wjbmattingly/lfm2-vl-medieval-page

Full fine-tune vs LiquidAI/LFM2-VL-450M. Fine-tune: 100% of the weights changed, across 16 of 16 layers.

Tensors changed: Embeddings 4/5, other 34/34, MLP 89/96, Norms 42/96, Attention 120/120

1 derivatives checked

• mradermacher/lfm2-vl-medieval-page-GGUF: Same weights, converted (fewer layers)

On the Hub · Downloads on Model Pulse