Hugging Face's logo

Undi95
/
MG-FinalMix-72B

Text Generation
Transformers
Safetensors
English
qwen2
mergekit
Merge
OG_finetune_merge
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
5

Very Very Good

#5
by NotaNunya - opened
Discussion
NotaNunya

I don't know why this seems to be overlooked but I can't sit back and just lurk with no praise for this model, I have been running with 3xxs and it's really smart, smarter than all the models I've tested so far.

Undi95
 Owner

Thank you!

Ainonake

@NotaNunya have you tested it against original magnum 72b v1/4? Or newest magnum SE? Those are very smart models by themselves, especially newest SE.

NotaNunya

@Ainonake yes, all of them actually, v1 v2 v3 v4, even further up the chain in source models, whatever I could find because usually the further from an original model you get the less intelligent it becomes but no, somehow this pushes just a little further imo.

Sign up or log in to comment