cocoindex 0.1.79__cp312-cp312-manylinux_2_28_x86_64.whl → 0.1.80__cp312-cp312-manylinux_2_28_x86_64.whl

cocoindex/functions.py

@@ -116,19 +116,62 @@ class SentenceTransformerEmbedExecutor:
 def _get_colpali_model_and_processor(model_name: str) -> ColPaliModelInfo:
     """Get or load ColPali model and processor, with caching."""
     try:
-        from colpali_engine.models import ColPali, ColPaliProcessor  # type: ignore[import-untyped]
+        from colpali_engine.models import (  # type: ignore[import-untyped]
+            ColPali,
+            ColPaliProcessor,
+            ColQwen2,
+            ColQwen2Processor,
+            ColQwen2_5,
+            ColQwen2_5_Processor,
+            ColIdefics3,
+            ColIdefics3Processor,
+        )
         from colpali_engine.utils.torch_utils import get_torch_device  # type: ignore[import-untyped]
         import torch
     except ImportError as e:
         raise ImportError(
-            "ColPali is not available. Make sure cocoindex is installed with ColPali support."
+            "ColVision models are not available. Make sure cocoindex is installed with ColPali support."
         ) from e
 
     device = get_torch_device("auto")
-    model = ColPali.from_pretrained(
-        model_name, device_map=device, torch_dtype=torch.bfloat16
-    ).eval()
-    processor = ColPaliProcessor.from_pretrained(model_name)
+
+    # Manual model detection based on model name
+    model_name_lower = model_name.lower()
+
+    try:
+        if "qwen2.5" in model_name_lower:
+            model = ColQwen2_5.from_pretrained(
+                model_name,
+                torch_dtype=torch.bfloat16,
+                device_map=device,
+            ).eval()
+            processor = ColQwen2_5_Processor.from_pretrained(model_name)
+        elif "qwen2" in model_name_lower:
+            model = ColQwen2.from_pretrained(
+                model_name,
+                torch_dtype=torch.bfloat16,
+                device_map=device,
+            ).eval()
+            processor = ColQwen2Processor.from_pretrained(model_name)
+        elif "colsmol" in model_name_lower or "smol" in model_name_lower:
+            # ColSmol models use Idefics3 architecture
+            model = ColIdefics3.from_pretrained(
+                model_name,
+                torch_dtype=torch.bfloat16,
+                device_map=device,
+            ).eval()
+            processor = ColIdefics3Processor.from_pretrained(model_name)
+        else:
+            # Default to ColPali
+            model = ColPali.from_pretrained(
+                model_name,
+                torch_dtype=torch.bfloat16,
+                device_map=device,
+            ).eval()
+            processor = ColPaliProcessor.from_pretrained(model_name)
+
+    except Exception as e:
+        raise RuntimeError(f"Failed to load model {model_name}: {e}")
 
     # Get dimension from the actual model
     dimension = _detect_colpali_dimension(model, processor, device)
@@ -167,17 +210,25 @@ def _detect_colpali_dimension(model: Any, processor: Any, device: Any) -> int:
 
 class ColPaliEmbedImage(op.FunctionSpec):
     """
-    `ColPaliEmbedImage` embeds images using the ColPali multimodal model.
+    `ColPaliEmbedImage` embeds images using ColVision multimodal models.
+
+    Supports ALL models available in the colpali-engine library, including:
+    - ColPali models (colpali-*): PaliGemma-based, best for general document retrieval
+    - ColQwen2 models (colqwen-*): Qwen2-VL-based, excellent for multilingual text (29+ languages) and general vision
+    - ColSmol models (colsmol-*): Lightweight, good for resource-constrained environments
+    - Any future ColVision models supported by colpali-engine
 
-    ColPali (Contextual Late-interaction over Patches) uses late interaction
-    between image patch embeddings and text token embeddings for retrieval.
+    These models use late interaction between image patch embeddings and text token
+    embeddings for retrieval.
 
     Args:
-        model: The ColPali model name to use (e.g., "vidore/colpali-v1.2")
+        model: Any ColVision model name supported by colpali-engine
+               (e.g., "vidore/colpali-v1.2", "vidore/colqwen2.5-v0.2", "vidore/colsmol-v1.0")
+               See https://github.com/illuin-tech/colpali for the complete list of supported models.
 
     Note:
         This function requires the optional colpali-engine dependency.
-        Install it with: pip install 'cocoindex[embeddings]'
+        Install it with: pip install 'cocoindex[colpali]'
     """
 
     model: str
@@ -189,7 +240,7 @@ class ColPaliEmbedImage(op.FunctionSpec):
     behavior_version=1,
 )
 class ColPaliEmbedImageExecutor:
-    """Executor for ColPaliEmbedImage."""
+    """Executor for ColVision image embedding (ColPali, ColQwen2, ColSmol, etc.)."""
 
     spec: ColPaliEmbedImage
     _model_info: ColPaliModelInfo
@@ -209,7 +260,7 @@ class ColPaliEmbedImageExecutor:
             import io
         except ImportError as e:
             raise ImportError(
-                "Required dependencies (PIL, torch) are missing for ColPali image embedding."
+                "Required dependencies (PIL, torch) are missing for ColVision image embedding."
             ) from e
 
         model = self._model_info.model
@@ -235,17 +286,25 @@ class ColPaliEmbedImageExecutor:
 
 class ColPaliEmbedQuery(op.FunctionSpec):
     """
-    `ColPaliEmbedQuery` embeds text queries using the ColPali multimodal model.
+    `ColPaliEmbedQuery` embeds text queries using ColVision multimodal models.
+
+    Supports ALL models available in the colpali-engine library, including:
+    - ColPali models (colpali-*): PaliGemma-based, best for general document retrieval
+    - ColQwen2 models (colqwen-*): Qwen2-VL-based, excellent for multilingual text (29+ languages) and general vision
+    - ColSmol models (colsmol-*): Lightweight, good for resource-constrained environments
+    - Any future ColVision models supported by colpali-engine
 
-    This produces query embeddings compatible with ColPali image embeddings
+    This produces query embeddings compatible with ColVision image embeddings
     for late interaction scoring (MaxSim).
 
     Args:
-        model: The ColPali model name to use (e.g., "vidore/colpali-v1.2")
+        model: Any ColVision model name supported by colpali-engine
+               (e.g., "vidore/colpali-v1.2", "vidore/colqwen2.5-v0.2", "vidore/colsmol-v1.0")
+               See https://github.com/illuin-tech/colpali for the complete list of supported models.
 
     Note:
         This function requires the optional colpali-engine dependency.
-        Install it with: pip install 'cocoindex[embeddings]'
+        Install it with: pip install 'cocoindex[colpali]'
     """
 
     model: str
@@ -257,7 +316,7 @@ class ColPaliEmbedQuery(op.FunctionSpec):
     behavior_version=1,
 )
 class ColPaliEmbedQueryExecutor:
-    """Executor for ColPaliEmbedQuery."""
+    """Executor for ColVision query embedding (ColPali, ColQwen2, ColSmol, etc.)."""
 
     spec: ColPaliEmbedQuery
     _model_info: ColPaliModelInfo
@@ -275,7 +334,7 @@ class ColPaliEmbedQueryExecutor:
             import torch
         except ImportError as e:
             raise ImportError(
-                "Required dependencies (torch) are missing for ColPali query embedding."
+                "Required dependencies (torch) are missing for ColVision query embedding."
             ) from e
 
         model = self._model_info.model

{cocoindex-0.1.79.dist-info → cocoindex-0.1.80.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cocoindex
-Version: 0.1.79
+Version: 0.1.80
 Requires-Dist: click>=8.1.8
 Requires-Dist: rich>=14.0.0
 Requires-Dist: python-dotenv>=1.1.0
@@ -39,7 +39,8 @@ Project-URL: Homepage, https://cocoindex.io/
 [![Documentation](https://img.shields.io/badge/Documentation-394e79?logo=readthedocs&logoColor=00B9FF)](https://cocoindex.io/docs/getting_started/quickstart)
 [![License](https://img.shields.io/badge/license-Apache%202.0-5B5BD6?logoColor=white)](https://opensource.org/licenses/Apache-2.0)
 [![PyPI version](https://img.shields.io/pypi/v/cocoindex?color=5B5BD6)](https://pypi.org/project/cocoindex/)
-[![PyPI - Downloads](https://img.shields.io/pypi/dm/cocoindex)](https://pypistats.org/packages/cocoindex)
+<!--[![PyPI - Downloads](https://img.shields.io/pypi/dm/cocoindex)](https://pypistats.org/packages/cocoindex) -->
+[![PyPI Downloads](https://static.pepy.tech/badge/cocoindex/month)](https://pepy.tech/projects/cocoindex)
 [![CI](https://github.com/cocoindex-io/cocoindex/actions/workflows/CI.yml/badge.svg?event=push&color=5B5BD6)](https://github.com/cocoindex-io/cocoindex/actions/workflows/CI.yml)
 [![release](https://github.com/cocoindex-io/cocoindex/actions/workflows/release.yml/badge.svg?event=push&color=5B5BD6)](https://github.com/cocoindex-io/cocoindex/actions/workflows/release.yml)
 [![Discord](https://img.shields.io/discord/1314801574169673738?logo=discord&color=5B5BD6&logoColor=white)](https://discord.com/invite/zpA9S2DR7s)
@@ -216,6 +217,8 @@ It defines an index flow like this:
 | [Paper Metadata](examples/paper_metadata) | Index papers in PDF files, and build metadata tables for each paper |
 | [Multi Format Indexing](examples/multi_format_indexing) | Build visual document index from PDFs and images with ColPali for semantic search |
 | [Custom Output Files](examples/custom_output_files) | Convert markdown files to HTML files and save them to a local directory, using *CocoIndex Custom Targets* |
+| [Patient intake form extraction](examples/patient_intake_extraction) | Use LLM to extract structured data from patient intake forms with different formats |
+
 
 More coming and stay tuned 👀!
 

{cocoindex-0.1.79.dist-info → cocoindex-0.1.80.dist-info}/RECORD

@@ -1,14 +1,14 @@
-cocoindex-0.1.79.dist-info/METADATA,sha256=1RyDQg5xjdf954JY3IBfZaAe5u6SQ2x-sDcPJDlSuPQ,11799
-cocoindex-0.1.79.dist-info/WHEEL,sha256=agcEWikPie1qUk10ElUHI4WcN5vs5MjhJbp7oethB0A,108
-cocoindex-0.1.79.dist-info/entry_points.txt,sha256=_NretjYVzBdNTn7dK-zgwr7YfG2afz1u1uSE-5bZXF8,46
-cocoindex-0.1.79.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+cocoindex-0.1.80.dist-info/METADATA,sha256=DNAMbPaFYBElOtqshXIvRW_MNmJgcqNHY6A5uN9NcQ8,12073
+cocoindex-0.1.80.dist-info/WHEEL,sha256=agcEWikPie1qUk10ElUHI4WcN5vs5MjhJbp7oethB0A,108
+cocoindex-0.1.80.dist-info/entry_points.txt,sha256=_NretjYVzBdNTn7dK-zgwr7YfG2afz1u1uSE-5bZXF8,46
+cocoindex-0.1.80.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
 cocoindex/__init__.py,sha256=sLpSVO5Cotgn_82lawxvXnaqfa-qj33rytWBAe2MTtU,2201
-cocoindex/_engine.cpython-312-x86_64-linux-gnu.so,sha256=wOmweA_mrr2KX_41KXrVfwbCLS9rcGoYP2tJ1Esx3vE,71022048
+cocoindex/_engine.cpython-312-x86_64-linux-gnu.so,sha256=QwiJs1O5W2RvEHPNRj7q-eVbTkkRKldgln8Lhxo5rt0,72116944
 cocoindex/auth_registry.py,sha256=PE1-kVkcyC1G2C_V7b1kvYzeq73OFQehWKQP7ln7fJ8,1478
 cocoindex/cli.py,sha256=-gp639JSyQN6YjnhGqCakIzYoSSqXxQMbxbkcYGP0QY,22359
 cocoindex/convert.py,sha256=HodeDl1HVX8nnBH02lQKarw5i3xmkjB0nGj-DXt7Ifc,18284
 cocoindex/flow.py,sha256=egKbBG2X9DjAqmcATcndyRhe9zMZHRd-YxKCpt9BsUg,36551
-cocoindex/functions.py,sha256=34sZWoS0zGnaKyooIODQgc6QEPZKiJoWhfb8jKIWwps,9528
+cocoindex/functions.py,sha256=-8tAW7_HhSw-A7M8U_C1vUfE9jxNPJ6j2yBRJvP16Tk,12302
 cocoindex/index.py,sha256=j93B9jEvvLXHtpzKWL88SY6wCGEoPgpsQhEGHlyYGFg,540
 cocoindex/lib.py,sha256=f--9dAYd84CZosbDZqNW0oGbBLsY3dXiUTR1VrfQ_QY,817
 cocoindex/llm.py,sha256=Pv_cdnRngTLtuLU9AUmS8izIHhcKVnuBNolC33f9BDI,851
@@ -28,4 +28,4 @@ cocoindex/tests/test_validation.py,sha256=X6AQzVs-hVKIXcrHMEMQnhfUE8at7iXQnPq8nH
 cocoindex/typing.py,sha256=qQ0ANF3iuQDeSqipHgL2SDiiXL2reTMUN0aj4ve_T0w,13359
 cocoindex/utils.py,sha256=hUhX-XV6XGCtJSEIpBOuDv6VvqImwPlgBxztBTw7u0U,598
 cocoindex/validation.py,sha256=PZnJoby4sLbsmPv9fOjOQXuefjfZ7gmtsiTGU8SH-tc,3090
-cocoindex-0.1.79.dist-info/RECORD,,
+cocoindex-0.1.80.dist-info/RECORD,,