]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
model : support vision LiquidAI LFM2-VL family (#15347)
authorTarek Dakhran <redacted>
Sat, 16 Aug 2025 21:33:54 +0000 (23:33 +0200)
committerGitHub <redacted>
Sat, 16 Aug 2025 21:33:54 +0000 (23:33 +0200)
commit65349f26f2299e06477ec8e85e46243046801358
treef4f63e132a34a02d91e9d7d27174f4ec552b953a
parent1fe00296f587dfca0957e006d146f5875b61e43d
model : support vision LiquidAI LFM2-VL family (#15347)

* wip lfm2 vision model

* Fix conv weight

* Implement dynamic resolution

* Fix cuda

* support LFM2-VL-450M

* happy CI

* Remove extra `ggml_conv` and put others into the right place

Co-authored-by: Sigbjørn Skjæret <redacted>
---------

Co-authored-by: Xuan Son Nguyen <redacted>
Co-authored-by: Sigbjørn Skjæret <redacted>
convert_hf_to_gguf.py
gguf-py/gguf/constants.py
gguf-py/gguf/tensor_mapping.py
tools/mtmd/clip-impl.h
tools/mtmd/clip.cpp