biondizzle 04245b664b Add warmup-based activation global scale computation in finalize_weights ...

The checkpoint input_scale is a calibration value that produces wrong gs
at runtime (too small → block scales saturate → garbage output → EOS).

Now calls compute_activation_global_scales() with sample data during weight
finalization, before cudagraph capture. This observes actual activation
magnitudes and computes correct L1 and L2 gs values.

2026-05-17 10:48:24 +00:00

..

patches

Add warmup-based activation global scale computation in finalize_weights

2026-05-17 10:48:24 +00:00

nvfp4_cutedsl.py

Fix: return 2D scale tensor for GEMM (shape[1] access)

2026-05-17 09:59:57 +00:00