







An overview of perceptual hashing
Robustness of Practical Perceptual Hashing Algorithms to...
Perceptual hashing algorithms (PHAs) are widely used for identifying illegal online content and are thus integral to various sensitive applications. However, due to their hasty deployment in...

Hash Functions for GPU Rendering – Nathan Reed’s coding blog
Pixels and polygons and shaders, oh my!
Hash-based Signatures: An illustrated Primer
Over the past several years I’ve been privileged to observe two contradictory and fascinating trends. The first is that we’re finally starting to use the cryptography that researchers h…

Random Hash Generator - BenNadel.com
The following hashes are generated from 300-bytes of random input using the "sha1prng" algorithm.
Fast Embeddings on GPUs
Fast and accurate search is vital to all of Perplexity, from Search and Computer to our API Platform. Behind the scenes, the heavy lifting is done by embedding

Similarity
Search, explore and see related notes based on their meaning (semantics). Using a tiny AI running fully on-device: data never leaves your machine, no API keys needed. Especially great for journaling.
Nectar - Find similar images across the ATmosphere
Nectar is an image search API for AT Protocol. Query images and records by perceptual hash. Find reposts, add annotations, track popularity.
Bab
Bab is a family of secure hash functions, heavily inspired by BLAKE3. Bab hashes allow for streaming verification of strings, similar to the BLAKE3-based Bao algorithm. We discuss several optimization techniques that go beyond the Bao specification. Further, unlike BLAKE3 and Bao, Bab digests allow for constant-sized length proofs of their strings.
Shared Geometry As A Rosetta Stone: Cross-Modal Alignment Without Paired Data
Multimodal representations enable zero-shot classification and retrieval, but aligning independently trained models usually requires large amounts of paired data. Yet, the Platonic Representation Hypothesis suggests that models trained on different modalities may converge spontaneously toward a shared representation geometry. But then, do we even need paired examples for cross-modal alignment? Remarkably, we show that paired examples are unnecessary for coarse cross-modal alignment. Our simple Wasserstein Procrustes method with a coarse geometric initialization aligns two disjoint embedding sets by estimating a single orthogonal map without seeing any pairs. Across datasets, modalities, and unimodal models, we show that we can consistently align independently trained representations without pairs, and standard geometric alignment metrics accurately predict when this is possible. Nevertheless, we can naturally benefit from paired examples. In the very few-pair regime, our method substantially outperforms existing ones, while staying competitive with pair-based methods with more added examples. Finally, we demonstrate that the resulting alignments can enable text-to-image generation without paired examples. These results show that independently trained models often share enough geometry to establish cross-modal correspondence with little or no paired data.

Google for Developers Blog - News about Web, Mobile, AI and Cloud
EmbeddingGemma 2 is a compact, open-source multimodal embedding model that maps text, code, images, video, and audio into a unified 768-dimensional space. Developers can use the sentence-transformers library to selectively load modular modality encoders—ranging from 270M to 740M parameters—to optimize memory usage. Additionally, Matryoshka Representation Learning enables dynamic dimension truncation down to 128d, significantly reducing vector database storage requirements while maintaining high retrieval performance.

Functional hash tables explained — Spritely Institute
For those of us that use the Lisp family of programming languages, we have much appreciation for the humble pair. Using pairs, we can construct singly-linked lists and key/value mappings called association lists. Lists built from pairs have the pleasant property of being immutable (if you abstain from using setters!) and persistent: extending a list with a new element creates a new list that shares all the data from the original list. Adding to a list is a constant time operation, but lookup is linear time. Thus, lists are not appropriate when we need constant time lookup. For that, we need hash tables.
Linkup - SPARSEUP: Linkup's Open-Source Sparse Embedding Model
Linkup Research releases SPARSEUP, an open-source sparse embedding model under 150M parameters. SOTA on BEIR-13, sub-millisecond retrieval, Apache 2.0 license.

Similar Songs Finder | Playlist Generator - Chosic
Need to find similar songs to a song you like? An awesome playlist created for you in seconds? Spotify playlist generator make it very easy

Tyler Angert on Twitter / X
we're working on a new photo search engine called Angles. it's focused on doing one thing well: finding photos + videos by visual similarity.here is raw, unedited footage of me going through my 80,000 photo camera roll *in realtime* with text-to-image search, image-to-image… pic.twitter.com/fsCqIHv00j— Tyler Angert (@tylerangert) March 24, 2026