Update README.md

README.md

@@ -76,7 +76,7 @@ Here's the thing with the 2 extra layers compared to my first model. When I trai
 
 I have any model generate a bunch of prompts that a teacher model answers with gusto (the chosen column), and then have NinjaMouse2 also answer them (as the rejects). **BAM**. Skibidibi doo. Have I made these DPO datasets? No. But the prompts, their evaluations, along with responses of its own, responses from better models, and evaluations of both of them are included in the training. You can find the dataset [here](https://huggingface.co/datasets/trollek/Self-Rewarding-Mouse).
 
-### Ollama
+## Ollama
 
 I have [quantised this model](https://huggingface.co/trollek/NinjaMouse2-2.5B-v0.1-GGUF) and made it available through [LM Studio](https://lmstudio.ai/) and [Ollama](https://ollama.com/) in Q4KM and Q6K.
 
@@ -88,6 +88,13 @@ ollama run trollek/ninjamouse2:34l-q4_K_M
 ollama run trollek/ninjamouse2:34l-q6_K
 ```
 
+## Quantizations
+
+[@cgus](https://huggingface.co/cgus) has done a great job with the quants. Reducing models from 16bit to ~2, and every bit inbetween, is Numberwang and much appreciated.
+
+- **GGUF iMatrix:** [cgus](https://huggingface.co/cgus)/[NinjaMouse2-2.5B-v0.1-iMat-GGUF](https://huggingface.co/cgus/NinjaMouse2-2.5B-v0.1-iMat-GGUF)
+- **Exllamav2:** [cgus](https://huggingface.co/cgus)/[NinjaMouse2-2.5B-v0.1-exl2](https://huggingface.co/cgus/NinjaMouse2-2.5B-v0.1-exl2)
+- **GGUF:** trollek/[NinjaMouse2-2.5B-v0.1-GGUF](https://huggingface.co/trollek/NinjaMouse2-2.5B-v0.1-GGUF)
 
 ## Notes