Tag:colpali
All the articles with the tag "colpali".
Mine the Way Your Model Scores: MaxSim Hard-Negative Mining for a Late-Interaction Student
Updated:The standard way to mine hard negatives for a late-interaction model uses a single-vector cosine teacher, even though the model itself scores with multi-vector MaxSim. So I rebuilt my miner to score the way my model does. Matched mining clearly beat training with no mined negatives, while the cosine approach was barely doing anything at all.
Diminishing Returns and the Art of Knowing When to Stop
I trained three generations of ColQwen3.5, each with more sophisticated optimization than the last. The most optimized version barely beat the previous one on the primary benchmark (+0.0011 nDCG@5). Individual tasks reshuffled substantially, with per-task swings an order of magnitude larger than the aggregate gain.
Implementing Spatially-Grounded Document Retrieval via Patch-to-Region Propagation
A deep dive into my recent research on spatially-grounded document retrieval using ColPali models and OCR bounding boxes, enabling precise region-level retrieval during inference time and without additional training.
Snappy: Your Vision Retrieval Buddy!
How Snappy evolved from the nextjs-fastapi-colpali template into a vision-first document retrieval system
The Most Beautiful RAG: Starring ColPali, Qdrant, Minio and Friends
Updated:An end-to-end, page-level Vision RAG template with ColPali-style embeddings, Qdrant multivector retrieval (with optional binary quantization), and MinIO-backed storage — dockerized and API-first.
Athrael.net