Mungert/InternVL3-8B-GGUF

Full fine-tune (fewer layers) vs OpenGVLab/InternVL3-8B-Instruct. Mungert/InternVL3-8B-GGUF matches part of OpenGVLab/InternVL3-8B-Instruct (339 of its 685 tensors; the rest of OpenGVLab/InternVL3-8B-Instruct has no counterpart). On the shared part: Mungert/InternVL3-8B-GGUF is a GGUF (Q8_0) of different weights. Fine-tune: 100% of the weights changed, across 28 of 28 layers.

Tensors changed: Embeddings 1/1, Norms 49/57, MLP 84/84, Attention 195/196, lm_head 1/1

On the Hub · Downloads on Model Pulse