What was the base model?
#7
by
pirolen
- opened
Hi, could you clarify the discrepancy: in your paper you write "we discuss the Vikhr model based on Mistral 7B". The current config says it is LLama. https://huggingface.co/Vikhrmodels/Vikhr-7B-instruct_0.2/blob/main/config.json Thanks!
Mistral and llama has same architecture except SWA, we didn't use it, so it's became llama))