Feed aggregator

Article URL: https://www.youtube.com/watch?v=6Ide5pRLR8Y

Comments URL: https://news.ycombinator.com/item?id=49141228

Points: 1

# Comments: 0

Last week a few of us pretrained a 15M-parameter language model from scratch — past its Chinchilla-optimal token budget — with no cluster, no server, and no budget. The training coordinator is a GitHub Actions cron job. The gradients are pull requests. By the end, someone we'd never met had joined and trained 23M tokens of it. github.com/commonsense-ai/coop — Model + card · huggingface.co/commonsense-ai/tinystories-15m — Gradient inbox — Leaderboard

Comments URL: https://news.ycombinator.com/item?id=49141174

Points: 1

# Comments: 0

Article URL: https://www.wsj.com/style/fashion/naked-dressing-hailey-bieber-madonna-d2a1d94b

Comments URL: https://news.ycombinator.com/item?id=49141127

Points: 1

# Comments: 0

Article URL: https://codeberg.org/markdascher/moochacha

Comments URL: https://news.ycombinator.com/item?id=49141121

Points: 2

# Comments: 0

Article URL: https://codeberg.org/amano.kenji/j3blocks

Comments URL: https://news.ycombinator.com/item?id=49141115

Points: 1

# Comments: 0

CCN Strips comments out of codebases, and comments ONLY, this isn't a regex parse that could potentially eat your code, no this actually makes a list and checks it twice.

I, as a real developer, often leave bytes of comments in my own codebases then forget about them, AI models leave MEGABYTES and has NO memory, sometimes literally using comments as a FORM OF IT.

Use this tool to clean your codebases of literal trash, or use it to review, very useful, end to end tested and verified against 2,700 iterations (not that you care or anything, it works, even on debian trixie with xfce desktop, and that's what matters, not the thousands of tests I ripped from the codebase before publishing it that end users dgaf about)

As Stallman infamously says...

Happy hacking ;)

Comments URL: https://news.ycombinator.com/item?id=49141107

Points: 1

# Comments: 0

Article URL: https://www.wafer.ai/blog/kimi-k3-mi355x

Comments URL: https://news.ycombinator.com/item?id=49141073

Points: 7

# Comments: 0

Article URL: https://www.hdlfactory.com/post/2026/07/19/cocoapuffs-booting-fuchsias-zircon-kernel-on-a-risc-v-core-in-artix-7-fpga/

Comments URL: https://news.ycombinator.com/item?id=49141057

Points: 3

# Comments: 0

I am an engineer and i spend so much of my time using claude code, and its started out being great because I was able to get a lot done but its also incredibly frustrating. I spend so much time waiting for the agents to complete running, during which time i end up moving to another task. The context switching is so tedious.

For more complex tasks, I would like for it to happen without my input once i have given it multiple steps of instructions and let it to run. None of this is typically extremely complex instructions so it is something that claude code should be able to come up with on its own. Its just that it needs to ensure that each step works before moving to the next step (including retries, feedback to make improvements etc).

I have setup rules to do this and it works to an extent but it doesnt work when i have to E2E test things in a separate environment (like on a pre production environment). I would like for it to run its own tests, look at the multiple sets of logs and verify that its working and then get to the next step or retry. Kind of like a feedback loop that keeps getting monitored at all hours of the day.

Comments URL: https://news.ycombinator.com/item?id=49141048

Points: 1

# Comments: 1

Remembered there were a few top YC pitch decks from early on, does anyone know a placed to download them?

Comments URL: https://news.ycombinator.com/item?id=49141045

Points: 3

# Comments: 0

Article URL: https://clickhouse.com/blog/strict-memory-overcommit-for-postgres

Comments URL: https://news.ycombinator.com/item?id=49141028

Points: 1

# Comments: 0

Article URL: https://isaacsu.com/2026/08/dont-credit-the-llm/

Comments URL: https://news.ycombinator.com/item?id=49141008

Points: 7

# Comments: 3

Article URL: https://arxiv.org/abs/2607.27951

Comments URL: https://news.ycombinator.com/item?id=49141003

Points: 1

# Comments: 0

Article URL: https://twitter.com/karpathy/status/2083749667410727319

Comments URL: https://news.ycombinator.com/item?id=49140998

Points: 2

# Comments: 0

Article URL: https://thehub.ca/2026/04/03/can-anyone-solve-canadas-brain-drain-problem/

Comments URL: https://news.ycombinator.com/item?id=49140996

Points: 1

# Comments: 0

Article URL: https://gurevich.ca/simple-outlier-detection-in-a-kalman-filter/

Comments URL: https://news.ycombinator.com/item?id=49140948

Points: 1

# Comments: 0

Article URL: https://runtimewire.com/article/futurism-says-meta-has-almost-nothing-to-show-for-ai-the-numbers-say-otherwise

Comments URL: https://news.ycombinator.com/item?id=49140938

Points: 3

# Comments: 6

Article URL: https://www.axios.com/2026/08/01/deepseek-model-cheap-ai-price-war

Comments URL: https://news.ycombinator.com/item?id=49140923

Points: 1

# Comments: 1

Article URL: https://www.science.org/content/article/deep-sea-vehicles-spot-alien-sharks-deep-beneath-waves-pacific#

Comments URL: https://news.ycombinator.com/item?id=49140599

Points: 2

# Comments: 0