7.2 KiB
title, draft, slug, short_description, description, preview_image, social_preview_image, date, author, featured, tags
| title | draft | slug | short_description | description | preview_image | social_preview_image | date | author | featured | tags | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Optimizing ColPali for Retrieval at Scale, 13x Faster Results | false | colpali-qdrant-optimization | Learn how we optimized ColPali, a document retrieval system, to be 13x faster for datasets of 20,000+ PDFs. Discover how pooling and reranking boost scalability while maintaining precision. | ColPali combines text and visual contexts for precise document retrieval, but scaling posed challenges. We achieved 13x faster retrieval using pooling and reranking, reducing vector counts by ~30x while retaining near-original accuracy. Explore the open-source demo to optimize your retrieval workflows! | /blog/colpali-optimization/preview-2.png | /blog/colpali-optimization/preview-2.png | 2024-11-27T00:40:24-03:00 | Evgeniya Sukhodolskaya | yes |
|
ColPali is a fascinating leap in document retrieval. Its precision in handling visually rich PDFs is phenomenal, but scaling it to handle real-world datasets comes with its share of computational challenges.
Here's how we solved these challenges to make ColPali 13x faster without sacrificing the precision it’s known for.
The Scaling Dilemma
ColPali generates ~1,030 vectors for just one page of a PDF. While this is manageable for small-scale tasks, in a real-world production setting where you may need to store hundreds od thousands of PDFs, the challenge of scaling becomes significant.
Consider this scenario:
- Dataset Size: 20,000 PDF pages.
- Vector Explosion: Each page generates ~1,030 vectors of 128 dimensions.
- Index Complexity: Trillions of comparisons to build the index comparing every vector pair.
Even advanced indexing algorithms like HNSW struggle with this scale, as computational costs grow quadratically.
We turned to a hybrid optimization strategy combining pooling (to reduce computational overhead) and reranking (to preserve accuracy).
Before we go any deeper, watch our Webinar video for the full demo walkthrough.
For those eager to explore, the codebase is available here.
Two-Stage Retrieval Process
Pooling
Pooling is well-known in machine learning as a way to compress data while keeping important information intact. For ColPali, we reduced ~1,030 vectors per page to just 38 vectors by pooling rows in the document's 32x32 grid.
The most popular types of pooling are:
- Mean Pooling: Averages values across rows.
- Max Pooling: Selects the maximum value for each feature.
32 vectors represent the pooled rows, while an additional 6 vectors encode contextual information derived from ColPali’s special tokens (e.g., for the beginning of the sequence, and task-specific instructions like “Describe the image”).
For our experiments, we chose to preserve these 6 additional vectors.
The "ColPali as a Reranker" Experiment
Pooling drastically reduces retrieval costs, but there’s a risk of losing fine-grained precision. To address this, we implemented a two-stage retrieval system with the pooled vectors for the initial retrieval stage and the original ColPali model for reranking:
- Pooled Retrieval: Quickly retrieves the top 200 candidates using lightweight pooled embeddings.
- Full Reranking: Refines these candidates using the original, high-resolution embeddings, delivering the final top 20 results.
This approach delivers speed without compromising on retrieval quality.
Implementation
We created a custom dataset with over 20,000 unique PDF pages by merging:
- ViDoRe Benchmark: Designed for document retrieval evaluation.
- UFO Dataset: Visually rich documents paired with synthetic queries.
- DocVQA Dataset: A large set of document-derived Q&A pairs.
Each document was processed into 32x32 grids, generating both full-resolution and pooled embeddings.
These embeddings were stored in the Qdrant vector database, configured for speed:
- Full-Resolution Embeddings: ~1,030 vectors per page.
- Pooled Embeddings: Mean and max pooling variants.
All embeddings were kept in RAM to ensure consistent retrieval performance.
Experiment Setup
We evaluated retrieval quality using 1,000 task-specific queries and the retrieval process followed the two-stage approach:
- Pooled embeddings retrieved the top 200 candidates.
- Full-resolution embeddings reranked these candidates to produce the final top 20 results.
To measure performance, we used:
- NDCG@20: Measures ranking quality (how well the top results align with expectations).
- Recall@20: Measures the overlap between pooled and full-resolution retrievals.
Results
The experiments gave us some very promissing results:
- Speed: Retrieval time improved 13x compared to full-resolution embeddings alone.
- Accuracy: Mean pooling preserved retrieval quality nearly identical to the original ColPali.
Metrics
| Pooling Type | NDCG@20 | Recall@20 |
|---|---|---|
| Mean | 0.952 | 0.917 |
| Max | 0.759 | 0.656 |
Mean pooling offered the ideal balance, combining speed and precision, while max pooling delivered faster results but at the cost of noticeable quality degradation.
Other Insights
We also explored additional optimizations, including removing <pad> tokens, which slightly improved speed without significant quality loss, and applying binary quantization, which increased retrieval speed but introduced minor accuracy drops.
What’s Next?
Future experiments could push these results even further:
- Investigating column-wise pooling for additional compression.
- Testing hybrid half-precision (float16) vectors to balance memory use and speed.
- Skipping special multivectors during prefetch to streamline retrieval.
- Combining quantization with oversampling for even faster search.
Try It Yourself
Curious to see this in action? Explore the full codebase and experiment with ColPali optimizations:
- Demo Notebook: GitHub Repository
- Webinar Walkthrough: Watch Here
Join the community and share your results!