stockmark/Stockmark-DocReasoner-Qwen2.5-VL-32B

Full fine-tune vs Qwen/Qwen2.5-VL-32B-Instruct. Fine-tune: 98% of the weights changed, across 64 of 64 layers.

Tensors changed: Attention 576/576, MLP 388/388, Norms 148/194, Embeddings 1/2, lm_head 1/1

2 derivatives checked

• stockmark/Stockmark-DocReasoner-Qwen2.5-VL-32B-FP8-dynamic: Quantization
• tokimoa/Stockmark-DocReasoner-Qwen2.5-VL-32B-MLX-4bit: Quantization

On the Hub · Downloads on Model Pulse