LocalLLaMA

1 readers

1 users here now

Community to discuss about Llama, the family of large language models created by Meta AI.

founded 10 months ago

MODERATORS

communick@poweruser.forum

"Base" models were actually trained with some GPT instruct datasets (alien.top)

submitted 10 months ago by Wonderful_Ad_5134@alien.top to c/localllama@poweruser.forum

9 comments fedilink hide all child comments

Look at this, apart Llama1, all the other "base" models will likely answer "language" after "As an AI". That means Meta, Mistral AI and 01-ai (the company that made Yi) likely trained the "base" models with GPT instruct datasets to inflate the benchmark scores and make it look like the "base" models had a lot of potential, we got duped hard on that one.

https://preview.redd.it/vqtjkw1vdyzb1.png?width=653&format=png&auto=webp&s=91652053bcbc8a7b50bced9bbf8638fa417387bb

you are viewing a single comment's thread
view the rest of the comments

[–] FPham@alien.top 1 points 10 months ago

Shouldn't be the proof in the pudding?

If Mistral 7B is better than most other 7b models, then they did something right, no?

I understand that the base model then can inherit some biases - but it's onto them that they didn't cleaned those "As and AI..." answers strings from their dataset. So despite this, it performs better.