janhq/Jan-v2-VL-max-gguf

Full fine-tune vs Qwen/Qwen3-VL-30B-A3B-Thinking. janhq/Jan-v2-VL-max-gguf is a GGUF (Q8_0) of different weights. Fine-tune: 100% of the weights changed, across 48 of 48 layers.

Tensors changed: Embeddings 1/1, Norms 0/193, MLP 48/48, Attention 192/192, lm_head 1/1

On the Hub · Downloads on Model Pulse