Kewk/Heretical-Qwen3.5-2B

Partial fine-tune vs Qwen/Qwen3.5-2B-Base. Partial change: 86% of the weights changed (attention, embeddings, mlp, norms), in 24 of 24 layers.

Tensors changed: Attention 142/264, Norms 81/177, Embeddings 2/4, MLP 76/172

2 derivatives checked

• mradermacher/Heretical-Qwen3.5-2B-GGUF: Same weights, converted
• mradermacher/Heretical-Qwen3.5-2B-i1-GGUF: Quantization

On the Hub · Downloads on Model Pulse