AIR-hl/Qwen2.5-1.5B-ultrachat200k

Full fine-tune vs Qwen/Qwen2.5-1.5B. Fine-tune: 100% of the weights changed, across 28 of 28 layers.

Tensors changed: Embeddings 1/1, Norms 42/57, MLP 84/84, Attention 195/196

1 derivatives checked

• mradermacher/Qwen2.5-1.5B-ultrachat200k-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse