biondizzle
a0bacb3cf6
- New scripts/quantize_nvfp4.py: runs full ModelOpt pipeline in-process - Saves calibrated state after calibration (insurance against export crashes) - Patches modelopt for V4: ModuleList quantizers, stale GPU tensor safety - --export-only flag to retry export from saved calibration state - Removed old model_opt_nvfp4_full.py (shell wrapper) - Updated README with new pipeline docs and bug #5/#6
11 lines
137 B
Plaintext
11 lines
137 B
Plaintext
# Dequantized BF16 weights (3TB)
|
|
DeepSeek-V4-Pro-BF16/
|
|
|
|
# Calibration state (huge, not for git)
|
|
*.pt
|
|
|
|
# Python
|
|
__pycache__/
|
|
*.pyc
|
|
.venv/
|