OpenGVLab/InternVL3_5-4B
Partial fine-tune vs OpenGVLab/InternVL3_5-4B-MPO. Partial change: 85% of the weights changed (attention, lm_head, mlp, norms), in 36 of 36 layers.
Tensors changed: Embeddings 0/5, Norms 102/241, MLP 113/210, Attention 144/240, other 0/48, lm_head 1/1
7 derivatives checked
- bartowski/OpenGVLab_InternVL3_5-4B-GGUF: Same weights, converted (fewer layers)
- mradermacher/InternVL3_5-4B-i1-GGUF: Quantization (fewer layers)
- mradermacher/InternVL3_5-4B-GGUF: Same weights, converted (fewer layers)
- lmstudio-community/InternVL3_5-4B-GGUF: Quantization (fewer layers)
- issai/Qolda: Full fine-tune
- RKNNAI/RK1828-VLM-InternVL3_5-4B: Quantization
- OpenGVLab/InternVL3_5-4B-Flash: Full fine-tune
1 other repos hold byte-for-byte identical weight files.