Fidelity to the BF16 original on an internal mixed EN/KO set (100 prompts, text next-token scoring).
For task-specific OCR prompts (tables, formulas, layout) see the original model card.
This is an OCR model: send an image. Text-only chat is not its purpose — in our vLLM check of this build both Korean text prompts repeated one phrase until the token limit, and "What is 17 * 23?" was answered "17 * 23 = 27".
The KV cache is kept in BF16 by default. KV-cache precision is a runtime choice, independent of the
weight quantization above; for long contexts or large batches FP8 roughly halves KV-cache memory: