Commit ·
b9cad69
1
Parent(s): 122e4f1
docs: link the arXiv paper (2611.15671) (#1)
Browse files- docs: link the arXiv paper (2611.15671) (5d3d4c0499586a4d3d8afe403038b61fc823a61c)
README.md
CHANGED
|
@@ -12,6 +12,8 @@ FP8 W8A8, built with llm-compressor, for vLLM. A quantized build of Cloudflare's
|
|
| 12 |
|
| 13 |
These builds keep the joint schema head, the vision tower, the norms and the Gated DeltaNet gate projections in BF16. The recipes are in the repository.
|
| 14 |
|
|
|
|
|
|
|
| 15 |
Measurements, recipes and per-record outputs: https://github.com/Level5Tech/clef-quant-bench
|
| 16 |
|
| 17 |
The base model is under Apache-2.0. Its LICENSE is included here, and NOTICE says these weights were modified by quantization.
|
|
|
|
| 12 |
|
| 13 |
These builds keep the joint schema head, the vision tower, the norms and the Gated DeltaNet gate projections in BF16. The recipes are in the repository.
|
| 14 |
|
| 15 |
+
Paper: https://arxiv.org/abs/2611.15671
|
| 16 |
+
|
| 17 |
Measurements, recipes and per-record outputs: https://github.com/Level5Tech/clef-quant-bench
|
| 18 |
|
| 19 |
The base model is under Apache-2.0. Its LICENSE is included here, and NOTICE says these weights were modified by quantization.
|