OpenGVLab/InternVL3-38B-Instruct
Full fine-tune vs OpenGVLab/InternVL3-38B-Pretrained. Fine-tune: 100% of the weights changed, across 64 of 64 layers.
Tensors changed: Embeddings 5/5, Norms 309/309, Attention 583/583, other 90/90, MLP 378/378, lm_head 1/1
5 derivatives checked
- OpenGVLab/InternVL3-38B: Full fine-tune
- OpenGVLab/InternVL3-38B-hf: Full fine-tune
- mradermacher/InternVL3-38B-Instruct-i1-GGUF: Quantization (fewer layers)
- unsloth/InternVL3-38B-Instruct-GGUF: Same weights, converted (fewer layers)
- mradermacher/InternVL3-38B-Instruct-GGUF: Quantization (fewer layers)
1 other repos hold byte-for-byte identical weight files.