Files
lancedb/python/pyproject.toml
Poornachandra.A.N 7d0127b376 feat(embeddings): add siglip embedding support to lancedb (#2499)
###  Summary

This PR adds **SigLIP** (Sigmoid Loss Image Pretraining) as a new
embedding model in the LanceDB embedding registry. SigLIP improves
image-text alignment performance using sigmoid-based contrastive loss
and offers robust zero-shot generalization.

Fixes #2498 

### What’s Implemented

#### 1. `SigLIP` Embedding Class

* Added `SigLIP` support under `python/lancedb/embeddings/siglip.py`
* Implements:

  * `compute_source_embeddings`
  * `_batch_generate_embeddings`
  * Normalization logic
  * Batch-wise progress logging for image embedding

#### 2. Registry Integration

* Registered `SigLIP` in `embeddings/__init__.py`
* `SigLIP` now usable via `connect(..., embedding="siglip")`

#### 3. Evaluation Benchmark Support

* Added SigLIP to `test_embeddings_slow.py` for side-by-side
benchmarking with OpenCLIP and ImageBind


###  New Test Methods

####  `test_siglip`

* End-to-end test to verify embeddings table creation and vector shape
for SigLIP
![WhatsApp Image 2025-07-10 at 18 00
27_a3368163](https://github.com/user-attachments/assets/e5582ee1-80a3-43d7-a7a1-26ceecce9f4d)


####  `test_siglip_vs_openclip_vs_imagebind_benchmark_full`

* Benchmarks:

  * **Recall\@1 / 5 / 10**
  * **mAP (Mean Average Precision)**
  * **Embedding & Search Latency**
  * Dimensionality reporting
![WhatsApp Image 2025-07-10 at 18 12
13_22c67a84](https://github.com/user-attachments/assets/455bf30f-62b7-4684-a3f3-ad52e2a1ffe5)


###  Notes

* SigLIP outputs 768D embeddings (vs 512D for OpenCLIP)
* Benchmark shows competitive performance despite higher dimensionality
* I'm still new to contributing to open-source and learning as I go.
Please feel free to suggest any improvements — I'm happy to make
changes!
2025-08-04 11:42:39 -07:00

140 lines
3.5 KiB
TOML

[project]
name = "lancedb"
# version in Cargo.toml
dynamic = ["version"]
dependencies = [
"deprecation",
"numpy",
"overrides>=0.7",
"packaging",
"pyarrow>=16",
"pydantic>=1.10",
"tqdm>=4.27.0",
]
description = "lancedb"
authors = [{ name = "LanceDB Devs", email = "dev@lancedb.com" }]
license = { file = "LICENSE" }
readme = "README.md"
requires-python = ">=3.9"
keywords = [
"data-format",
"data-science",
"machine-learning",
"arrow",
"data-analytics",
]
classifiers = [
"Development Status :: 3 - Alpha",
"Environment :: Console",
"Intended Audience :: Science/Research",
"License :: OSI Approved :: Apache Software License",
"Operating System :: OS Independent",
"Programming Language :: Python",
"Programming Language :: Python :: 3",
"Programming Language :: Python :: 3 :: Only",
"Programming Language :: Python :: 3.9",
"Programming Language :: Python :: 3.10",
"Programming Language :: Python :: 3.11",
"Programming Language :: Python :: 3.12",
"Topic :: Scientific/Engineering",
]
[project.urls]
repository = "https://github.com/lancedb/lancedb"
[project.optional-dependencies]
pylance = [
"pylance>=0.25",
]
tests = [
"aiohttp",
"boto3",
"pandas>=1.4",
"pytest",
"pytest-mock",
"pytest-asyncio",
"duckdb",
"pytz",
"polars>=0.19, <=1.3.0",
"tantivy",
"pyarrow-stubs",
"pylance>=0.25",
"requests",
"datafusion",
]
dev = [
"ruff",
"pre-commit",
"pyright",
'typing-extensions>=4.0.0; python_version < "3.11"',
]
docs = ["mkdocs", "mkdocs-jupyter", "mkdocs-material", "mkdocstrings-python"]
clip = ["torch", "pillow", "open-clip-torch"]
siglip = ["torch", "pillow", "transformers>=4.41.0","sentencepiece"]
embeddings = [
"requests>=2.31.0",
"openai>=1.6.1",
"sentence-transformers",
"torch",
"pillow",
"open-clip-torch",
"cohere",
"colpali-engine>=0.3.10",
"huggingface_hub",
"InstructorEmbedding",
"google.generativeai",
"boto3>=1.28.57",
"awscli>=1.29.57",
"botocore>=1.31.57",
'ibm-watsonx-ai>=1.1.2; python_version >= "3.10"',
"ollama>=0.3.0",
"sentencepiece"
]
azure = ["adlfs>=2024.2.0"]
[tool.maturin]
python-source = "python"
module-name = "lancedb._lancedb"
[build-system]
requires = ["maturin>=1.4"]
build-backend = "maturin"
[tool.ruff.lint]
select = ["F", "E", "W", "G", "PERF"]
[tool.pytest.ini_options]
addopts = "--strict-markers --ignore-glob=lancedb/embeddings/*.py"
markers = [
"slow: marks tests as slow (deselect with '-m \"not slow\"')",
"asyncio",
"s3_test",
]
[tool.pyright]
include = [
"python/lancedb/index.py",
"python/lancedb/rerankers/util.py",
"python/lancedb/rerankers/__init__.py",
"python/lancedb/rerankers/voyageai.py",
"python/lancedb/rerankers/jinaai.py",
"python/lancedb/rerankers/openai.py",
"python/lancedb/rerankers/cross_encoder.py",
"python/lancedb/rerankers/colbert.py",
"python/lancedb/rerankers/answerdotai.py",
"python/lancedb/rerankers/cohere.py",
"python/lancedb/arrow.py",
"python/lancedb/__init__.py",
"python/lancedb/types.py",
"python/lancedb/integrations/__init__.py",
"python/lancedb/exceptions.py",
"python/lancedb/background_loop.py",
"python/lancedb/schema.py",
"python/lancedb/remote/__init__.py",
"python/lancedb/remote/errors.py",
"python/lancedb/embeddings/__init__.py",
"python/lancedb/_lancedb.pyi",
]
exclude = ["python/tests/"]
pythonVersion = "3.12"