HuggingFaceH4/starchat2-15b-sft-v0.1

Full fine-tune vs bigcode/starcoder2-15b. Fine-tune: 100% of the weights changed, across 40 of 40 layers.

Tensors changed: Norms 162/162, MLP 160/160, Attention 320/320

3 derivatives checked

• bartowski/starchat2-15b-v0.1-GGUF: Full fine-tune
• mradermacher/starchat2-15b-sft-v0.1-GGUF: Quantization
• HuggingFaceH4/starchat2-15b-v0.1: Full fine-tune

On the Hub · Downloads on Model Pulse