README.md

metadata

base_model:
  - qingy2024/Qwark-4B
datasets:
  - qingy2024/qwark-corpus

Qwarkstar 4B

Training complete!

This is a fine-tuned frankenmerge of Qwen2.5 3B on 1.3B+ tokens from qingy2024/qwark-corpus to smooth out any rough patches.

It is currently a base model (not instruct-tuned yet).