maveryn/sphinx-qwen3-8b
Partial fine-tune vs Qwen/Qwen3-VL-8B-Instruct. Partial change: 86% of the weights changed (attention, embeddings, mlp, norms), in 36 of 36 layers.
Tensors changed: Embeddings 3/4, Norms 167/261, MLP 213/232, Attention 252/252, lm_head 0/1
2 derivatives checked
- mradermacher/sphinx-qwen3-8b-i1-GGUF: Quantization (fewer layers)
- mradermacher/sphinx-qwen3-8b-GGUF: Same weights, converted (fewer layers)