TheBloke

/

Yi-34B-GGUF

like 73

Thanks

Hmm I’m not sure, isn’t this a base model with no prompt template?

That's what I originally thought. But @prasanna2003 is correct that in the Github it shows it being queried with Human: {prompt} Assistant:
https://github.com/01-ai/Yi/blob/main/finetune/sft/prompt_eval.py#L163-L169

Similarly we can see it used in this example training data:
https://github.com/01-ai/Yi/blob/main/finetune/yi_example_dataset/data/train.jsonl

Although It's a bit unclear in their text generation demo, it does not: https://github.com/01-ai/Yi/blob/main/demo/text_generation.py
and the dataset is an 'example dataset' so it's also possible that this is showing an example of how you could fine tune the base model?

I'd welcome any further input or testing on this.