Hacker News: duchenne

New comment by duchenne in "What AI did to stackoverflow in a graph"

duchenne — Sat, 18 Jul 2026 14:02:35 +0000

I am a long-time user of stackoverflow with 16k points, and even I got all my questions of the last five years downvoted into oblivion.

New comment by duchenne in "First atmosphere found on Earth-like planet in habitable zone of distant star"

duchenne — Sat, 18 Jul 2026 09:23:38 +0000

Nuclear-powered ion thrusters could solve this issue. They provide low acceleration for a long time consuming very little consumable. This would allow the telescope to stay at the right position for observation.

New comment by duchenne in "Mistral's Robostral Navigate: a state of the art robotics navigation model"

duchenne — Wed, 08 Jul 2026 21:16:21 +0000

Yes, it would work.

New comment by duchenne in "Mistral's Robostral Navigate: a state of the art robotics navigation model"

duchenne — Wed, 08 Jul 2026 16:24:40 +0000

I am working in Mistral robotics team. I confirm this is map-less. The only inputs are the text prompt and the front camera rgb image.

New comment by duchenne in "Local AI needs to be the norm"

duchenne — Mon, 11 May 2026 03:38:04 +0000

Cloud models can use batch processing which is significantly more efficient. A local model has basically a batch of one which takes as much time to process as a batch of 100 because the gpu is memory bound and spend most of its time loading the model from vram to the gpu cache while the gpu cores are idle. With a batch of 100 the model loading time and compute time are roughly similar. So local Models have a first 100x lower efficiency. Secondly, local models are idle most of the time waiting for the user to write a prompt, so the efficiency gap is probably more around 1000x.

New comment by duchenne in "California's new bill requires DOJ-approved 3D printers that report themselves"

duchenne — Thu, 19 Feb 2026 22:20:45 +0000

If they are worried about firearms, why don't they target CNC mills rather than 3d printer? Can you even make a firearm in plastic?

Some US company specialize in selling CNC mills specifically for firearms.

Ex: https://realghostguns.com/product/gg3-s-cnc-deposit/

    It is sold with the cut codes for the AR-15, AR-.308, 1911, Polymer80 and AK-47 receivers and frames.

New comment by duchenne in "OpenAI's new open-source model is basically Phi-5"

duchenne — Fri, 08 Aug 2025 10:13:29 +0000

I have done that at meta/FAIR and it is published in the Llama 3 paper. You usually start from a seed. It can be a randomly picked piece of website/code/image/table of contents/user generated data, and you prompt the model to generate data related to that seed. After, you also need to pass the generated data through a series of verifiers to ensure quality.

New comment by duchenne in "DumPy: NumPy except it's OK if you're dum"

duchenne — Sat, 24 May 2025 17:40:59 +0000

Looks awesome. Can we get the same thing for pytorch?

New comment by duchenne in "Ask HN: How do you propose to rebuild industry in a post-apocalypse world?"

duchenne — Tue, 25 Mar 2025 03:39:22 +0000

There is a manga/anime about this: doctor stone.

For the knowledge preservation, I guess that a copy of deepseek has most of the required information. But, it would be hard to run it in a primitive world.

New comment by duchenne in "Ask HN: How do you propose to rebuild industry in a post-apocalypse world?"

duchenne — Tue, 25 Mar 2025 03:36:10 +0000

But we have landfills which are full of great raw materials. I would argue that it is easier to collect steel from a landfill than from a mine during the industrial revolution.

New comment by duchenne in "MacBook Air M4"

duchenne — Thu, 06 Mar 2025 04:38:10 +0000

I had one for years. Never had overheating issues, except if I put it on my blanket for long.

New comment by duchenne in "MacBook Air M4"

duchenne — Thu, 06 Mar 2025 02:28:01 +0000

The asus zenbook pro is great. The 16inch version is not really bulky. It is 2.4kg, 2TB, 3.2k resolution, great design and build quality. $2200

The 14.5 inch version is 1.6kg, 2TB, 2.9k resolution, also great design and build quality. $1700

https://www.asus.com/laptops/for-creators/zenbook/zenbook-pr...

New comment by duchenne in "DeepSeek open source DeepEP – library for MoE training and Inference"

duchenne — Tue, 25 Feb 2025 16:00:33 +0000

Come on... Meta has been refining pytorch for more than a decade. It basically contains all that you need to train LLMs, including the latest technologies. What more do you need? The part of the code that is specific to Meta infrastructure?

New comment by duchenne in "OpenAI says it has evidence DeepSeek used its model to train competitor"

duchenne — Wed, 29 Jan 2025 23:04:07 +0000

The reasoning happens in the chain of thoughts. But OpenAI (aka ClosedAI) doesn't show this part when you use the o1 model, whether through the API or chat. They hide it to prevent distillation. Deepseek, though, has come up with something new.

New comment by duchenne in "AMD Open-Source 1B OLMo Language Models"

duchenne — Sat, 02 Nov 2024 02:23:18 +0000

Training a 1B model on 1T tokens is cheaper than people might think. A H100 GPU can be rented for 2.5$ per hour and can train around 63k tokens per second for a 1B model. So you would need around 4,400 hours of GPU training costing only $11k And costs will keep going down.

New comment by duchenne in "OpenAI to become for-profit company"

duchenne — Thu, 26 Sep 2024 14:10:25 +0000

But, if the non-profit gives all its assets to the new legal entity, shouldn't the new legal entity be taxed heavily? The gift tax rate goes up to 40% in the US. And 40% of the value of openAI is huge.

New comment by duchenne in "FAA grounds SpaceX ahead of historic Polaris Dawn commercial spacewalk"

duchenne — Wed, 28 Aug 2024 21:49:50 +0000

Except that a plane has passengers. But this rocket had none. It did not even have cargo. And it crashed in a pre-evacuated zone. There is no need to have the same level of security for these two situations.

New comment by duchenne in "Large Enough"

duchenne — Wed, 24 Jul 2024 18:28:47 +0000

Counter-intuitively, larger models are cheaper to train. However, smaller models are cheaper to serve. At first, everyone was focusing on training, so the models were much larger. Now, so many people are using AI everyday, so companies spend more on training smaller models to save on serving.

New comment by duchenne in "Llama 3.1"

duchenne — Tue, 23 Jul 2024 15:32:05 +0000

Most SMBs would be able to run it. This is already a huge win for decentralized AI.

New comment by duchenne in "Chang'e-6: Moon dark side samples collected and launched into lunar orbit"

duchenne — Wed, 05 Jun 2024 12:59:16 +0000

In French, it is called the "hidden face of the Moon", obviously because we cannot see it from the Earth point of view.