Update README.md

README.md

@@ -7,9 +7,25 @@ tags:
 - language
 - granite-3.1
 base_model:
-- ibm-granite/granite-3.1-8b-base
+- ibm-granite/granite-3.1-8b-instruct
 ---
-
+# Granite-3.1-8B-Instruct-exl2
+Original model: [granite-3.1-8b-instruct](https://huggingface.co/ibm-granite/granite-3.1-8b-instruct)  
+Creator: [Granite Team, IBM](https://huggingface.co/ibm-granite)
+## Quants
+[4bpw h6 (main)](https://huggingface.co/cgus/granite-3.1-8b-instruct-exl2/tree/main)  
+[4.5bpw h6](https://huggingface.co/cgus/granite-3.1-8b-instruct-exl2/tree/4.5bpw-h6)  
+[5bpw h6](https://huggingface.co/cgus/granite-3.1-8b-instruct-exl2/tree/5bpw-h6)  
+[6bpw h6](https://huggingface.co/cgus/granite-3.1-8b-instruct-exl2/tree/6bpw-h6)  
+[8bpw h8](https://huggingface.co/cgus/granite-3.1-8b-instruct-exl2/tree/8bpw-h8)  
+
+## Quantization notes
+Made with Exllamav2 0.2.7 with the default dataset. This is the first version that supports Granite architecture.  
+So this model will work only with apps that have exllamav2 0.2.7 or newer.  
+It should work with newest TabbyAPI or Text-Generation-WebUI.  
+Exl2 models require to be fully loaded into GPU, offloading to RAM isn't natively supported.  
+Also they require either a Nvidia RTX card on Windows or Nvidia RTX/AMD ROCm on Linux. 
+# Original model card
 # Granite-3.1-8B-Instruct
 
 **Model Summary:**