OpenGVLab/InternVL3-8B-Instruct
Full fine-tune vs OpenGVLab/InternVL3-8B-Pretrained. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 5/5, Norms 145/153, MLP 186/186, Attention 292/292, other 48/48, lm_head 1/1
8 derivatives checked
- OpenGVLab/InternVL3-8B: Full fine-tune
- OpenGVLab/InternVL3-8B-hf: Different architecture
- ggml-org/InternVL3-8B-Instruct-GGUF: Same weights, converted (fewer layers)
- unsloth/InternVL3-8B-Instruct-GGUF: Same weights, converted (fewer layers)
- Mungert/InternVL3-8B-GGUF: Full fine-tune (fewer layers)
- mradermacher/InternVL3-8B-Instruct-i1-GGUF: Quantization (fewer layers)
- mradermacher/InternVL3-8B-Instruct-GGUF: Same weights, converted (fewer layers)
- SandLogicTechnologies/Internvl3-8b-instruct-GGUF: Quantization (fewer layers)
1 other repos hold byte-for-byte identical weight files.