github/tinygrad
1
1
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-04-29 03:00:14 -04:00
Code Issues Packages Projects Releases Wiki Activity

use BEAM=2 instead of BEAM=4 in cuda ci gpt2 (#3089)

Browse Source 
BEAM=2 is faster and less search time. investigating why BEAM2+BEAM4 is slower than BEAM2 alone
This commit is contained in:
chenyu
2024-01-11 13:21:06 -05:00
committed by GitHub
parent f502c9b08f
commit 93e3f952aa
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
.github/workflows/benchmark.yml vendored
View File

@@ -87,7 +87,7 @@ jobs:
- name: Run GPT2 w HALF
run: CUDA=1 JIT=1 HALF=1 python3 examples/gpt2.py --count 10 --temperature 0 --timing
- name: Run GPT2 w HALF/BEAM
run: CUDA=1 JIT=1 HALF=1 BEAM=4 CACHELEVEL=0 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half_beam.txt
run: CUDA=1 JIT=1 HALF=1 BEAM=2 CACHELEVEL=0 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half_beam.txt
- uses: actions/upload-artifact@v4
with:
name: Speed (NVIDIA)