Hacker News: leodriesch

New comment by leodriesch in "Qwen VLo: From “Understanding” the World to “Depicting” It"

leodriesch — Fri, 27 Jun 2025 16:25:55 +0000

That's what I really like about Flux Kontext, it has similar editing capabilities to the multimodal models, but doesn't mess up the details. The editing with gpt-image-1 only really works for complete style changes like "make this ghibli", but not adding glasses to a photorealistic image and have it retain all the details.

Absolute Zero: Reinforced Self-Play Reasoning with Zero Data

leodriesch — Sun, 11 May 2025 07:07:26 +0000

Article URL: https://arxiv.org/abs/2505.03335

Comments URL: https://news.ycombinator.com/item?id=43951985

Points: 88

# Comments: 19

Does RL Incentivize Reasoning in LLMs Beyond the Base Model?

leodriesch — Tue, 22 Apr 2025 10:24:37 +0000

Article URL: https://limit-of-rlvr.github.io/

Comments URL: https://news.ycombinator.com/item?id=43760625

Points: 84

# Comments: 38

New comment by leodriesch in "SAM 2: Segment Anything in Images and Videos"

leodriesch — Tue, 30 Jul 2024 12:33:33 +0000

I got SAM 1 to work with MPS device on my MacBook Pro M1, don’t know if it works with this one too.

New comment by leodriesch in "SAM 2: Segment Anything in Images and Videos"

leodriesch — Tue, 30 Jul 2024 12:10:41 +0000

I ran it with 3040x3040px images on my MacBook M1 Pro in about 9 seconds + 200ms or so for the masking.

New comment by leodriesch in "SAM 2: Segment Anything in Images and Videos"

leodriesch — Tue, 30 Jul 2024 07:35:51 +0000

I’d guess testing hardware is same as training hardware, so A100. If it was on a mobile device they would have definitely said that.

New comment by leodriesch in "Apple's On-Device and Server Foundation Models"

leodriesch — Mon, 10 Jun 2024 23:52:15 +0000

I think it’s fair to leave it out in the on-device model comparison. 3b is much smaller than 8b, it is obviously not going to be as good as llama 3 if they did not make groundbreaking advancements with the technology.

New comment by leodriesch in "Using a LLM to compress text"

leodriesch — Sun, 05 May 2024 11:39:32 +0000

The model is always wrong, since it predicts a propability distribution over all possible tokens, but the target has 100% possibility for one token and 0 for all others.

New comment by leodriesch in "Level of Gaussians: Real-Time View Synthesis for Millions of Square Meters"

leodriesch — Wed, 01 May 2024 00:59:24 +0000

I am really impressed by the Apple Maps implementation. I think it also uses textured polygons, but does so in a very good looking way and at 120 fps on an iPhone, showing even a whole city in textured 3d.

New comment by leodriesch in "Can we RAG the whole web?"

leodriesch — Tue, 30 Apr 2024 18:04:43 +0000

Brave Search also has one https://brave.com/search/api/

New comment by leodriesch in "CoreNet: A library for training deep neural networks"

leodriesch — Wed, 24 Apr 2024 03:56:26 +0000

How does this compare to MLX? As far as I understand MLX is equivalent to PyTorch but optimized for Apple Silicon.

Is this meant for training MLX models in a distributed manner? Or what is its purpose?

New comment by leodriesch in "Stable-Audio-Demo"

leodriesch — Tue, 13 Feb 2024 09:48:18 +0000

I'd say most are thinking of Midjourneys success in image generation when talking about this kind of progress.

New comment by leodriesch in "OpenAI drops ban on military tools to partner with The Pentagon"

leodriesch — Wed, 17 Jan 2024 00:21:36 +0000

I’d say talent? Outside of OpenAI no team has been able to release a model as capable as GPT-4, and I’m unsure if the CIA has been prioritizing LLM experts in their hiring.

New comment by leodriesch in "Mistral "Mixtral" 8x7B 32k model [magnet]"

leodriesch — Sat, 09 Dec 2023 00:03:10 +0000

Could you link me to a finetune optimized for function calling? I was looking for one a few weeks ago but did not find any.

Nothing Chats – Bring on the blue bubbles

leodriesch — Tue, 14 Nov 2023 16:17:24 +0000

Article URL: https://us.nothing.tech/pages/nothing-chats

Comments URL: https://news.ycombinator.com/item?id=38265251

Points: 1

# Comments: 0

New comment by leodriesch in "Show HN: LLaVaVision: An AI "Be My Eyes"-like web app with a llama.cpp backend"

leodriesch — Mon, 06 Nov 2023 12:46:29 +0000

The purple icon in the shared transcript indicates GPT-4.

Grok, an AI Modeled After the Hitchhiker's Guide to the Galaxy

leodriesch — Sun, 05 Nov 2023 12:50:52 +0000

Article URL: https://twitter.com/xai/status/1721027348970238035?s=12

Comments URL: https://news.ycombinator.com/item?id=38150695

Points: 5

# Comments: 2

New comment by leodriesch in "Dot by New Computer"

leodriesch — Wed, 01 Nov 2023 19:47:54 +0000

It’s a movie about a humanlike personal AI companion.

https://m.imdb.com/title/tt1798709/

New comment by leodriesch in "Dot by New Computer"

leodriesch — Wed, 01 Nov 2023 19:46:50 +0000

Their privacy policy lists OpenAI as one of their partners for data processing, which indicates that this is happening not on your device, and data is also shared with third parties.

For me this is the main counterargument against apps like these. I want to feel free to post any information into this without thinking about who may read or use it.

Local is the only way to go for software like this in my opinion.

New comment by leodriesch in "Google to invest up to $2B in Anthropic"

leodriesch — Sat, 28 Oct 2023 18:05:01 +0000

This is just a prompting hack you can use with any LLM, not exclusive to Claude. But I do like the fact that they include these tricks in their documentation.