How about a scaled down version like 7B/8B?

#2
by ZeroWw - opened

How about a scaled down version like 7B/8B?

7/8B versions are not good for complex medical reasoning, that's my experience with using Llama 3.1 8B. So here might be similar....

7/8B versions are not good for complex medical reasoning, that's my experience with using Llama 3.1 8B. So here might be similar....

Mistral 7B v03 was not bad in medical reasoning (as it was) with finetuning it could improve a lot.

7/8B versions are not good for complex medical reasoning, that's my experience with using Llama 3.1 8B. So here might be similar....

Mistral 7B v03 was not bad in medical reasoning (as it was) with finetuning it could improve a lot.
I'm planning to finetune Mistral 7B v0.3 4bit quantized for nutrition, diet and exercise recommendation. Can you recommend any datasets I can use for this task?

Looking forward to the distilled version. If the version can be distilled and deployed locally in medical colleges, it should benefit many people.

Your need to confirm your account before you can post a new comment.

Sign up or log in to comment