Update README.md

README.md

@@ -6,4 +6,11 @@ pipeline_tag: text-generation
 tags:
 - text-generation-inference
 ---
-This is [Llama2-22b](https://huggingface.co/chargoddard/llama2-22b) in a couple of GGML formats. I have no idea what I'm doing so if something doesn't work that's likely on me, not the models themselves.
+This is [Llama2-22b](https://huggingface.co/chargoddard/llama2-22b) in a couple of GGML ormats. I have no idea what I'm doing so if something doesn't work that's likely on me, not the models themselves.
+
+Approximate VRAM requirements
+MODEL  | SIZE   | VRAM
+q5_1   | 16.0GB | 21.5GB
+q4_K_M | 12.8GB | 18.3GB
+q3_K_M | 10.0GB | 16.1GB
+q2_K   | 9.0GB  | 14.5GB