OpenGVLab/InternVL3-78B

Full fine-tune vs OpenGVLab/InternVL3-78B-Instruct. Fine-tune: 100% of the weights changed, across 80 of 80 layers.

Tensors changed: Embeddings 4/5, Norms 225/341, Attention 693/695, other 88/90, MLP 417/426, lm_head 1/1

3 derivatives checked

• unsloth/InternVL3-78B-GGUF: Same weights, converted (fewer layers)
• mradermacher/InternVL3-78B-i1-GGUF: Quantization (fewer layers)
• mradermacher/InternVL3-78B-GGUF: Quantization (fewer layers)

1 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse