diff --git a/.github/workflows/benchmark.yml b/.github/workflows/benchmark.yml index d16a6c8a7f..3242451956 100644 --- a/.github/workflows/benchmark.yml +++ b/.github/workflows/benchmark.yml @@ -82,8 +82,8 @@ jobs: run: | python3.11 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing --quantize int8 | tee llama_int8.txt python3.11 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing --quantize nf4 | tee llama_nf4.txt - - name: Run LLaMA 7B on 4 (virtual) GPUs - run: python3.11 examples/llama.py --gen 1 --size 7B --shard 4 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_four_gpu.txt + #- name: Run LLaMA 7B on 4 (virtual) GPUs + # run: python3.11 examples/llama.py --gen 1 --size 7B --shard 4 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_four_gpu.txt - name: Run GPT2 run: | JIT=0 python3.11 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt