aisingapore/Qwen-SEA-LION-v4-4B-VL
Full fine-tune vs Qwen/Qwen3-VL-4B-Instruct. Fine-tune: 91% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 1/4, Norms 136/249, MLP 108/220, Attention 144/240
4 derivatives checked
- aisingapore/Qwen-SEA-Guard-4B-2602: Full fine-tune
- mradermacher/Qwen-SEA-LION-v4-4B-VL-i1-GGUF: Quantization (fewer layers)
- mradermacher/Qwen-SEA-LION-v4-4B-VL-GGUF: Same weights, converted (fewer layers)
- MagicalAlchemist/Qwen-SEA-LION-v4-4B-VL-Magic_decensored: Abliterated