Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: gemma
|
| 3 |
+
base_model:
|
| 4 |
+
- google/gemma-3-27b-it
|
| 5 |
+
tags:
|
| 6 |
+
- gemma
|
| 7 |
+
- google
|
| 8 |
+
- instruct
|
| 9 |
+
- llm
|
| 10 |
+
- gguf
|
| 11 |
+
- GGUF
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
# Quant Infos (Uploading in progress ETA 25mins)
|
| 15 |
+
|
| 16 |
+
- Requires llama.cpp [b4875](https://github.com/ggml-org/llama.cpp/releases/tag/b4875)
|
| 17 |
+
- LLM ONLY (No vision support)
|
| 18 |
+
- quants done with an importance matrix for improved quantization loss
|
| 19 |
+
- Quantized ggufs & imatrix from hf bf16, through bf16. `safetensors bf16 -> gguf bf16 -> quant` for *optimal* quant loss.
|
| 20 |
+
- Wide coverage of different gguf quant types from Q\_8\_0 down to IQ1\_S (WIP)
|
| 21 |
+
- Imatrix generated with [this](https://gist.github.com/bartowski1182/eb213dccb3571f863da82e99418f81e8) multi-purpose dataset by [bartowski](https://huggingface.co/bartowski).
|
| 22 |
+
```
|
| 23 |
+
./imatrix -m $model_name-bf16.gguf -f calibration_datav3.txt -o $model_name.imatrix
|
| 24 |
+
```
|