search

TransformerLensOrg / TransformerLens

A library for mechanistic interpretability of GPT-style language models
https://transformerlensorg.github.io/TransformerLens/
MIT License
1.48k stars 288 forks source link

ValueError: microsoft/Phi-3-mini-128k-instruct not found. #670

Open joykirat18 opened 3 months ago

joykirat18 commented 3 months ago

Not able to load microsoft/Phi-3-mini-128k-instruct.

Code snippet



def loadTransformerLensModel(modelPath):
    tokenizer = AutoTokenizer.from_pretrained(modelPath)
    hf_model = AutoModelForCausalLM.from_pretrained( 
    "microsoft/Phi-3-mini-128k-instruct",  
    device_map="cuda",  
    torch_dtype="auto",  
    trust_remote_code=True,  
) 
    model = HookedTransformer.from_pretrained(modelPath, hf_model=hf_model, device='cpu')

    return model, tokenizer

model, tokenizer = loadTransformerLensModel(model_name)```
joykirat18 commented 3 months ago

What changes should be made for running Phi-3-mini-128k-instruct