Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
license_name: minimax-h3-license
|
| 4 |
+
license_link: https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/LICENSE
|
| 5 |
+
base_model: MiniMaxAI/MiniMax-H3
|
| 6 |
+
tags: [svdquant, w4a4, nvfp4, video, text-to-video, quantized]
|
| 7 |
+
---
|
| 8 |
+
|
| 9 |
+
# MiniMax-H3 · SVDQuant NVFP4 (rank 32, RTN) — reference format
|
| 10 |
+
|
| 11 |
+
The RTX 50-series (sm_120) sibling of
|
| 12 |
+
[the int4 release](https://huggingface.co/ModelsLab/MiniMax-H3-svdquant-int4_r32):
|
| 13 |
+
e2m1 4-bit residual + bf16 rank-32 low-rank branch, quantized fresh from
|
| 14 |
+
BF16 (int4 and fp4 grids do not nest, so transcoding is never used).
|
| 15 |
+
|
| 16 |
+
**Status: reference (unpacked) format.** Runs through svdquant's
|
| 17 |
+
reference backend today; the packed fp4 export for the Blackwell
|
| 18 |
+
tensor-core kernels is in progress. RTN rounding (fp4-GPTQ pending).
|
| 19 |
+
Same license inheritance and credits as the int4 release.
|