langchain-aneforge 0.1.0__tar.gz

langchain_aneforge-0.1.0/.github/workflows/publish.yml

@@ -0,0 +1,22 @@
+name: Publish to PyPI
+
+# Publishes on a GitHub Release via PyPI Trusted Publishing (OIDC): no API token.
+# One-time setup on PyPI -> the project's Publishing settings -> add a trusted
+# publisher with owner sbryngelson, repo langchain-aneforge, workflow publish.yml,
+# environment pypi.
+
+on:
+  release:
+    types: [published]
+
+jobs:
+  publish:
+    runs-on: ubuntu-latest
+    environment: pypi
+    permissions:
+      id-token: write          # mint the OIDC token Trusted Publishing checks
+    steps:
+      - uses: actions/checkout@v4
+      - uses: astral-sh/setup-uv@v5
+      - run: uv build
+      - uses: pypa/gh-action-pypi-publish@release/v1

langchain_aneforge-0.1.0/.gitignore

@@ -0,0 +1,6 @@
+__pycache__/
+*.pyc
+dist/
+build/
+*.egg-info/
+.pytest_cache/

langchain_aneforge-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Spencer Bryngelson
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

langchain_aneforge-0.1.0/PKG-INFO

@@ -0,0 +1,63 @@
+Metadata-Version: 2.4
+Name: langchain-aneforge
+Version: 0.1.0
+Summary: LangChain embeddings that run on the Apple Neural Engine, via ANEForge (no CoreML).
+Project-URL: Homepage, https://github.com/sbryngelson/langchain-aneforge
+Project-URL: Repository, https://github.com/sbryngelson/langchain-aneforge
+Project-URL: ANEForge, https://github.com/sbryngelson/ANEForge
+Author-email: Spencer Bryngelson <sbryngelson@gmail.com>
+License-Expression: MIT
+License-File: LICENSE
+Keywords: ane,aneforge,apple-neural-engine,apple-silicon,embeddings,langchain
+Classifier: Development Status :: 3 - Alpha
+Classifier: Intended Audience :: Developers
+Classifier: Operating System :: MacOS :: MacOS X
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.10
+Requires-Dist: aneforge>=0.1.3
+Requires-Dist: langchain-core>=0.3
+Description-Content-Type: text/markdown
+
+# langchain-aneforge
+
+LangChain embeddings that run on the Apple Neural Engine, via
+[ANEForge](https://github.com/sbryngelson/ANEForge). The encoder runs on the ANE
+with no CoreML, matching the fp32 reference to cosine ~1.0 at a fraction of the
+GPU's energy.
+
+```sh
+pip install langchain-aneforge
+```
+
+Apple Silicon macOS only (ANEForge reaches the engine through a built dispatch
+shim; see the ANEForge install notes).
+
+## Use
+
+```python
+from langchain_aneforge import ANEEmbeddings
+
+emb = ANEEmbeddings(model="sentence-transformers/all-MiniLM-L6-v2")
+vec = emb.embed_query("the Apple Neural Engine runs networks at low power")
+docs = emb.embed_documents(["one passage", "another passage"])
+```
+
+`ANEEmbeddings` is a standard `langchain_core.embeddings.Embeddings`, so it drops
+into any LangChain retriever or vector store:
+
+```python
+from langchain_community.vectorstores import FAISS
+
+store = FAISS.from_texts(["...", "..."], ANEEmbeddings())
+hits = store.similarity_search("query", k=3)
+```
+
+`model` is any BERT-family sentence encoder on the Hugging Face hub. Its pooling
+mode and normalize flag are read from the model's own sentence-transformers config,
+so mean-pooled (MiniLM, E5) and cls-pooled (BGE, GTE) models both produce correct
+vectors. Pass `int8=True` to stream int8 weights (half the size, cosine ~0.9999).
+
+## License
+
+[MIT](LICENSE). Not affiliated with or endorsed by Apple or LangChain.

langchain_aneforge-0.1.0/README.md

@@ -0,0 +1,42 @@
+# langchain-aneforge
+
+LangChain embeddings that run on the Apple Neural Engine, via
+[ANEForge](https://github.com/sbryngelson/ANEForge). The encoder runs on the ANE
+with no CoreML, matching the fp32 reference to cosine ~1.0 at a fraction of the
+GPU's energy.
+
+```sh
+pip install langchain-aneforge
+```
+
+Apple Silicon macOS only (ANEForge reaches the engine through a built dispatch
+shim; see the ANEForge install notes).
+
+## Use
+
+```python
+from langchain_aneforge import ANEEmbeddings
+
+emb = ANEEmbeddings(model="sentence-transformers/all-MiniLM-L6-v2")
+vec = emb.embed_query("the Apple Neural Engine runs networks at low power")
+docs = emb.embed_documents(["one passage", "another passage"])
+```
+
+`ANEEmbeddings` is a standard `langchain_core.embeddings.Embeddings`, so it drops
+into any LangChain retriever or vector store:
+
+```python
+from langchain_community.vectorstores import FAISS
+
+store = FAISS.from_texts(["...", "..."], ANEEmbeddings())
+hits = store.similarity_search("query", k=3)
+```
+
+`model` is any BERT-family sentence encoder on the Hugging Face hub. Its pooling
+mode and normalize flag are read from the model's own sentence-transformers config,
+so mean-pooled (MiniLM, E5) and cls-pooled (BGE, GTE) models both produce correct
+vectors. Pass `int8=True` to stream int8 weights (half the size, cosine ~0.9999).
+
+## License
+
+[MIT](LICENSE). Not affiliated with or endorsed by Apple or LangChain.

langchain_aneforge-0.1.0/langchain_aneforge/__init__.py

@@ -0,0 +1,4 @@
+"""LangChain integration for ANEForge: embeddings on the Apple Neural Engine."""
+from langchain_aneforge.embeddings import ANEEmbeddings
+
+__all__ = ["ANEEmbeddings"]

langchain_aneforge-0.1.0/langchain_aneforge/embeddings.py

@@ -0,0 +1,36 @@
+"""LangChain embeddings backed by the Apple Neural Engine, via ANEForge."""
+from __future__ import annotations
+
+from typing import List
+
+from langchain_core.embeddings import Embeddings
+
+
+class ANEEmbeddings(Embeddings):
+    """A LangChain `Embeddings` that runs a sentence encoder on the Apple Neural Engine.
+
+    The encoder runs on the ANE through ANEForge (no CoreML), matching the fp32
+    reference to cosine ~1.0 at a fraction of the GPU's energy. `model` is any
+    BERT-family sentence encoder on the Hugging Face hub; its pooling mode and
+    normalize flag are read from the model's own sentence-transformers config, so
+    mean-pooled (MiniLM, E5) and cls-pooled (BGE, GTE) models both come out right.
+
+        from langchain_aneforge import ANEEmbeddings
+        emb = ANEEmbeddings(model="BAAI/bge-small-en-v1.5")
+        vec = emb.embed_query("hello world")
+
+    Pass `int8=True` to stream int8 weights (half the size, cosine ~0.9999).
+    Runs only on Apple Silicon macOS, where ANEForge can reach the engine.
+    """
+
+    def __init__(self, model: str = "sentence-transformers/all-MiniLM-L6-v2", int8: bool = False) -> None:
+        from aneforge.sentence_transformers import SentenceTransformer  # lazy: Apple-Silicon only
+        self.model = model
+        self._st = SentenceTransformer(model, int8=int8)
+
+    def embed_documents(self, texts: List[str]) -> List[List[float]]:
+        vecs = self._st.encode(list(texts))
+        return vecs.astype(float).tolist()
+
+    def embed_query(self, text: str) -> List[float]:
+        return self._st.encode(text).astype(float).tolist()

langchain_aneforge-0.1.0/pyproject.toml

@@ -0,0 +1,40 @@
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[project]
+name = "langchain-aneforge"
+version = "0.1.0"
+description = "LangChain embeddings that run on the Apple Neural Engine, via ANEForge (no CoreML)."
+readme = "README.md"
+requires-python = ">=3.10"
+license = "MIT"
+license-files = ["LICENSE"]
+authors = [{ name = "Spencer Bryngelson", email = "sbryngelson@gmail.com" }]
+keywords = ["langchain", "embeddings", "apple-neural-engine", "ane", "apple-silicon", "aneforge"]
+
+# aneforge reaches the ANE and pulls the encoder deps via its own extras; it runs
+# only on Apple Silicon macOS. langchain-core is the integration surface.
+dependencies = [
+    "langchain-core>=0.3",
+    "aneforge>=0.1.3",
+]
+
+classifiers = [
+    "Development Status :: 3 - Alpha",
+    "Programming Language :: Python :: 3",
+    "Operating System :: MacOS :: MacOS X",
+    "Intended Audience :: Developers",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+]
+
+[project.urls]
+Homepage = "https://github.com/sbryngelson/langchain-aneforge"
+Repository = "https://github.com/sbryngelson/langchain-aneforge"
+"ANEForge" = "https://github.com/sbryngelson/ANEForge"
+
+[tool.hatch.build.targets.wheel]
+packages = ["langchain_aneforge"]
+
+[tool.ruff]
+line-length = 120

langchain_aneforge-0.1.0/tests/test_embeddings.py

@@ -0,0 +1,27 @@
+"""Interface + on-device tests for ANEEmbeddings.
+
+The interface test runs anywhere. The embedding tests need Apple Silicon with a
+working ANEForge install and are skipped otherwise.
+"""
+import importlib.util
+
+import pytest
+
+from langchain_aneforge import ANEEmbeddings
+
+_HAS_ANE = importlib.util.find_spec("aneforge") is not None
+
+
+def test_is_langchain_embeddings():
+    from langchain_core.embeddings import Embeddings
+    assert issubclass(ANEEmbeddings, Embeddings)
+
+
+@pytest.mark.skipif(not _HAS_ANE, reason="needs aneforge on Apple Silicon")
+def test_embed_query_and_documents():
+    emb = ANEEmbeddings("sentence-transformers/all-MiniLM-L6-v2")
+    q = emb.embed_query("hello world")
+    assert isinstance(q, list) and len(q) == 384 and all(isinstance(x, float) for x in q)
+
+    docs = emb.embed_documents(["a passage", "another passage"])
+    assert len(docs) == 2 and len(docs[0]) == 384