frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

A General Goal-Conditioned Minecraft Model

https://pantograph.com/journal/pan-1
17•agajews•1h ago

Comments

xnx•1h ago
Reminds me of Google's AI Dreamer that could mine diamonds in Minecraft (https://www.scientificamerican.com/article/google-deepmind-t...) or OpenAI training on video to play Minecraft (https://openai.com/index/vpt/)
agajews•1h ago
Very much inspired by those papers! One of the things that's interesting about our model is it's goal-conditioned, so it can do any task at inference time without training on it. We had a lot of fun making eval environments after we trained the model trying to find interesting things it can do, and that was all after we trained the model. More like prompting an LLM.

(Versus Dreamer, which needs to be trained on a hand-written reward function for each task that you want to do.)

agajews•1h ago
Hey everyone! I'm Alex, one of the founders of Pantograph. We've spent the last six months building a pretty smart Minecraft model, coming soon to a server near you!

We trained it on about 500k hours of Minecraft videos, and it learned how to fight creepers, build walls and other structures, and explore to find visual goals.

We're considering putting up a public API for larger models like this one, let us know if you'd like to be able to put Pan in your own server :)

What's most interesting about the model isn't the performance that it gets in Minecraft, but how general the method is. When we scale it up, it should be able to act in any kind of video game, as well as robots in the real world (which are really just another video game).

codeulike•40m ago
Does it have language skills? I always thought it would be interesting to train a model within minecraft as a sortof proxy for 'embodiment'. You could then try asking it about its experiences. "whats your favourite food?", "How does it feel when you hear a spider?", "how low does your food bar need to go before it feels really urgent?" etc
agajews•31m ago
Not yet, but we'll add language as a modality to the larger models! The models are trained end-to-end on video data, so we'll need datasets that mix video and language, e.g. transcripts of game streams. When the models are scaled up to cross-videogames and robots there will definitely be a bunch of language data.
nee1r•16m ago
how do you pick good goal conditioning images/do you have to hand pick a dataset of good goal images? seems hard if you don't have full context. really cool though!
agajews•3m ago
you don't need to pick good images manually! it's similar to next-token prediction, many prediction tasks aren't especially interesting, but there are enough hard ones that the model can spend the most time learning from those. the simplest thing scaled up works very well.
Onavo•11m ago
I disagree with the approach. It's a good approach for limited domain problems, but not for general purpose. Take something like this where you will need to be able to refer to wikis and research and ask questions on Reddit and Discord to optimize playthrough, none of the goal conditioning will be useful.

https://www.feed-the-beast.com/modpacks/125-ftb-evolution

I think a properly fine-tuned VLA with access to tool calls can scale way better.

agajews•55s ago
The goal conditioning is just a training objective! You could combine it with tool use, web searches, etc. and still train end-to-end on goal conditioning. (One way of thinking about it is goal conditioning defines a good distribution of tasks, and lends itself well to pretraining with an unlabelled video dataset.)

Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

https://www.neomindlabs.com/2026/06/08/running-gemma-4-26b-at-5-tokens-sec-on-a-13-year-old-xeon-...
106•neomindryan•2h ago•55 comments

Codex Micro

https://openai.com/supply/co-lab/work-louder/
73•davidbarker•2h ago•72 comments

Mysteries of Telegram Data Centers

https://dev.moe/en/3025
173•theanonymousone•4h ago•54 comments

Show HN: misa77 - a codec that decodes 2x faster than LZ4 (at better ratios)

https://github.com/welcome-to-the-sunny-side/misa77
61•nonadhocproblem•2h ago•17 comments

Open-source memory for coding agents, synced over SSH

https://github.com/vshulcz/deja-vu/
51•vshulcz•2h ago•13 comments

Collection of Digital Clock Designs

https://clocks.dev
39•levmiseri•1h ago•10 comments

Launch HN: Coasty (YC S26) – An API for computer-use agents

https://coasty.ai/docs
19•nkov47•2h ago•0 comments

Jurassic Park computers in excruciating detail

https://fabiensanglard.net/jurrasic_park_computers/index.html
821•vinhnx•15h ago•212 comments

