Update README.md

README.md

@@ -8,7 +8,9 @@ pipeline_tag: text-generation
 
 # Quantization
 
-3.75bpw quant, using default exllama parameters.
+3.75bpw quantization using default settings, for good amount of context on a 24GB GPU.
+
+This is the base model, not instruct! Base models tend to be better for raw completion, especially at long context.
 
 # Qwen2.5-32B