Training a coding model to paint watercolours with TRL and OpenEnv

Hugging Face engineers reproduced a viral AI art project in which a language model trained with reinforcement learning writes JavaScript to paint watercolours, publishing all artifacts openly. The project, led by Surya Narreddi's original idea, uses TRL and OpenEnv to train a Qwen model to generate p5.brush code, with the full pipeline running on Hugging Face and the training recipe available as a single command.

Reinforcement Learning • Updated Training a coding model to paint watercolours with TRL and OpenEnv Update on GitHub https://github.com/huggingface/blog/blob/main/train-to-paint-with-code.md On 23 August, Surya Narreddi https://x.com/kickingkeys/status/2091570990048276897 posted a beautiful video of watercolours painted by a language model. The model writes JavaScript through p5.brush https://github.com/acamposuribe/p5.brush , a library that "adds natural drawing tools to p5.js". The video went viral fast, over 1.5M views at the time of writing. The video came with a blog post https://surya.website/rling-qwen-to-paint-with-code explaining the training behind an earlier and narrower stage of the project, close-up flowers rather than the full compositions in the video, sadly without open artifacts yet. His site says a full technical report is coming, so ensure you follow him. The original idea is his, coming from the art and design side, where his skills are way beyond mine https://x.com/kickingkeys/status/2094901433149612118 . My attempt is on the engineering side, reproducing the recipe in the open with every piece published. Note:for the context behind the project, told by Surya himself, watch this video of his thesis . In this article I try and reproduce his idea with TRL https://huggingface.co/docs/trl and OpenEnv https://github.com/huggingface/OpenEnv . The reference pool dataset, the RL environment, the training scripts and the trained models, all open. The whole pipeline runs on Hugging Face, end to end: - training on Jobs https://huggingface.co/docs/huggingface hub/guides/jobs - the RL environment and the scorer model as Spaces https://huggingface.co/docs/hub/spaces - the pairwise judge through Inference Providers https://huggingface.co/docs/inference-providers - and every artifact on the Hub, gathered in one collection https://huggingface.co/collections/HuggingEnvs/paint-with-code-6a955b79d63f67f1631d9be6 Once the two Spaces are up, the recipe is one command. Duplicate the environment https://huggingface.co/spaces/HuggingEnvs/watercolour-env and the scorer model https://huggingface.co/spaces/HuggingEnvs/watercolour-hpsv3 , set two environment variables for the reward mix, and launch: hf jobs uv run train/watercolour grpo.py --flavor h200 --timeout 48h --secrets HF TOKEN -- \ --env-url https://<you -watercolour-env.hf.space \ --model Qwen/Qwen3.5-35B-A3B --lora --all-linear --bf16 --gradient-checkpointing \ --subject 'a peach hibiscus' --references 4 \ --top-p 0.95 --top-k 20 \ --lr 5e-5 --lr-scheduler constant with warmup --warmup-steps 5 \ --scale-rewards none \ --steps 110 --n-episodes 240 --num-generations 8 \ --per-device-batch-size 1 --gradient-accumulation-steps 8 \ --max-completion-length 8192 \ --run-tag my-run --out <you /watercolour-grpo --push-to-hub The rest of this article is the story of getting there, and every piece is in the repo https://github.com/adithya-s-k/HuggingEnvs/tree/main/02-watercolour . I have followed the original blog step by step, and only changed something when strictly needed. Every idea of my own went into a list instead of into the experiment, and that list became "What I would try next" at the end, next to the full list of published artifacts. If you have already read his post, the framing and the reward design will be familiar. The new material is the open implementation, the hand-rated pool, and three reward mixes trained and compared, and it starts at The RL environment you need to build the-rl-environment-you-need-to-build . Why people loved it The paintings look loose, imperfect, handmade, at a moment when image models produce perfect statistically average pictures. My guess is that this contrast is a big part of why the video went viral. It reminded me of the early days of generative AI art, when the point was to explore the medium. DeepDream https://research.google/blog/inceptionism-going-deeper-into-neural-networks/ 2015 was a debugging tool that people turned into art, works like Edmond de Belamy https://en.wikipedia.org/wiki/Edmond de Belamy 2018 came from artists probing what a GAN could do, and artists like Mario Klingemann https://quasimondo.com spent those years making dreamy portraits with neural networks https://artsandculture.google.com/asset/memories-of-passerby-i-mario-klingemann/aAHG7iV3aXme8g . This project feels closer to those early days. In his thesis, Surya describes the path that led here. He started by prompting text-to-image models, where the prompt is the only lever you can pull, and more detail buys more control only up to a point. Training the model itself goes further. The other half of the idea is the medium. The model writes a program of about 150 lines of JavaScript that paints the image. That model output is code. You can read it, edit it and run it again, and the decision behind each brushstroke is visible. And the style comes from a restriction where the model is only allowed ten of the library's methods . More on that below. In that same period, Anna Ridler https://annaridler.com/works/myriad-tulips photographed thousands of tulips, hand-labelled every one, exhibited the dataset itself as the artwork, and later trained a model on it. I found her work through the references AI agents brought back while building this project and loved it because this project does something very similar by curating a set of images by hand, and then training against them. RL over taste Most of the recent RL work on language models uses rewards you can verify. For example, math problems with a known answer, code that passes tests, or graders that are right or wrong and cheap to run. This project is closer to the older exception, RLHF, where the model learns a reward model from human preferences. Here the reward is aesthetic preference. There is no correct answer. The real question of the project is whether you can do RL over taste. The reward, as his blog defines it and as the RL environment I built implements it: | term | weight | what it measures | |---|---|---| gate | 0.05 | the sketch compiles, paints something, does not cheat | length | 0.05 | a soft push towards longer code snippets | | pairwise judge | 0.60 | style, compared against references drawn from a pool | | HPSv3 https://huggingface.co/MizzenAI/HPSv3 is an open 7B preference model. Give it an image and a text description, and it returns a score for how much a person would prefer that image. It was trained on a large set of human choices between pairs of images, so its score is an average of many people's taste. The pairwise judge is Qwen3-VL-30B-A3B-Instruct https://huggingface.co/Qwen/Qwen3-VL-30B-A3B-Instruct , a general vision model called through HF Inference Providers. The pairwise judge sees the candidate painting next to four references randomly selected from the pool, guided by a written description of what to weigh bleeds, translucent washes, soft edges , each comparison in both presentation orders, and its score is the share of comparisons the candidate wins. Its only standard is the pool, so its score is my taste, as encoded in those ratings. Those are the weights Narreddi converged on. The pool defines taste here. That moves the work from tuning hyperparameters to building the set that decides what is beautiful. I trained three runs with this reward. They differ only in how the weight splits between the two model judges: | run | pairwise judge | HPSv3 | role | |---|---|---|---| judge-led | 0.60 | 0.30 | the original mix, stopped at step 110 | hps-led | 0.30 | 0.60 | the middle point, stopped at step 110 | hps-only | 0.00 | 0.90 | the validation run, stopped at step 60 | I started with hps-only to validate that the pipeline could learn at all. Once the reward was going up and the metrics were healthy, there was no reason to run it longer, so I launched the two longer runs instead. The question that the longer runs ask is how much of HPSv3's power can you hand to the pairwise judge? The more weight the judge carries, the more the reward means my taste instead of everyone's, and the harder it should be to climb. Incidentally, if you push it far enough or your style is too far from the average, the model could stop entirely. Fortunately, it did not stop and both runs with the pairwise judge on learned too. The hand-rated pool can steer the policy, at least as far as the metrics and the final paintings show. The numbers are below. Disclaimer.If we use a frontier model, it can already generate the JavaScript code that paints watercolour from a prompt. That's the starting point. The work here is about teaching a smaller model to do it combined with a person's own artistic preferences. The RL environment you need to build The environment wraps everything that sits between the model and the reward, including the JavaScript library that the model uses for painting, the system prompt that restricts it, the headless Chromium that renders each sketch, and the gate that rejects cheats. The library does more of the work than it seems. p5.brush https://github.com/acamposuribe/p5.brush , by @acamposuribe https://x.com/acamposuribe , simulates a medium rather than drawing shapes: pigment bleeds past the edges of a fill, paper has texture, strokes have mass, flow fields drag brushwork around. When the model calls brush.fillBleed 0.25 it is deciding how far the ink runs. Note.The author of p5.brush had been trying to teach a machine to paint long before any of this. In 2022 he made a generative art series that hides a diary about teaching p5.js to draw like a child:"It is barely able to use the crayons ... It cannot follow simple commands. I'm done for today, very infuriating."The series was meant to have three pieces, and he made two. When Surya's video went viral, he quoted it , sharing that diary and saying this work is the third piece arriving on its own. p5.brush exposes 47 methods. The prompt allows 10: scaleBrushes , noStroke , fill , noFill , fillBleed , fillTexture , beginShape , vertex , endShape and circle . What the other thirty-seven add, lines, hatching, custom brushes, would break the watercolour look. With these ten the model can only paint filled shapes, and the library adds the bleed to every one of them. His blog post saved me a lot of time I could have