --output_path Meta-Llama-3-8B-Instruct_gsm8k_temperature=0.0_results.json --log_samples
lm_eval --model vllm --model_args pretrained=meta-llama/Meta-Llama-3-8B-Instruct,tensor_parallel_size=8,dtype=auto,gpu_memory_utilization=0.8,data_parallel_size=1,trust_remote_code=True --batch_size auto --tasks gsm1k --seed 0 --gen_kwargs temperature=0.0 --output_path Meta-Llama-3-8B-Instruct_gsm1k_temperature=0.0_results.json --log_samples
'''
```

**Note:** due to API changes (latest model API versions changing over time), nondeterminism of API reuslts, and slightly different library versions of the HuggingFace transformers libraries required to run each model, the precise results may not be fully exactly reproducible.
Loading