Text Generation
Transformers
Safetensors
English
llama
conversational
Inference Endpoints
text-generation-inference

Do Korean dataset used when Finetune?

#15
by maywell - opened

Hi, I just wondering if Korean dataset used.
I tried this and base model but the Korean output was awful.
but in upstage ko llm leaderboard SOLAR finetuned model is #1

  • I'm training SOLAR model with my Korean dataset now. and It's fits well crazy.
upstage org

Korean output was awful.
We did not use Korean datasets for fine-tuning.

hunkim changed discussion status to closed

Sign up or log in to comment