Feed aggregator

Article URL: https://gawk.dev

Comments URL: https://news.ycombinator.com/item?id=48089145

Points: 1

# Comments: 0

Article URL: https://kevinullyott.com/blog/2026-05-05-composer-lock-gitattributes/

Comments URL: https://news.ycombinator.com/item?id=48089100

Points: 1

# Comments: 0

Article URL: https://www.theguardian.com/us-news/2026/may/09/los-angeles-subway-public-transportation

Comments URL: https://news.ycombinator.com/item?id=48089098

Points: 1

# Comments: 0

Article URL: https://jola.dev/posts/running-local-models-on-m4

Comments URL: https://news.ycombinator.com/item?id=48089091

Points: 2

# Comments: 0

Article URL: https://economistwritingeveryday.com/2024/12/13/the-mythology-of-rice-and-beans/

Comments URL: https://news.ycombinator.com/item?id=48089069

Points: 1

# Comments: 0

Article URL: https://dunkels.com/adam/claude-user-space-ip-stack-ping/

Comments URL: https://news.ycombinator.com/item?id=48089049

Points: 1

# Comments: 0

Article URL: https://www.youtube.com/watch?v=c1LjnFKGDhQ

Comments URL: https://news.ycombinator.com/item?id=48089041

Points: 1

# Comments: 0

Article URL: https://xcancel.com/rebelprazz/status/2053606378238009832#m

Comments URL: https://news.ycombinator.com/item?id=48089018

Points: 4

# Comments: 0

Article URL: https://timelock.sh

Comments URL: https://news.ycombinator.com/item?id=48089009

Points: 3

# Comments: 1

Article URL: https://en.wikipedia.org/wiki/Proprioception

Comments URL: https://news.ycombinator.com/item?id=48088974

Points: 2

# Comments: 0

Article URL: https://www.bloomberg.com/news/articles/2026-05-07/dc-s-metro-makes-a-case-for-driverless-red-line-subway-trains

Comments URL: https://news.ycombinator.com/item?id=48088955

Points: 2

# Comments: 0

Google doesn't allow non-business accounts to receive invoices. I must sign up for a Workspace subscription (which I neither want nor need) to be able to receive an invoice in the format I need for my Gemini subscription to be charged as a business expense. It's silly, as I do not need to do such things on OpenAI's and Anthropic's respective services - A two second settings change. Therefore, I have to leave Gemini.

Comments URL: https://news.ycombinator.com/item?id=48088914

Points: 3

# Comments: 2

And if so, what kinds of short stories tend to be most successful in this regard?

Comments URL: https://news.ycombinator.com/item?id=48088891

Points: 2

# Comments: 2

Article URL: https://hackaday.com/wp-content/uploads/2024/02/WEIZENBAUM-1966-ELIZA-A-Computer-Program-For-the-Study-of-Natural-Language-Communication-Between-Man-And-Machine.pdf

Comments URL: https://news.ycombinator.com/item?id=48088860

Points: 3

# Comments: 1

What are some great resources for people who looking to break into this field or improve their skillset in general

I would love to hear some recommendations

Comments URL: https://news.ycombinator.com/item?id=48088849

Points: 4

# Comments: 3

I built 1e4.ai - a chess web app where you play against neural networks trained to mimic human Lichess players at specific Elo ranges. There's a separate model for each 100-point rating bucket from ~800 to 2200+, and the bots not only choose human-like moves but also burn clock time, play worse under time pressure, and blunder in human-like ways.

Live demo: https://1e4.ai Code: https://github.com/thomasj02/1e4_ai

A few things that might be interesting:

- Trained on almost a full year of Lichess blitz games, around 1B total games

- Architecture is an a small (~9MM parameters) transformer-based network that takes the board, recent move history, the player's rating, and remaining clock time as input. Three separate models per rating bucket: move, clock-usage, and win probability. The clock model is what makes the bots feel humanish under time pressure rather than instant. Because the move model takes the clock as one input parameter, it also learns to blunder under time pressure like a human might.

- Because the network is so tiny, no GPU is needed for inference - it runs easily on a local CPU

- Downside of the tiny network is that it's a bit weak as you turn up the rating past around 1700. It can spot short tactics but not long multi-move combinations.

- Initial training on a rented 8xH100 cluster, then fine-tunes on my local GPU for different rating ranges

- Inspired by Maia-2 and DeepMind's "Grandmaster-Level Chess Without Search". On a held-out Lichess blitz benchmark, the it beats Maia-2 blitz on top-1 move prediction (56.7% vs 52.7%) and pretty substantially on win-probability calibration (Brier 0.176 vs 0.272). Numbers and code in https://github.com/thomasj02/1e4_ai/tree/master/experiments/...

- The data pipeline is C++ via nanobind, then training with Pytorch. Getting this right was actually the thing I spent the most time on. Pre-shuffling the dataset and then being able to read the shuffled dataset sequentially at training time kept the GPU utilization high. Without this it spent a huge percentage of time on I/O while the GPU sat idle.

Happy to answer questions about the rating-conditioning, the clock model, or the data pipeline.

Comments URL: https://news.ycombinator.com/item?id=48088819

Points: 5

# Comments: 0

Article URL: https://footbeen.com/blog/i-built-a-production-app-with-ai-no-developers

Comments URL: https://news.ycombinator.com/item?id=48088764

Points: 1

# Comments: 0

Article URL: https://electrek.co/2026/05/10/frankfurt-expands-commercial-ev-fleet-with-10-new-vocational-trucks/

Comments URL: https://news.ycombinator.com/item?id=48088737

Points: 1

# Comments: 0

Article URL: https://mkenely.com/publications/preprints/large-scale-photogrammetric-st-johns.pdf

Comments URL: https://news.ycombinator.com/item?id=48088700

Points: 1

# Comments: 0