Rendered at 14:05:22 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
stusmall 18 hours ago [-]
This is so interesting to watch. For a couple minutes I was in awe of how quick and cheap it was. Then I saw just how bad the decision are and how it would get stuck in strange loops of going in and out of the same door to no end.
This seems like a technology heading in the right direction but not quiet there yet. Excited for what they are cooking up but probably won't start building around it yet.
binlog 17 hours ago [-]
This entire conversation around Jev seems weird to me. Like... we started from neural nets that could do basic decision making and classifications pretty well, then trained larger and larger language models to get to where we are now. Now suddenly everyone is going crazy because someone trained a smaller model that is adequate at making decisions? We already went through the "look this AI can play pokemon terribly" phase like a decade ago.
c7b 17 hours ago [-]
A pre-trained universal classifier that can replace specifically-trained ones would have been considered just as much science fiction in the 2010's as the capabilities of modern LLMs. I'm not sure Jev is actually there yet, but at least it sounds theoretically doable today.
That being said, one thing having been unrealistic 10 years ago and just about possible today doesn't mean that it's going to change the world the same way another technically related, previously-impossible thing did. The Jev hype gives me a bit of the "you're still early to crypto" vibes of some later altcoins. I really like the idea, I think it's going to open up possibilities for using classifiers where we wouldn't or couldn't have trained one before. I'm crossing my fingers for an open weights version to drop. But it's still just a classifier, people have built similar things before Jev, the one thing that really stands out about it is their ability to generate hype.
someothherguyy 16 hours ago [-]
> but at least it sounds theoretically doable today
why
thornewolf 15 hours ago [-]
We have a bad universal classifier now (via Jev). 0->1, one might say.
A bad universal classifier does suggest a good one later. And that is exactly what I would call "theoretically doable"
That said, I don't think that Jev is a magic breakthrough or anything. I think it is just a particularly good narrative with an easy way to try it out.
lukev 13 hours ago [-]
Jev is interesting in that it's much cheaper and faster than a frontier LLM.
But I've seen nothing to indicate that the upper bound on classification tasks of a Jev-like model can exceed a frontier LLM with reasoning tokens. That seems nearly impossible even in principle (since Jev-style models are still based on LLM pretraining).
So while they're definitely on the Pareto frontier, which is valuable, they're at the "cheap" end of the spectrum more than the "good" end and I don't expect that to change.
15 hours ago [-]
c7b 14 hours ago [-]
LLMs are like lossy compression of ~all of written text ever produced, with useful recall. To the extent that the corpus contains labelled examples of the given classification task, it's not unreasonable to think that we'll be able to build a decoder for that, just like we already have a useful decoder for next-token prediction. Extend to image classification the same way we already have multimodal LLMs.
Garlef 5 hours ago [-]
The difference is that you don't need training here;
The decision graphs can be built on the fly by an LLM and contain instructions in plain language.
The magic moment for me from the Jev release was not that there was some system playing doom: Rather it was the moment, they just changed a part of the prompt to "don't shoot, just dodge" and the behavior changed immediately.
This means you can have a system with fast decision-making but still interact with it via language.
solidasparagus 15 hours ago [-]
The cheap, fast and smart-enough LLM space has been wildly neglected. Jev is one of the few players truly targeting that space. And for a lot of people it is the first time they are asking "what could I build if llms were interaction-speed fast?". The answers are cool, the problem is that Jev is not, I think, smart-enough yet to have that many applications, but it's smart enough that you can start to see what they will look like.
osener 16 hours ago [-]
It is impressive, but all the hype and fake demos are selling it as a model that is as smart as frontier reasoning LLMs in the decisions it makes yet much cheaper and much faster, which is not true.
pancomplex 16 hours ago [-]
Nothing fake here and fully open source if you wanna take a peek. It does make a bunch of mistakes, often. But it eventually recovers!
For those curious how it works, it’s essentially a script that plays the game but uses jev as a source of rng to make it stochastic
Jach 8 hours ago [-]
Yeah, there's a lot hard-coded into the typescript files that make this much less impressive than the other "AI plays" versions that have come and gone that play with less help. A Claude one that only used screenshots would always get stuck in the rocket hideout...
osener 8 hours ago [-]
For the record I did not mean to say that about your work, nice job.
gchamonlive 15 hours ago [-]
I think that misses the point of Jev being ridiculously efficient while maintaining adequate intelligence for automation tasks. We have to train our minds to filter out branding and marketing.
9 hours ago [-]
mtford 6 hours ago [-]
This happens all the time in tech. A few years ago everybody got excited about static websites and server-side rendering as if we hadn't been doing that with PHP long ago.
gchamonlive 15 hours ago [-]
For what it does, it classifies, orchestrates, operates and delegates tasks exceedingly well for its size and weight. It's ridiculously cheap and efficient, but if you can only see progress in terms of raw cognitive power then you'll surely miss how interesting this is.
michaelchisari 12 hours ago [-]
Performance and efficiency have been neglected as everyone threw every available GPU and trillions of dollars trying (and failing) to create AGI. Though the results have been impressive.
But good enough for pennies in an instant is very useful.
aurareturn 9 hours ago [-]
Yea but this model wasn’t trained to play Pokémon but it can. It’s general.
ford 17 hours ago [-]
I agree it's overhyped, but the transition to a general purpose classifier (vs a narrow scope classifier) is new and noteworthy.
Ie the famous "Hotdog" clip from Silicon Valley [0]
Making decisions quickly, cheaply and without having to train your own model.
joshuat 17 hours ago [-]
Math.random can make poor decisions quickly and cheaply
azan_ 17 hours ago [-]
Benchmark it against jev and you'll have your answer.
zahlman 16 hours ago [-]
I mean,
> get stuck in strange loops of going in and out of the same door to no end
Math.random is statistically unlikely to do this.
joshuat 13 hours ago [-]
mathrandomplayspokemon.org
irregularbowels 10 hours ago [-]
[flagged]
jbjbjbjb 15 hours ago [-]
Jev is for single shot classification, not multi-step RL environments with delayed reward and explore/exploit. My guess is it would go through the door with high confidence every time unless you change the input to add the history.
Garlef 6 hours ago [-]
I think the people behind jev made their intended use clear by labeling as "system one" - so they anticipate that there's a slower "system two" mediating jev
pancomplex 15 hours ago [-]
This runs entirely on Jev as the only AI with a typescript harness that feeds it selective context.
mitxela 3 hours ago [-]
feels like pre-LLM AI playing a video game but more expensive
ralusek 17 hours ago [-]
The exact message I sent my friend this morning:
> the most interesting thing about this jev stuff
> is that people are seemingly like
> completely disinterested in how smart it actually is
> I haven't even heard it mentioned a single time how it actually compares to other LLMs coming up with their own classifications. Just: it's fast and cheap
After watching a few minutes of this it makes me think that maybe we should be a little more interested in how smart it is.
johnsmith1840 14 hours ago [-]
They know it's not. The company actually has or had public statements that they didn't like public benchmarks for comparison.
My main wonder is the difference between it and having a small llm no thinking output a single number only as a choice. Isn't that nearly the same here?
stusmall 15 hours ago [-]
There is a lot of room for a lot of different models. For many use cases, intelligence beats out all.
For me in my day job, having extremely fast low quality decision makers over noisy inputs is very valuable. I work in security and having something that can help triage alerts, classify items and group things together is extremely valuable. It doesn't need to be perfect. Just being able to take a set of inputs from deterministic tooling and to be make general priority classifications goes a long way on helping humans look at the most important items first.
refactor_master 10 hours ago [-]
Why not spend a few tokens or so on making a frontier LLM build the classifier from scratch? Then you also know exactly what the classifier is capable of, and whether it’s easily learnable/informative data.
Why would you be confident in feeding garbage to a “cheap and fast” classifier with unknown domain-specific performance?
I know we kind assume omniscience for frontier models, but at this point the evidence is kind of out there.
raincole 13 hours ago [-]
Because it's not that smart especially when you compare it to other LLMs. The top LLMs have completely change the baseline of being smart.
pancomplex 17 hours ago [-]
Like others have mentioned in this post, I think a mix of models like Jev for simple stuff + a smarter reasoning model for more strategic thinking is the optimal solution. This experiment however is purely Jev. Which sometimes can be kinda dumb.
stusmall 39 minutes ago [-]
Oh I want to be clear, I don't want my post to come off as negative. I really do mean things are headed the right direction and I think this harness is super cool. So don't take it that way and thank you for bringing something cool into this world.
JamesSwift 11 hours ago [-]
Well in a lot of ways this demo actually is a mix of jev for simple stuff + a very intelligent harness to fill in the rest. Pay attention to the "jev counter" on when api calls actually happen and the kinds of decisions its making. Its very rarely in a tight loop, and when it is (eg in an item menu) it tends to randomly walk through options.
Its very cool, but the harness is doing a _ton_ of heavy lifting.
brenschluss 2 hours ago [-]
The seminal (lol) Twitch Plays Pokémon was twelve years ago, so just posting this amazing moment of internet history/lore just in case folks don’t know or have forgotten: https://en.wikipedia.org/wiki/Twitch_Plays_Pok%C3%A9mon
MitPitt 18 hours ago [-]
This is kinda chill to have in the background. I wish there were livestreams showing live reasoning of top models which are currently trying to solve cancer or whatever. Imagine the pogs in chat when it does.
pancomplex 18 hours ago [-]
People need to be live streaming their AI more!
tehnoslow 18 hours ago [-]
Actually, yes, that would be at least interesting
someothherguyy 16 hours ago [-]
then it would cost human lives, no more fun i made this thing with no real effort vibes
ac2u 15 hours ago [-]
Cool project, comes with a little too much guidance in the harness though IMO (pathfinding, textual milestones etc). (The author is very upfront about this in their README though)
I think if it was combined with a regular vLLM it could be really interesting, especially watching the reasoning logs.
Bonus points if it was one of the latest open models that somehow had all prior training knowledge of Pokemon abliterated so it was reasoning as an intelligent persona that had no knowledge of even the concept of Pokemon.
budoso 7 hours ago [-]
I also tried to create a “Jev plays Pokémon” but with minimal additional context outside a move history and what is available in memory from the emulator, so no pathfinder, predetermined game path, etc. I can safely say that this experiment failed however, and Jev was not able to even get to Professor Oak’s lab to get a starter.
Props to OP for getting a working version, but it does not seem that this model is capable enough to play Pokémon at this point in time.
3 hours ago [-]
jackbrookes 2 hours ago [-]
I've been watching GPT_Plays_Games on twitch and with Astra now it can beat pretty much all of them with very few mistakes, with vision only. One downside is that when pathfinding it can only plan high level paths, wait for result, then try again if it fails. Its a ~10-30 second loop.
Has anyone tried a combined LLM + Jev? So the LLM directs the high level goal ("reach the door of the pokemon center while avoiding NPCs") then allow it to instruct Jev to do the actual movements? That seems like a good balance between the high-level, slow planning of the LLM to the fast but limited Jev. That kind of mirrors how humans work too. The harness could even allow Jev to delegate back to the LLM when it doesn't have a confident answer.
Its almost like when humans drive, we kind of let our subconscious take over once we know where to go. But if we see something unexpected in the road, we can go back to a conscious planning mode to decide what to do.
with such a fat harness, this is more like watching a walk thru play the game.
laszlokorte 17 hours ago [-]
Yeah I would have expected it to only decide which button to press, not something abstract like the choice of "go east to lavender town" for the goal of "in lavender town, climb the pokemon tower"
pancomplex 16 hours ago [-]
That was how I first implemented it. Jev sadly never left Pallet Town.
rickintoplace 23 hours ago [-]
That's actually fun to watch. Did you experiment with nicknaming before you turned it off? I'd be a little curious to see how it behaves.
pancomplex 23 hours ago [-]
Jev can't come up with original text, but I did consider giving it a list of hilarious names.
JamesSwift 18 hours ago [-]
Sure it can, just ask it for next char or "done" in a loop
pancomplex 17 hours ago [-]
I just took your advice and added it to the list of Jev decisions. Watch it name its next Pokemon!
pancomplex 18 hours ago [-]
True!
IanCal 18 hours ago [-]
Maybe letter by letter spelling?
18 hours ago [-]
singularity2001 4 hours ago [-]
Jev is dumb a rock, so it's basically just a random selection of a smart model's suggested plans?
dang 14 hours ago [-]
Dare we have two Pokemon-playing-AI threads at the same time?
