fwgpiyawudk/chandra-ocr-2-bf16-mlx

Partial fine-tune vs datalab-to/chandra-ocr-2. Partial change: 0% of the weights changed (norms), in 32 of 32 layers.

Tensors changed: Embeddings 0/1, Norms 81/105, Attention 0/200, MLP 0/96

2 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse