tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity

defaults

Browse Source
This commit is contained in:
Dan Saunders
2025-09-19 16:34:45 +00:00
parent d2f1e23bcd
commit 336616d659
1 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
scripts/bench_moe.py
View File

@@ -74,11 +74,11 @@ def load_hf_block(
def main() -> None:
p = argparse.ArgumentParser(description="Qwen2 MoE grouped_mm benchmark")
p.add_argument("--bsz", type=int, default=8)
p.add_argument("--seq", type=int, default=1024)
p.add_argument("--hidden", type=int, default=4096)
p.add_argument("--inter", type=int, default=14336)
p.add_argument("--experts", type=int, default=32)
p.add_argument("--top_k", type=int, default=4)
p.add_argument("--seq", type=int, default=512)
p.add_argument("--hidden", type=int, default=1024)
p.add_argument("--inter", type=int, default=2948)
p.add_argument("--experts", type=int, default=8)
p.add_argument("--top_k", type=int, default=2)
p.add_argument("--dtype", choices=["bf16", "fp16", "fp32"], default="bf16")
p.add_argument("--iters", type=int, default=50)
p.add_argument("--warmup", type=int, default=10)