Skip to content

Latest commit

 

History

History
1 lines (1 loc) · 232 Bytes

README.md

File metadata and controls

1 lines (1 loc) · 232 Bytes

python run_llama_int8_0307.py -m Llama-2-7b-hf --quantized-model-path "./saved_results_w8a8/best_model.pt" --benchmark --int8 --num-iter 1 --num-warmup 0 --profile --max-new-tokens 10 --corenums 64 --greedy --batch-size 1 --jit