]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
mtmd: Add DeepSeekOCR 2 Support (#20975)
authorSaba Fallah <redacted>
Fri, 29 May 2026 14:13:51 +0000 (16:13 +0200)
committerGitHub <redacted>
Fri, 29 May 2026 14:13:51 +0000 (16:13 +0200)
commitda3f990a47ec8c25ff3d2154d3dea46ee3f4f334
tree093d63067c488ed2668dd6cd10fd2bbec30137ca
parent6ed481eea4cf4ed40777db2fa29e8d08eb712b3b
mtmd: Add DeepSeekOCR 2 Support (#20975)

* mtmd: DeepSeek-OCR 2 support, with multi-tile dynamic resolution

* introduced clip_image_f32::add_viewsep

* address PR review

- drop redundant ggml_cpy ops in both deepseekocr versions build
- drop no-op ggml_cont in build_sam
- assert num_image_tokens deepseekocr2
- view_seperator as (1, n_embd) at conversion (for both versions)
- drop redundant ggml_reshape_2d

* Update tools/mtmd/models/deepseekocr2.cpp

Co-authored-by: Xuan-Son Nguyen <redacted>
---------

Co-authored-by: Xuan-Son Nguyen <redacted>
16 files changed:
conversion/__init__.py
conversion/base.py
conversion/deepseek.py
gguf-py/gguf/constants.py
gguf-py/gguf/tensor_mapping.py
tools/mtmd/CMakeLists.txt
tools/mtmd/clip-impl.h
tools/mtmd/clip-model.h
tools/mtmd/clip.cpp
tools/mtmd/models/deepseekocr.cpp
tools/mtmd/models/deepseekocr2.cpp [new file with mode: 0644]
tools/mtmd/models/models.h
tools/mtmd/mtmd-image.cpp
tools/mtmd/mtmd-image.h
tools/mtmd/mtmd.cpp
tools/mtmd/tests/test-deepseek-ocr.py