Update README.md

README.md

@@ -11,8 +11,8 @@ tags:
 # merged
 
 This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit). It is a prune of Meta-Llama-3-8B-Instruct down to 20 layers, or about 5.4B models.
-This should allow Q8 or Q6 inference on 6GB VRAM, Q5 inference on 4GB VRAM, and full-weight fine-tuning (fp16) with a context length of XX and batch size YY on 24GB VRAM.
-It is in dire need of healing through tuning (the purpose of this experiment - can I prune&tune an *instruct* model?).
+Mostly, this is a test of pruning & healing an instruct-tuned model.
+This size should allow Q8 or Q6 inference on 6GB VRAM, Q5 inference on 4GB VRAM, and full-weight fine-tuning ... well, with less VRAM than an 8B model.
 
 ## Merge Details
 ### Merge Method