# Copyright 2026 Gentoo Authors # Distributed under the terms of the GNU General Public License v2 # LocalAI generic HuggingFace backend: text generation, embeddings and # assorted pipelines through the transformers library, plus # SentenceTransformer embedding models. The venv skeleton lives in # local-ai-python.eclass; the wheels tarball carries only # sentence-transformers and diffusers (noise schedulers for # VibeVoice-style configs) — everything else is tree packages via # system-site, including sci-ml/bitsandbytes for 4/8-bit quantized # loading. diffusers is pinned to 0.40.0: 0.41 raised its # huggingface-hub floor to 1.32, past the tree's 1.26; 0.40 accepts # >=1.23. # Deliberately NOT shipped: llvmlite/numba (upstream's requirements # list them for audio feature extractors nothing in backend.py or the # shipped libraries imports) and the Intel/OpenVINO stack (lazily # imported for request types this packaging never serves). EAPI=8 LOCAL_AI_PYTHON_SMOKE_IMPORTS="backend transformers sentence_transformers diffusers bitsandbytes scipy" inherit local-ai-python DESCRIPTION="LocalAI generic HuggingFace transformers backend (gRPC server)" LICENSE="MIT Apache-2.0" RDEPEND+=" sci-ml/transformers[${PYTHON_SINGLE_USEDEP}] sci-ml/accelerate[${PYTHON_SINGLE_USEDEP}] sci-ml/bitsandbytes[${PYTHON_SINGLE_USEDEP}] $(python_gen_cond_dep ' dev-python/scipy[${PYTHON_USEDEP}] dev-python/scikit-learn[${PYTHON_USEDEP}] dev-python/pillow[${PYTHON_USEDEP}] dev-python/soundfile[${PYTHON_USEDEP}] ') " DEPEND="${RDEPEND}"