Update README.md

README.md

@@ -16,7 +16,7 @@ datasets:
 
 # StableLM-Tuned-Alpha 3B 8Bit
 
-3B model converted to int8 by rockerBOO. May require `bitsandbytes` dependency and using `load_in_8bit=True`.
+3B model converted to int8 by rockerBOO. May require `bitsandbytes` dependency. Tested on a 2080 8GB.
 
 ## Model Description
 
@@ -29,9 +29,9 @@ Get started chatting with `StableLM-Tuned-Alpha` by using the following code sni
 ```python
 from transformers import AutoModelForCausalLM, AutoTokenizer, StoppingCriteria, StoppingCriteriaList
 
-tokenizer = AutoTokenizer.from_pretrained("StabilityAI/stablelm-tuned-alpha-7b")
-model = AutoModelForCausalLM.from_pretrained("StabilityAI/stablelm-tuned-alpha-7b")
-model.half().cuda()
+tokenizer = AutoTokenizer.from_pretrained("StabilityAI/stablelm-tuned-alpha-3b")
+model = AutoModelForCausalLM.from_pretrained("rockerBOO/stablelm-tuned-alpha-3b-8bit")
+model.cuda()
 
 class StopOnTokens(StoppingCriteria):
     def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool: