In browser PPO training demo, made possible by tinygrad: TinyJit -> WebGPU kernels.

Requires WebGPU.


• Mariajaved906 an hour ago

This is seriously impressive. Running PPO training directly in the browser through WebGPU feels like a glimpse into where lightweight AI experimentation is headed.

• ldoughty a day ago

Really cool! But right as it was nearing 4,000, it seems to have corrupted itself and no longer got any scores above 0. Not sure if that's a code bug or a neural net issue.

avg500 -4.6 last 500 episodes

peak 3959.3 best window

roll/s 20.68 20-step avg

progress 4388 562749 episodes

• c1b a day ago

Yes it just collapses eventually — never stabilizes. The training process is flawed, I suspect it has to do with the fact that some weights blow up over time, you can see in “weights” tab.

But at around 4K avg score you should see it solve the env almost every time.

Just a demo :) optimized for speed over stability.

Reward structure: Step: -1 Dot: +100 Win: +1000 so ~4k is max theoretical score on 6x6.

• ticulatedspline a day ago

maybe because it doesn't understand "done"? perfect play is impossible, random variance will cause scores to drop even if the model plays well and "wins". feels like it would get stuck in a loop trying to improve what can't be improved.

• fc417fc802 a day ago

The optimizer doesn't need to understand anything it's just an iterated mathematical construct. The author simply didn't bother to implement the necessary details to ensure numerical stability.

Alternatively it might be a problem with the scoring model in the end game.

• jesuo 21 hours ago

feels like it would get stuck in a loop trying to improve what can't be improved.

That is the point, there is nothing on an intention that we cannot improve, the goal here is no more than 1 unique iteration of the same path

• r3trohack3r a day ago

I think I noticed it reach “end game.” The snake reaches a point where, if it gets any longer, it is out of squares and hits its own tail. So it finds the route through the squares that it can infinitely loop, never eats the ball, and score starts dropping and goes negative.

• simedw a day ago

Cool project!

I noticed that if you go from training to watch and then back, the training temporarily drop significantly in score.

• bguberfain a day ago

It seems to be something related the moving average calculation. So it is just a glitch on the chart.

• joshka a day ago

A previous similar idea running as a Ratatui based TUI: https://github.com/bones-ai/rust-snake-ai-ratatui

• starshadowx2 19 hours ago

FYI this website sets off a bunch of Bitdefender alerts as being a suspicious web page. I assume probably false positives or something but still something you might want to look into.

"The page https://ppo.gradexp.xyz/ has been detected with suspicious activity. It is not recommended to continue browsing this website."

Same for:

https://ppo.gradexp.xyz/version.js

https://ppo.gradexp.xyz/dist/sizes.js

https://ppo.gradexp.xyz/dist/size_6/manifest.j

https://ppo.gradexp.xyz/dist/size_6/weights.safetens

https://ppo.gradexp.xyz/dist/sokol/demo.wa

• jesuo 13 hours ago
• lucrbvi 13 hours ago

it's using webgpu kernels, probably a false positive

• anthonycoslett 19 hours ago

Mesmerizing - could be its own digital art showcase XD Love what you've done here, friend. Looking forward to what you do next. <3

• snats a day ago

did a pretty similar thing last month for the text rendering library last month.

trained and made a viz for the model and then made it displace text.

should probably do a proper write-up:https://x.com/i/status/2038367016969724259

• ziofill a day ago

I noticed snake gets penalized for not getting to the apple early, is that what you really want? Snake is about how long it gets not about the balance between length and wall clock time

• foo12bar a day ago

But if not the snake could go into an infinite loop, never growing, never eating.

• ziofill 14 hours ago

Why? It should get the reward for getting longer, but not for getting longer quicker

• foo12bar 9 hours ago

Because the sessions would last forever. Think of a 1 or 2 length snake, figuring out that left down up right over and over again doesn't lose any points. You're now trapped in a local minimum. You need to make the AI get impatient (lose points) or it'll never learn.

• ziofill 5 hours ago

I see what you are saying but then wouldn’t it miss out on the best strategies, which do require patience and not going straight for the apple?

• foo12bar an hour ago

Maybe you could make it lose points for repeating a board state, I guess.

• jesuo a day ago

Poorly programmed, it doesn't learn from its mistakes, the games get stuck in a loop because the snake doesn't capture a piece but the piece remains and there's a gap, constantly moving the snake along the same path with negative scores in an infinite loop leaving an unaltered yin and yang ;) there's a repetitive pattern in these infinite games between the position of the gap and the piece

• spectre9 a day ago

Did you let it train? This doesn’t happen for me

• jesuo a day ago

Yes, thousands of games, you can see how it happens in the displayed game matrix, there comes a point when they all enter those loops https://ibb.co/bM4RPzPb

• spectre9 a day ago

Makes sense, author mentioned training collapses eventually

• jesuo 12 hours ago

Give the neural network the sense of sight, to know where the point is located.

• beardsciences a day ago

My average eventually made it to about 3900, and then stagnated between 3600-3900. I'm curious if this is universal behavior or not. I'm up to about 5k steps.

• dennisy 9 hours ago

Very cool! Not GitHub repo?

• neduma a day ago

More details and implementation notes please?

• cshimmin a day ago

It's on the page, if you click the little info icon in the upper-right. Here's the text but there's some nice graphics there too:

  Snake Game, training entirely in the browser. Built on tinygrad: the rollout / targets / train graphs are TinyJits authored in Python, then compiled once to WGSL and replayed here under WebGPU.

  Observation: flat 10×10 board (100) + 4-dim prev-action one-hot = 104 dims. fc_pi.weight is zero-init so the opening policy is uniform over the legal actions; fc_v uses tinygrad's default Kaiming init.

  Per rollout: T=24 × N=384 parallel snakes (9,216 transitions), then K=3 epochs × 4 mini-batches of PPO updates. GAE γ=0.99, λ=0.95; AdamW wd=0.01; ratio clip ε=0.1; grad-norm 0.5; Huber value β=1, val_coef=1; entropy bonus 0.008333333333333333.

  Action mask + value clip + KL early stop. The 4-dim prev_a obs tail lets fc_pi zero the U-turn logit (the env silently overrides same-axis reversals anyway). Value loss is max(huber(v_new−td), huber(v_clip−td)) at ε=0.2. Approx-KL is sampled after each epoch and breaks the loop at 1.5·kl_target.
• ionwake 10 hours ago

damn this was really interesting and really well executed

• mavdol04 a day ago

That's cool, i did exactly the same few years ago

• modinfo 16 hours ago

Will be open-sourced?

• LowLevelKernel a day ago

Link to repo?

• th1nhng0 a day ago

cool project

• bozhark a day ago

Crashed

• jmclnx a day ago

> WebGPU not available in this browser

Looks like this is for Linux and Windows, on NetBSD I get this issue :(

• NoboruWataya a day ago

I got this in Firefox on Linux, just had to enable WebGPU in about:config (`dom.webgpu.enabled` = true).

• jmclnx a day ago

Did not know that existed, I enabled it but no luck. Must be a NetBSD thing based upon this new message:

> WebGPU is not yet available in Release or late Beta builds.

• redshiftza a day ago

If you are using brave (which i assume also applies to chrome) , there is a menu at brave://flags , you can enable unsafe web GPU from there

• insane_dreamer 20 hours ago

sound cool; would like to show my kid for education; doesn't work on Mac/Safari though (no webGPU)

• c1b 16 hours ago

You can enable it in settings; works on my older iPhone.

• austinthetaco a day ago

my training on a 10x10 just randomly broke. i got to like 3600 then the graph went flat, the viewer on the left just showed it constantly restarting the game, and the scores in the negative. my average is now -10.