Looking at the diagram in the gh repo, it looks like this is entirely jev. Are there any examples of people having a big model like Fable handle high level goals?
Hm wondering what a first pass optimal setup might be - jev for overworld navigation, escalate to sonnet for easy battles, opus for medium difficulty battles, and fable for gym bosses could probably have jev also manage all the escalation / de-escalation to different models.
stusmall 18 hours ago [-]
>but not fast enough to play Doom yet sadly
Did I miss something? I thought one of the demo videos was it doing pretty decent at the first level of Doom?
pancomplex 18 hours ago [-]
In my experience it was too slow to do an FPS with 30 ticks per second reliably. It gets killed too fast.
13 hours ago [-]
djhworld 17 hours ago [-]
What's not clear to me on the video is whether jev is doing the button presses for controlling the character to move around.
The "Jev calls" counter only seems to increment at junction points like battles, conversation prompts, menus etc.
Is something else moving the character around?
pancomplex 17 hours ago [-]
It's connected to the ROM of the actual game, so it can see a lot of things. It has multiple tools available, including being able to move to coordinates.
Considering it just made Charizard forget its only fire-type move "Ember" to learn "Counter", I note no signs of intelligence.
pancomplex 18 hours ago [-]
Rookie mistake clearly..
DylanMerigaud 7 hours ago [-]
Struggles with loops show complexity isn't fully mastered yet.
flockonus 14 hours ago [-]
The decisions are quick, but look good as random w/ tons of back and forth.
Are you at least feeding back some of its previous decisions on to state?
hummusFiend 16 hours ago [-]
Great work!
Super cool to see it do the whole game. I spent my fable budget building something similar this week but only drove it to Brock. I like the "current focus" framing too.
pancomplex 15 hours ago [-]
Thank you! And interesting data point on Fable burning out on tokens so early.
18 hours ago [-]
dochaus 13 hours ago [-]
He confirms what I long suspected, all you need is to cheese Charizard and flamethrower
361994752 18 hours ago [-]
watched it stuck at rocket hideout for 10 mins.... let me check 1hr later to see if it can find a way out
jumploops 18 hours ago [-]
It’s currently stuck at an elevator and deciding to teach Pokemon various TMs and HMs instead of progressing… pretty hilarious!
pancomplex 17 hours ago [-]
it made it through!
rubicon33 12 hours ago [-]
Pretty awesome! Just curious are you vibe coding this?
theturtletalks 18 hours ago [-]
Is Frigade going to use Jev to do object detection?
pancomplex 18 hours ago [-]
Working on it :)
theturtletalks 18 hours ago [-]
Nice. When I saw your connection to Frigade, I knew there was a connection haha
zaik 17 hours ago [-]
Seems like it got stuck on a Ghost enemy.
avaer 17 hours ago [-]
I wish jev took in images so we could do this generically for any game, without memhacks. I'm sure that's coming.
You could front this with an image -> text model but that would be much lower quality vs latency, and the whole point of doing it with a decision model is remove the latency.
Games are a really interesting testing ground for robotics; if we can solve game playing (incl 3d) we could embody "system one" intelligence into robots that have something emulating general reflexes without needing to fine tune.
pancomplex 16 hours ago [-]
Agreed this would be super cool and I do see that coming in the future. But Jev-level latency just isn't there yet with full images.
fzysingularity 13 hours ago [-]
We just did exactly this - added TypeSafe-compatible API (incl. websocket support) for various VLMs. Latency right now is <250ms, but will be able to get it to <150ms (p95).
This seems like a technology heading in the right direction but not quiet there yet. Excited for what they are cooking up but probably won't start building around it yet.
That being said, one thing having been unrealistic 10 years ago and just about possible today doesn't mean that it's going to change the world the same way another technically related, previously-impossible thing did. The Jev hype gives me a bit of the "you're still early to crypto" vibes of some later altcoins. I really like the idea, I think it's going to open up possibilities for using classifiers where we wouldn't or couldn't have trained one before. I'm crossing my fingers for an open weights version to drop. But it's still just a classifier, people have built similar things before Jev, the one thing that really stands out about it is their ability to generate hype.
why
A bad universal classifier does suggest a good one later. And that is exactly what I would call "theoretically doable"
That said, I don't think that Jev is a magic breakthrough or anything. I think it is just a particularly good narrative with an easy way to try it out.
But I've seen nothing to indicate that the upper bound on classification tasks of a Jev-like model can exceed a frontier LLM with reasoning tokens. That seems nearly impossible even in principle (since Jev-style models are still based on LLM pretraining).
So while they're definitely on the Pareto frontier, which is valuable, they're at the "cheap" end of the spectrum more than the "good" end and I don't expect that to change.
The magic moment for me from the Jev release was not that there was some system playing doom: Rather it was the moment, they just changed a part of the prompt to "don't shoot, just dodge" and the behavior changed immediately.
This means you can have a system with fast decision-making but still interact with it via language.
https://github.com/christianmat/jev-pokemon
But good enough for pennies in an instant is very useful.
Ie the famous "Hotdog" clip from Silicon Valley [0]
https://www.youtube.com/watch?v=ACmydtFDTGs
Example (2022):
https://developers.openai.com/cookbook/examples/zero-shot_cl...
> get stuck in strange loops of going in and out of the same door to no end
Math.random is statistically unlikely to do this.
> the most interesting thing about this jev stuff
> is that people are seemingly like
> completely disinterested in how smart it actually is
> I haven't even heard it mentioned a single time how it actually compares to other LLMs coming up with their own classifications. Just: it's fast and cheap
After watching a few minutes of this it makes me think that maybe we should be a little more interested in how smart it is.
My main wonder is the difference between it and having a small llm no thinking output a single number only as a choice. Isn't that nearly the same here?
For me in my day job, having extremely fast low quality decision makers over noisy inputs is very valuable. I work in security and having something that can help triage alerts, classify items and group things together is extremely valuable. It doesn't need to be perfect. Just being able to take a set of inputs from deterministic tooling and to be make general priority classifications goes a long way on helping humans look at the most important items first.
Why would you be confident in feeding garbage to a “cheap and fast” classifier with unknown domain-specific performance?
I know we kind assume omniscience for frontier models, but at this point the evidence is kind of out there.
Its very cool, but the harness is doing a _ton_ of heavy lifting.
I think if it was combined with a regular vLLM it could be really interesting, especially watching the reasoning logs.
Bonus points if it was one of the latest open models that somehow had all prior training knowledge of Pokemon abliterated so it was reasoning as an intelligent persona that had no knowledge of even the concept of Pokemon.
Props to OP for getting a working version, but it does not seem that this model is capable enough to play Pokémon at this point in time.
Has anyone tried a combined LLM + Jev? So the LLM directs the high level goal ("reach the door of the pokemon center while avoiding NPCs") then allow it to instruct Jev to do the actual movements? That seems like a good balance between the high-level, slow planning of the LLM to the fast but limited Jev. That kind of mirrors how humans work too. The harness could even allow Jev to delegate back to the LLM when it doesn't have a confident answer.
Its almost like when humans drive, we kind of let our subconscious take over once we know where to go. But if we see something unexpected in the road, we can go back to a conscious planning mode to decide what to do.
1. https://www.twitch.tv/gpt_plays_games
Teaching a World Model to Play Pokemon - https://news.ycombinator.com/item?id=49849907
The interesting thing here imo is the cost and latency. So far we're at 4 badges for less than $0.5
However, https://x.com/TynanSylvester/status/2096965749369720970 Astra was able to beat RimWorld. So LLMs are definitely able to drive these sorts of games to completion with their current abilities.
Did I miss something? I thought one of the demo videos was it doing pretty decent at the first level of Doom?
The "Jev calls" counter only seems to increment at junction points like battles, conversation prompts, menus etc.
Is something else moving the character around?
All in the OSS repo if you wanna play around with it: https://github.com/christianmat/jev-pokemon
Super cool to see it do the whole game. I spent my fable budget building something similar this week but only drove it to Brock. I like the "current focus" framing too.
You could front this with an image -> text model but that would be much lower quality vs latency, and the whole point of doing it with a decision model is remove the latency.
Games are a really interesting testing ground for robotics; if we can solve game playing (incl 3d) we could embody "system one" intelligence into robots that have something emulating general reflexes without needing to fine tune.
Take a look at a snake demo with streaming image inputs: https://x.com/spillai/status/2103630735425089957