Update README.md

README.md

@@ -1,3 +1,28 @@
 ---
 license: llama2
 ---
+
+These are exl2 quants of [Goliath-longLORA-120b-rope8-32k-fp16](https://huggingface.co/grimulkan/Goliath-longLORA-120b-rope8-32k-fp16).  
+Main branch includes `measurement.json` files for the new default dataset and the one used by [goliath-120b-exl2-rpcal](https://huggingface.co/Panchovix/goliath-120b-exl2-rpcal).
+
+# Available versions
+[2.65bpw](https://huggingface.co/aikitoria/Goliath-longLORA-120b-rope8-32k-exl2/tree/2.65bpw) using default dataset  
+[4.35bpw](https://huggingface.co/aikitoria/Goliath-longLORA-120b-rope8-32k-exl2/tree/4.35bpw) using default dataset  
+[4.35bpw-rpcal](https://huggingface.co/aikitoria/Goliath-longLORA-120b-rope8-32k-exl2/tree/4.35bpw-rpcal) using PIPPA dataset
+
+
+# Memory usage tests
+
+### 2.65bpw
+context 16k, cache 16: 46.9GiB (fits in 2x 3090)  
+context 32k, cache 8: 47GiB (fits in 2x 3090)  
+
+### 4.35bpw
+context 16k, cache 16: 70.1GiB (fits in 3x 3090)  
+context 32k, cache 8: 70.3GiB (fits in 3x 3090)  
+context 32k, cache 16: 78.7GiB (fits in A100 80GB)  
+
+# Super epic scientific test results
+The 2.65bpw version suffered greatly, it's not completely broken, but it's not good either.  
+The 4.35bpw version is worse than normal goliath but better than goliath with rope scale applied for long context.  
+The version using the PIPPA dataset produces worse results than the one using the default dataset on any context length.