241
Abusing vector search for texts, maps, and chess
Vector search isn't just for AI embeddings. Use HNSW for geo-spatial queries, stock covariance, chess positions, text tokens, and multi-modal recommendations with custom metrics.
New
The latest blog posts to take off on Hacker News. 371 posts.
242
Prolog for data science
I demonstrate a widely applicable pattern which integrates Prolog as a critical component in a data science analysis. Analytic methods are used to generate properties about the data under study and Prolog is used to reason about the data via the generated properties. The post includes some examples of piece-wise…
243
I downloaded all 1.6M posts on Bluesky
I downloaded all the posts on Bluesky as of 2023-05-01. Then I did some data analysis on the 1680399 posts from 45457 accounts.
244
What is a Vector Database? (2021)
Discover Vector Databases: How They Work, Examples, Use Cases, Pros & Cons, Selection and Implementation. They have combined capabilities of traditional databases and standalone vector indexes while specializing for vector embeddings.
245
Fixing Hallucination with Knowledge Bases
Large Language Models (LLMs) have a data freshness problem. Even some of the most powerful models, like GPT-4, have no idea about recent events. Let's learn how to fix this.
246
Investigating Linux phantom disk reads
An investigation of weird hardware utilization highlighting some interesting Linux kernel behaviors.
247
Storage on Vercel
New to the Vercel dashboard: Vercel KV, Vercel Postgres, and Vercel Blob. Announcing serverless storage solutions available on Vercel.
248
249
250
Datomic is Free
Nubank is making all editions of Datomic free of licensing fees, released under Apache 2.0 and available directly from Maven Central, no signup required.
251
The part of Postgres we hate the most: Multi-version concurrency control
Andy Pavlo - Carnegie Mellon University,
252
Ways to shoot yourself in the foot with Postgres
Personal website of Phil Booth
253
Modern Transactional Stack
Transactional databases have long been the most critical component of application design. Why? Because a steadfast database is generally the ultimate enforcement point for correctness in a messy, distributed world. Without them we’d overpay and undercharge. We’d lose riders trying to get home from the airport, and…
254
The Sell ∀ ∃ as ∃ ∀ Scam
Artificial intelligence, like machine learning before it, is making big money off what I call the “sell ∀ ∃ as ∃ ∀ scam.” The scam works as follows. Build a sy…
255
256
Faster LZ is not the answer to 150-250 GB video game downloads
When the JPEG folks were working on image compression, they didn't create a better or faster LZ. Instead they developed new approaches. I se...
257
How RocksDB Works
Over the past years, the adoption of RocksDB increased dramatically. It became a standard for embeddable key-value stores.
258
The Inner Workings of Distributed Databases
Comparison of replication options of time-series databases.
259
33TB of text data for a 1T-parameter model
Important: This page summarizes data scaling only, using tokens to parameters as a ratio, and as derived from large language models like GPT-3, Chinchilla, and beyond, linked to the Compute-Optimal scaling laws like Kaplan and Hoffmann/Chinchilla. Please note that compute scaling laws are outside the scope of my…
260
261
262
Temporal quality degradation in AI models
A recent study from MIT, Harvard, The University of Monterrey, and other top institutions showed an experiment where 91% of their ML models degrade over time . This study is one of the first of its kind, where researchers focus on studying machine learning models' behavior after deployment and how their performance…
263
SQL:2023 is finished: Here is what's new
SQL:2023 has been wrapped. The final text has been submitted by the working group to ISO Central Secretariat, and it’s now up to the ISO gods when it will be published. Based on past experience, it could be between a few weeks and a few months.
264
Do you need a vector database?
Spoiler alert: the answer is maybe! Although, my inclusion of the word “actually” betrays my bias. Vector databases are having their day right now. Three different vector DB companies have raised money on valuations up to $700 million (paywall link). Surprisingly, their rise in popularity is not for their “original”…
265
SQLite performance tuning: concurrent reads, multiple GBs and 100k SELECTs/s
SQLite is an embedded SQL database. It’s extremely easy to setup, buildable as a single C file with libraries existing for basically all common programming languages. It doesn’t need any server setup or configuration since the SQL logic is run in the host process, and the database consists of only
266
FerretDB: open-source MongoDB alternative
After many months of development, FerretDB is now production-ready. We are excited to announce the general availability of FerretDB v1.0.
267
SQL Maxis: Why We Ditched RabbitMQ and Replaced It with a Postgres Queue
Embracing our SQL Maxi nature: how we ripped out RabbitMQ from our infrastructure and replaced it with a Postgres queue.
268
The end of a myth: Distributed transactions can scale
This paper appeared in VLDB'17. The paper presents NAM-DB, a scalable distributed database system that uses RDMA (mostly 1-way RDMA) and a ...
269
From deep to long learning?
For the last two years , a line of work in our lab has been to increase sequence length. We thought longer sequences would enable a new era of machine learning foundation models: they could learn from longer contexts, multiple media sources, complex demonstrations, and more. All data ready and waiting to be learned…
270