Artie (YC S23) Is Hiring Software Engineers

https://jobs.ashbyhq.com/artie
1•tang8330•1h ago

Stripe, Advent offer to buy PayPal for more than $53B

https://www.reuters.com/business/finance/stripe-advent-offer-buy-paypal-more-than-53-billion-sour...
49•rvz•14h ago•16 comments

Sleep regularity is a stronger predictor of mortality risk than sleep duration (2023)

https://academic.oup.com/sleep/article/47/1/zsad253/7280269
563•bilsbie•6h ago•276 comments

Towards a Harness That Can Do Anything

https://eardatasci.github.io/c/ambiance/index.html
98•evakhoury•4h ago•56 comments

My midlife crisis Corolla is fast, furious, and modded

https://www.zocalopublicsquare.org/my-midlife-crisis-corolla-fast-furious-fully-modded/
70•gmays•3h ago•141 comments

Prioritize mental health, and why communication is so important

https://ramones.dev/posts/mental-health/
204•ramon156•6h ago•127 comments

When A.I. Is a Member of the Family

https://www.newyorker.com/magazine/2026/07/20/when-ai-is-a-member-of-the-family
34•fortran77•2h ago•32 comments

The Memory Heist

https://www.ayush.digital/blog/the-memory-heist
48•eieio•22h ago•3 comments

A General Goal-Conditioned Minecraft Model

https://pantograph.com/journal/pan-1
17•agajews•1h ago•9 comments

OpenAI loses trademark dispute at EU court

https://dpa-international.com/economics/urn:newsml:dpa.com:20090101:260715-930-389143/
130•hermanzegerman•3h ago•104 comments

The Three-Second Theft: Why AI Voice Fraud Outruns Every Defence

https://smarterarticles.co.uk/the-three-second-theft-why-ai-voice-fraud-outruns-every-defence
125•dxs•5h ago•160 comments

The well-calibrated Bayesian [pdf] (1982)

https://fitelson.org/seminar/dawid.pdf
39•Murfalo•4h ago•12 comments

Unsolved Problems in MLOps

https://spawn-queue.acm.org/doi/pdf/10.1145/3762989
16•gnyeki•2h ago•3 comments

Today I Rescued 7,234 Old GIFs

https://danq.me/2026/07/10/rescuing-7234-gifs/
28•birdculture•2d ago•1 comments

The Conservationist Who Turned 40 Terabytes of Public Data into a Video Game

https://blog.exe.dev/meet-the-conservationist-who-turned-40-terabytes-of-government-data-into-a-v...
63•bryanmikaelian•1d ago•10 comments

Weathergotchi – an open-source climate Tamagotchi

https://github.com/Michael-Manning/E-Paper-Climate-Logger
94•luanmuniz•7h ago•22 comments

What designing 54 computer science cards taught me about graphic design

https://fhoehl.com/designing-algodeck
14•marukodo•1h ago•4 comments

FreeBSD 16 Retires the Last of Its GPL Code from Its Base System

https://www.phoronix.com/news/FreeBSD-16-Goes-GPL-Free
40•lr0•1h ago•0 comments

What Every Python Developer Should Know About the CPython ABI

https://labs.quansight.org/blog/python-abi-abi3t
33•matt_d•3d ago•4 comments

DEA to Temporarily Schedule 7-Oh and Related Substances to Protect Public Safety

https://www.dea.gov/press-releases/2026/07/01/dea-temporarily-schedule-7-oh-and-related-substance...
63•gnabgib•3h ago•124 comments

SpaceX bond worth 10% less than issue price – heading for junk bond status

https://www.ft.com/content/3a023b95-66c3-41e1-b0ce-df752a499541
472•youngtaff•5h ago•412 comments

What's the most popular number in Hacker News titles?

https://blog.omgmog.net/post/most-popular-numbers-in-hn-post-titles/
40•omgmog•5h ago•21 comments