mradermacher

/

model_requests

like 31

Requesting quants for my weird model_stock merge of all the LLaMa-3-8B fine-tunes I could find which seemed to have at least some effort put into the fine-tuning process, without adding extra tokens. 54 models to be exact, which is ~864gb worth of models lol.

The idea of the merge was to get a sort of average of all the models for use as a ClownMoE base or a fine-tuning base, but it's probably an interesting model by itself as well.

https://huggingface.co/PJMixers/LLaMa-3-CursedStock-v1.6-8B

Thanks for the nearly 2400 model pages so far. Putting in work.