Hacker News: drubs

New comment by drubs in "Beam: Reflection's 501B open-weight model"

drubs — Mon, 05 Oct 2026 21:14:14 +0000

Outside of ML metrics, you're monitoring the health of every piece of hardware in the system. You need to make sure that you have every GPU, every CPU, the PCIe buses, the networking fabric are all working without any errors. You need to ensure that you can respond as fast as possible to any possible error. One bad component can bottleneck the entire job.

New comment by drubs in "Beam: Reflection's 501B open-weight model"

drubs — Mon, 05 Oct 2026 19:41:25 +0000

I remember being in the room with pretraining day 1 to help monitor the training job launch. Watching this model train from day 1 has been an amazing experience!

New comment by drubs in "Less is more: When agents learn not because but despite my doings"

drubs — Mon, 26 Jan 2026 20:08:52 +0000

Fantastic work! This was a really fun collaboration.

New comment by drubs in "Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris"

drubs — Wed, 31 Dec 2025 17:36:17 +0000

Star the puffer https://github.com/PufferAI/PufferLib

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Sat, 08 Mar 2025 21:45:43 +0000

Wouldn't make much sense. We generally train with 288 environments simultaneously. I've been thinking about ways to nicely stream all 288 environments though.

New comment by drubs in "Reflection – AlphaGo / Gemini team building superintelligent coding agents"

drubs — Fri, 07 Mar 2025 17:21:27 +0000

Really excited to be a part of the team!

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Thu, 06 Mar 2025 14:52:42 +0000

Sounds cool to me.

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Thu, 06 Mar 2025 05:46:26 +0000

Yup!

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Thu, 06 Mar 2025 00:36:33 +0000

It's silly, but signs were a way to incentivize the agent to explore deeper into the Safari Zone among other areas.

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Wed, 05 Mar 2025 23:32:27 +0000

My first version of this project 5 years ago involved a python-lua named pipe using Bizhawk actually. No clue where that code went

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Wed, 05 Mar 2025 22:24:52 +0000

There's a ton of applications for AI. Back when I was at Spotify, I co-authored Basic Pitch (https://basicpitch.spotify.com/), an audio-to-midi library. There are a ton of uses for AI outside of what's heavily publicized.

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Wed, 05 Mar 2025 19:32:32 +0000

There's an entire section on how the decompilations were used :)

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Wed, 05 Mar 2025 19:31:16 +0000

Wrote about this in the results section. I think there is a way to mix the two and simplify the rewards in the process. A lot of the magic behind getting the agent to teach and use cut probably could have been handled by an LLM.

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Wed, 05 Mar 2025 19:04:04 +0000

The environments wouldn't concentrate enough in the Rocket Hideout beneath Celadon Game Corner. The agent would have the player wander the world reward hacking. With wild battles enabled, the environments would end up in Lavender Tower fighting Gastly.

> (and how on earth did you port Pokémon red to a RL environment? O.o)

Read and find out :)

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Wed, 05 Mar 2025 19:01:08 +0000

and...fixed!

New comment by drubs in "Show HN: Beating Pokemon Red with RL and <10M Parameters"

drubs — Wed, 05 Mar 2025 18:42:02 +0000

Thanks for the heads up. I just pushed a fix.

Show HN: Beating Pokemon Red with RL and <10M Parameters

drubs — Wed, 05 Mar 2025 17:07:09 +0000

Hi everyone!

After spending hundreds of hours, we're excited to finally share our progress in developing a reinforcement learning system to beat Pokémon Red. Our system successfully completes the game using a policy under 10M parameters, PPO, and a few novel techniques. With the release of Claude Plays Pokémon, now feels like the perfect time to showcase our work.

We'd love to get feedback!

Comments URL: https://news.ycombinator.com/item?id=43269330

Points: 183

# Comments: 68