wasted iterating on the prompt. A long API reference makes the model invent methods that do not exist, and his 200 GEPA iterations converged on a strict allowlist with no documentation. I saw the same failures and wrote the allowlist by hand. My only addition to that recipe is one sentence: paint each petal two or three times, a big pass first and a smaller, more opaque one inside it. This small change made my outputs a lot more colorful. Note.If it's the first time you hear about GEPA , it is an automatic prompt optimizer. A language model reflects in plain words on where the current prompt failed and proposes a better one, and the loop repeats. The gate is the final piece. The sketch has to compile, use the library instead of direct p5 calls, put real pigment on the canvas, and not try to trick the scorer, for example by writing text on the canvas. The pool is the reward function The pool consists of 178 paintings https://huggingface.co/datasets/HuggingEnvs/watercolour-reference-pool divided into two tiers based on my personal preferences, love and okay . All of them are actually generated by a model. Four open-weight models, called through Inference Providers, wrote p5.brush sketches, each one working from a real, openly licensed photo of a hibiscus from iNaturalist. A vision model gave written feedback on every sketch, over three refinement iterations. Every final render was then rated one at a time, by me, and 178 made the cut. | generator | number of paintings | |---|---| | Kimi-K3 https://huggingface.co/moonshotai/Kimi-K3 Qwen3-Coder-Next https://huggingface.co/Qwen/Qwen3-Coder-Next Qwen3.5-122B-A10B https://huggingface.co/Qwen/Qwen3.5-122B-A10B Here I chose four different families of models to test their different styles. These four were the open models that produced a valid sketch every time in a quick reliability check, and two other candidates were dropped for failing it. If you want to produce your own pool, you might choose others. The tiers do real work in the reward. When the pairwise judge draws four references, half come from love and half from okay , so the policy always faces some rivals it can sometimes beat, and a win pays the same against either tier. This is one of my few deliberate changes: the original compares against its top tier only, and I kept the easier tier in the draw so a weak early policy still gets signal. No human-made painting is in there, which is a real limitation. p5.brush is a niche library, and the human work that exists in it with accessible code is a handful of pieces, nowhere near what a training corpus would need, as his blog also notes. The interesting idea here, as I've already discussed previously, is that the model will learn to imitate what the pool contains. If we point the environment at a different dataset, the reward would automatically change without touching a single line of code. For the dataset I generated and openly share, I also include the source sketch. If we look closer at the judges, the two of them answer different questions. HPSv3 decides whether it is a flower, and the pairwise judge decides whether it is well painted in the style I chose. Just one more yolo run Before anything worked, there was a long stretch of flat reward curves. If you've tried to reproduce a research paper/blog without open artifacts, you probably can relate. Every run tested what I thought was a reasonable theory about what was wrong. A run takes a lot of time, so I queued the next one while still going through the last one's results. As always, starting from something easier that works and then building on top of that was the answer. A simple control task, with no browser and no judges, was the first attempt that learned, and the reason was that my learning rate was just too low. Another change that cost me time to find was adjusting the LoRA parameters correctly. The usual target modules list assumes a dense model, and Qwen/Qwen3.5-35B-A3B https://huggingface.co/Qwen/Qwen3.5-35B-A3B is a mixture of experts that names most of its projections differently, so the adapter was training ten layers out of forty. I solved this by changing it to all-linear , which reaches every linear layer. The routed experts in this architecture are fused tensors that even all-linear leaves frozen, but everything else gets an adapter, and that was enough to learn.The fix was four changes in TRL's GRPOTrainer https://huggingface.co/docs/trl/grpo trainer : | setting | from | to | why | |---|---|---|---| | learning rate | 2e-5 | 5e-5 | the ceiling LoRA Without Regret uses for GRPO | | scheduler | linear | constant with warmup | linear decay had spent most of the learning rate by mid-run, so the reward never took off | scale rewards | group | none | one gate rejection was shrinking every other advantage in the group | target modules | hand list | all-linear | reach every linear layer | These four changes unlocked the first successful run hps-only , with the reward clearly improving. With that configuration, all three runs learn. Both judge runs were launched for 200 steps and stopped at 110, with the reward still climbing slowly. A step takes fifteen to eighteen minutes, and the comparison between mixes was already stable, so I stopped both to save compute. Mean group reward over the first and final third of each run: | run | steps | first third | final third | Δ | |---|---|---|---|---| hps-only | 60 | 0.58 | 0.71 | +0.13 | judge-led | 110 | 0.45 | 0.72 | +0.27 | hps-led | 110 | 0.57 | 0.82 | +0.24 | The three curves line up with how much weight my taste carries. The more the judge weighs, the lower the start and the noisier the climb. judge-led spent its first thirty steps nearly flat before it moved. It is the same move that solved the debugging. Shrink the problem until something learns, then add the hard parts back one at a time. The pairwise judge term itself climbed in the two runs that used it. The model wins more comparisons against the pool as training advances, which is the claim hps-only could not make. No group in any run collapsed to identical rewards, the GRPO failure mode that kills the gradient. For the curious, the per-metric curves HPSv3, paint coverage, entropy are in the repository as CSV https://github.com/adithya-s-k/HuggingEnvs/tree/main/02-watercolour/results . The full launch command, the hardware and the two environment variables that turn this into the other two runs are in the recipe https://github.com/adithya-s-k/HuggingEnvs/tree/main/02-watercolour . What it actually learned In every run, the first thing the model learned was to stop producing bad paintings , the near-blank canvases and shapeless washes that score under 0.3 in total reward. In hps-only , three quarters of the rise in the group mean comes from bad paintings becoming rare. In the judge runs the collapse is even steeper: rollouts under 0.3 fall from 99 to 16 across judge-led 's thirds, and from 37 to 4 in hps-led . This is why the obvious visual, the best painting of each step, shows almost no difference in hps-only . It moves +0.034 across the run while the median moves +0.155 . The learning is visible in the middle of the distribution. What the pairwise judge changes is the top. In hps-only , paintings got more reliable without getting better. The quality of the good ones added just +0.03 to the group mean, and once HPSv3 saw petals around a centre and a stem, it stopped asking for more pigment. With the judge on, the other half of the story appears. Better here means closer to the pool, so closer to what I rated as "good" or something I liked more. That added +0.12 in judge-led and +0.16 in hps-led , the best of each step rose too, and paint coverage doubled in both runs 0.11 to 0.23, and 0.13 to 0.30 where hps-only barely moved it. With a reference left to beat, a good painting can still get better, and the model starts being rewarded for using more pigment. One more finding. The model ignores an explicit instruction, and it is right to. The system prompt asks for fifteen to thirty filled shapes. If we look at the real mean, it is between 7 and 9, and n shapes barely correlates with reward in any run +0.000, −0.14, +0.07 . The policy is not rewarded for obeying that sentence, so it does not obey it. There is also a ceiling on the hps-only route. If every rollout matched its good ones, that run's mean would sit at 0.771. Whether more steps would break it is an open question. The paintings also show something that the tables miss. Within each run, they all look similar. As training advances, the rewards inside each group get closer together, and the median paintings in the opening video look like takes of the same flower. That is GRPO doing what it is designed to do with a pool built from one subject. The pool decides what counts as variety, the same way it decides what counts as quality. If the reward only pays for matching one flower, the model learns to paint that one flower. More diverse output would need a more diverse pool, and building one is more curation work. Surya's newer compositions are an example of this. Alex Yango's animal paintings https://x.com/alexyango/status/2091696296931574217 are the same recipe with different choices in the pool. This is the biggest difference between an aesthetic reward and a maths grader. Behind the number there is a very human job, deciding what belongs in the reward set. Jason Liu's essay on taste https://x.com/jxnlco/status/2073819508729684462 says the general version in one line. AI shifted the bottleneck from making to noticing. Surya closes his blog with some of his favourites. Instead of picking mine, below is a wall with the 178 paintings the reward scored highest across the two judge runs, the same number the reference pool holds, in no particular order. Open it and pick your own. To choose, you looked at many and kept a few, and that is exactly the job that built the reward of this project. Every painting of every run, with its sketch and its reward, is in the rollouts datasets, and browsable in this gallery https://huggingface.co/spaces/HuggingEnvs/watercolour-gallery . Since the reward was partly based on my taste, it is fair to close with my verdict as a viewer. To my eye, judge-led is the run that ends up the most diverse and the most artistically interesting. hps-led paints convincing watercolours, but its best ones share a soft, wet-on-wet look that is almost a style of its own. hps-only converges the hardest, and most of its paintings settle on the same colours. You can judge for yourself in the gallery https://huggingface.co/spaces/HuggingEnvs/watercolour-gallery , which has every painting of every run, sortable by step and by reward. Infra is hard This project is mostly infra. A run needs a