biondizzle 7a4403fa98 feat: FP4 staging kernel - BF16 → E2M1 packed + UE4M3 block16 scales ...

mxf4nvf4 requires FP4×FP4, not FP8×FP4.
- New staging kernel: E2M1 nearest-neighbor quantization
- Output: uint8 packed (2 E2M1 per byte) + UE4M3 packed int32 scales
- Added CUDA sync diagnostics for error localization

2026-05-11 20:29:36 +00:00

..

deepseek_v4.py

feat: FP4 staging kernel - BF16 → E2M1 packed + UE4M3 block16 scales

2026-05-11 20:29:36 +00:00

deepseek_v4.py.bak

S11: Fixed substr mapping, stacking, suffix, and o_a_proj - loads weights but attention forward uses FP8 einsum incompatible with NVFP4

2026-05-10 17:45:53 +00:00

deepseek_v4.py.s11

S11: Fixed substr mapping, stacking, suffix, and o_a_proj - loads weights but attention forward uses FP8 einsum incompatible with NVFP4

2026-05-10 17:45:53 +00:00

patch_finegrained_fp8_blackwell.py

Add BF16 upcast script and Blackwell DeepGEMM patch

2026-05-07 14:25:30 +00:00

patch_vllm_weights.py

vLLM serving: patched deepseek_v4.py, disabled mega_moe, updated docs

2026-05-10 16:14:17 +00:00

quant_module_patched.py

Add ModelOpt NVFP4 pipeline: patch, run script, README

2026-05-07 07:22:54 +00:00

staging_kernel.py

feat: FP4 staging kernel - BF16 → E2M1 packed + UE4M3 block16 scales

2026-05-11 20:29:36 +00:00