Commit Graph

4 Commits

Author SHA1 Message Date
Jeffrey Morgan
66831dcf70 x/imagegen: fix image editing support (#13866)
- Fix panic in ollama show for image gen models (safe type assertion)
- Add vision capability for Flux2KleinPipeline models at create time
- Flatten transparent PNG images onto white background for better results
2026-01-23 15:37:17 -08:00
Patrick Devine
148a1be0a3 Clean up the manifest and modelpath (#13807) 2026-01-21 11:46:17 -08:00
Jeffrey Morgan
03bf241c33 x/imagegen: add FP4 quantization support for image generation models (#13773)
Add --quantize fp4 support to ollama create for image generation models
(flux2, z-image-turbo), using MLX's affine 4-bit quantization.

Changes:
- Add fp4 to validation in CreateImageGenModel
- Add FP4 case to quantizeTensor (group_size=32, bits=4, affine mode)
- Add GetQuantization() to WeightSource interface for dynamic params
- Update LoadLinearLayer to use quantization params from model metadata
2026-01-19 00:54:54 -08:00
Patrick Devine
a077d996e3 Fix create and show commands for experimental models (#13741)
* x: make `ollama create --experimental` import from safetensors

This change allows pulling in safetensors models into the new experimental model format, and also
fixes the `ollama show` command to be able to correctly display the model information.

* gofumpt the linter

* gofumpt the linter again

* validate the model name
2026-01-16 14:31:55 -08:00