balnazzar/Smolly-248M

Full fine-tune vs Felladrin/TinyMistral-248M-Chat-v4. Fine-tune: 100% of the weights changed, across 12 of 12 layers.

Tensors changed: Norms 25/25, MLP 36/36, Attention 48/48

1 derivatives checked

• mradermacher/Smolly-248M-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse