lukealonso commited on
Commit
29f45f9
·
verified ·
1 Parent(s): 4466515

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +0 -2
README.md CHANGED
@@ -45,8 +45,6 @@ Calibration datasets:
45
 
46
  All 512 routed experts per layer were activated during calibration — no experts were missed entirely. Post-calibration, 256 out of ~184k routed expert quantizers (~0.14%) had their amaxes floored to median/10 of their peer group to stabilize rarely-hit experts. Gate/up projection weight scales were tied for fused w13 export compatibility.
47
 
48
- Compared to [nvidia/Qwen3.5-397B-A17B-NVFP4](https://huggingface.co/nvidia/Qwen3.5-397B-A17B-NVFP4) (calibrated on cnn_dailymail + Nemotron), the input activation scales in this checkpoint show significantly higher entropy per layer (avg 3.92 vs 2.35 bits across all 60 layers), indicating more differentiated per-expert scales. This suggests the diverse calibration mix produced scales that better reflect each expert's actual activation range, rather than collapsing toward similar values from homogeneous calibration data.
49
-
50
  ### Quality
51
 
52
  MMLU Pro results:
 
45
 
46
  All 512 routed experts per layer were activated during calibration — no experts were missed entirely. Post-calibration, 256 out of ~184k routed expert quantizers (~0.14%) had their amaxes floored to median/10 of their peer group to stabilize rarely-hit experts. Gate/up projection weight scales were tied for fused w13 export compatibility.
47
 
 
 
48
  ### Quality
49
 
50
  MMLU Pro results: