posicube/KolmOCR_v251129

Full fine-tune vs Qwen/Qwen2.5-VL-7B-Instruct. Fine-tune: 100% of the weights changed, across 32 of 32 layers.

Tensors changed: Norms 110/122, MLP 280/280, Attention 324/324, Embeddings 1/1

2 derivatives checked

• mradermacher/KolmOCR_v251129-i1-GGUF: Quantization (fewer layers)
• mradermacher/KolmOCR_v251129-GGUF: Same weights, converted (fewer layers)

On the Hub · Downloads on Model Pulse