Add organization card with flint-nvfp4-coldgptq

README.md

@@ -1,10 +1,26 @@
 ---
 title: README
-emoji: 🐠
-colorFrom: green
-colorTo: yellow
+colorFrom: indigo
+colorTo: blue
 sdk: static
 pinned: false
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# FutureMLS Lab
+
+## Models
+
+| Model | Base model | Format | Size |
+|---|---|---|---:|
+| [flint-nvfp4-coldgptq](https://huggingface.co/FutureMLS-Lab/flint-nvfp4-coldgptq) | [RecursiveSI/flint](https://huggingface.co/RecursiveSI/flint) | NVFP4 (W4A4) routed experts, BF16 elsewhere | 23.5 GB (BF16: 79.6 GB) |
+
+### flint-nvfp4-coldgptq
+
+B300, SGLang 0.5.16 with `flint_sglang` 0.1.4, the model's default sampling. Generation length counts completion tokens,
+reasoning included.
+
+| Benchmark | Accuracy, BF16 → NVFP4 | Generation length p50 / p90 (tokens), BF16 → NVFP4 |
+|---|---:|---:|
+| GSM8K, 1,319 questions | 96.1 → 96.0 | 451 / 2,796 → 459 / 3,004 |
+| GPQA Diamond, 198 questions × 3 samples | 69.2 → 68.9 | 6,718 / 30,918 → 7,440 / 30,553 |
+| AIME 2026, 30 problems × 16 samples | 84.4 → 83.8 | 7,927 / 32,768 → 8,294 / 32,768 |