Update README.md
Browse files
README.md
CHANGED
|
@@ -28,7 +28,7 @@ PROUDLY PRESENTS
|
|
| 28 |
|
| 29 |
<b>Quantization Note: Use repetition penalty (--repeat-penalty on llama.cpp) of 1.05 - 1.15 for best results </b>
|
| 30 |
|
| 31 |
-
Quantized from fp16.
|
| 32 |
* Weighted quantizations were creating using fp16 GGUF and [groups_merged-enhancedV2-TurboMini.txt](https://github.com/ggerganov/llama.cpp/discussions/5263#discussioncomment-9432658) in 189 chunks and n_ctx=512
|
| 33 |
* This method of calculating the importance matrix showed improvements in some areas for Mistral 7b and Llama3 8b models, see above post for details
|
| 34 |
* The enhancedv2-turbomini file appends snippets from turboderp's calibration data to the standard groups_merged.txt file
|
|
|
|
| 28 |
|
| 29 |
<b>Quantization Note: Use repetition penalty (--repeat-penalty on llama.cpp) of 1.05 - 1.15 for best results </b>
|
| 30 |
|
| 31 |
+
Quantized from fp16 with love.
|
| 32 |
* Weighted quantizations were creating using fp16 GGUF and [groups_merged-enhancedV2-TurboMini.txt](https://github.com/ggerganov/llama.cpp/discussions/5263#discussioncomment-9432658) in 189 chunks and n_ctx=512
|
| 33 |
* This method of calculating the importance matrix showed improvements in some areas for Mistral 7b and Llama3 8b models, see above post for details
|
| 34 |
* The enhancedv2-turbomini file appends snippets from turboderp's calibration data to the standard groups_merged.txt file
|