Update README.md
Browse files
README.md
CHANGED
|
@@ -6,6 +6,8 @@ tags:
|
|
| 6 |
- df11
|
| 7 |
- lossless compression
|
| 8 |
- 70% size, 100% accuracy
|
|
|
|
|
|
|
| 9 |
language:
|
| 10 |
- en
|
| 11 |
base_model:
|
|
@@ -18,4 +20,7 @@ The bf16 layers are compressed using the Dfloat11 lossless compression mixed wit
|
|
| 18 |
|
| 19 |
This isnt the perfect balance between nvfp4 layers and Dfloat11 compressed layers and changes a good amount from model to model but it is a start.
|
| 20 |
|
| 21 |
-
`flux-2-klein-4b-nvfp4_nvfp4_dfloat11.safetensors`
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
- df11
|
| 7 |
- lossless compression
|
| 8 |
- 70% size, 100% accuracy
|
| 9 |
+
- 86% size 100% accuracy
|
| 10 |
+
- 74.4% size 100% accuracy
|
| 11 |
language:
|
| 12 |
- en
|
| 13 |
base_model:
|
|
|
|
| 20 |
|
| 21 |
This isnt the perfect balance between nvfp4 layers and Dfloat11 compressed layers and changes a good amount from model to model but it is a start.
|
| 22 |
|
| 23 |
+
`flux-2-klein-4b-nvfp4_nvfp4_dfloat11.safetensors`
|
| 24 |
+
|
| 25 |
+
Other models I have done get 86% size 100% accuracy doing plain Dfloat11 compression.
|
| 26 |
+
and around 74.4% size 100% accuracy doing nvfp4 mixed with Dfloat11 compression.
|