X-Reasoner-7B-f32-GGUF

Microsoft X-Reasoner-7B is a 7B-parameter vision-language model from Microsoft Research, post-trained solely on general-domain text using a two-stage approach to achieve generalizable reasoning across modalities and domains without vision-specific fine-tuning. Designed for robust performance on multimodal benchmarks like mathematical reasoning, visual question answering, and cross-domain tasks, it leverages instruction tuning and reinforcement learning to produce structured reasoning traces that transfer zero-shot to unseen visual and textual challenges. The model demonstrates strong generalization by outperforming vision-augmented baselines on held-out domains through text-only post-training, making it particularly valuable for research into modality-agnostic reasoning capabilities under an Apache 2.0 license with Transformers deployment support.

LightOnOCR-2-1B [GGUF]

File Name	Quant Type	File Size	File Link
LightOnOCR-2-1B-BF16.gguf	BF16	1.2 GB	Download
LightOnOCR-2-1B-F16.gguf	F16	1.2 GB	Download
LightOnOCR-2-1B-F32.gguf	F32	2.39 GB	Download
LightOnOCR-2-1B-Q8_0.gguf	Q8_0	639 MB	Download
LightOnOCR-2-1B.mmproj-BF16.gguf	mmproj-BF16	829 MB	Download
LightOnOCR-2-1B.mmproj-F16.gguf	mmproj-F16	819 MB	Download
LightOnOCR-2-1B.mmproj-F32.gguf	mmproj-F32	1.64 GB	Download

Quants Usage

(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)

Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better):