Add the moss-image-quant collection
Browse files
README.md
CHANGED
|
@@ -20,5 +20,8 @@ pinned: false
|
|
| 20 |
- [**pplx-embed-v2-late ONNX: index with 9B, query with 0.6Bs**](https://huggingface.co/collections/nynefold/pplx-embed-v2-late-onnx-index-with-9b-query-with-06b-6ae86f55b8e45d57841eb425):
|
| 21 |
the late-interaction pplx-embed-v2 0.6B and 9B; they share one token space, so index with the 9B
|
| 22 |
and query with the 0.6B.
|
|
|
|
|
|
|
|
|
|
| 23 |
|
| 24 |
Everything here is open. Questions and suggestions are welcome in each repo's Community tab.
|
|
|
|
| 20 |
- [**pplx-embed-v2-late ONNX: index with 9B, query with 0.6Bs**](https://huggingface.co/collections/nynefold/pplx-embed-v2-late-onnx-index-with-9b-query-with-06b-6ae86f55b8e45d57841eb425):
|
| 21 |
the late-interaction pplx-embed-v2 0.6B and 9B; they share one token space, so index with the 9B
|
| 22 |
and query with the 0.6B.
|
| 23 |
+
- [**moss-image-quant: MOSS-Image-SFT GGUF quants**](https://huggingface.co/collections/nynefold/moss-image-quant-moss-image-sft-gguf-quants-6afcc4529187d7a33c68ec40):
|
| 24 |
+
37 GGUF quant recipes of the MOSS-Image-SFT text-to-image model, each measured against FP32 on the same
|
| 25 |
+
prompts and seeds; the 13.8 GB pick keeps FP32's pictures, the 11.4 GB one its quality.
|
| 26 |
|
| 27 |
Everything here is open. Questions and suggestions are welcome in each repo's Community tab.
|