trainer, two Spaces, an inference router and a websocket to stay healthy for hours straight, and every piece that fails quietly turns into a wrong number somewhere else. Half the work is checking that the number you read matches what actually happened. Failures of the infrastructure were entering the reward as zeros. A render that timed out or a scorer that did not answer scored the same as a bad painting, 0.0 inside the group. Across all my runs that was about 1.5% of rollouts, and in the worst run it reached 5.2%. That trains the model on noise, so those paths now return None and the rollout is excluded from the group. I also found a bug in OpenEnv, and sent the fix upstream. The client keeps one persistent websocket, and a socket closed by the far end stayed cached, so every later call failed even though the environment was healthy. It cost me two half-finished runs to find it. The fix is submitted upstream https://github.com/huggingface/OpenEnv/pull/1103 , and the runs launched with it have been running clean since. The reward of a step depends on which references it drew. The pairwise judge samples four references per step, so every step faces a different set of rivals, and some draws are simply harder. GRPO itself is mostly safe, because advantages are computed inside the group and a hard draw moves the whole group together. The curve I was reading was not safe, and some of what looked like a bad step was just a hard draw. The image above is one example. Step 12 scored half a point below step 11 mostly because it drew the hardest references of the run, while the paintings themselves look close. What it costs Rounded numbers, and only for the runs that finished. | piece | what it needs | |---|---| | trainer | 1 H200. 18 hours for 60 steps, about 34 for 110 | | HPSv3 | an a100-large Space, up for the whole run | | the environment | a cpu-upgrade Space, which renders comfortably in time | | the pairwise judge | Inference Providers quota for Qwen/Qwen3-VL-30B-A3B-Instruct | | the pool, one-off | openly licensed photos from iNaturalist, Inference Providers quota for the four generators, and rating is your own hours | A step is eight rollouts and takes fifteen to eighteen minutes, of which 70 to 80% is rendering . A single render takes 69 to 96 seconds against a 90 second deadline. Part of that is expected, the Space has no GPU, so Chromium renders the WEBGL canvas in software and p5.brush's bleeds and textures are heavy pixel work. Even so, I expected it to be faster, and I have not found the full cause. A scorer can cost more than the training that uses it: HPSv3 has to be up for the whole run, so pause the Space, or set its sleep timer, when the run ends. Everything runs on HF Jobs https://huggingface.co/docs/huggingface hub/guides/jobs , with the environment as a Docker Space https://huggingface.co/docs/hub/spaces-sdks-docker and metrics in trackio https://huggingface.co/docs/trackio . What I would try next The rule of this project was to reproduce the recipe with every resource open, not to improve it, so a list of untried ideas piled up along the way. These are the ones I would actually try, in order of how much evidence there is. Multi-step, and letting the model see what it paints. This is the first thing I would try. The original blog trains single turn, so I trained single turn, and in this setup the model paints with its eyes closed. No image ever goes in, and the only feedback it gets is one number. The evidence that a feedback loop works is the pool itself. The reference paintings came from models iterating three rounds under a vision critic, and the later rounds are better. The material that defines the reward was made with a loop the policy never gets. Smaller models. There is evidence that 35B is more than needed. In my side experiments a 4B already wrote valid sketches that passed the gate. If a 4B can learn this, the cost of the experiment drops by an order of magnitude. Other ideas on the list are SFT on the pool sources before starting RL, rewarding pigment explicitly, moving the judge's reference mix from easy to hard as the run advances until only love remains, widening the ten-method allowlist for more visual range my attempts on this crashed more sketches and broke the watercolour look , and checking how consistent the pairwise judge really is by scoring the same image twice. And the method is not specific to flowers. Alex Yango painted animals with the same mechanism https://x.com/alexyango/status/2091696296931574217 , and Brendan Hogan trained canvas animations https://x.com/brendanh0gan/status/2092650655789855222 against a pool of hand-rated clips. I had also played with something similar before using Simon Willison's pelican benchmark https://huggingface.co/blog/sergiopaniego/pelican-env-openenv , where code is rendered to an image and scored. And underneath all of it sits the question this project cannot close. 178 paintings made by models define what this trained model considers beautiful. The pool is the bottleneck, and it is the part of the pipeline with no principled answer. What I changed from the original For anyone reproducing this, there are two deliberate divergences from Narreddi's recipe, both argued earlier in the article: - The pairwise judge draws its references half from love and half from okay , instead of comparing against the top tier only, so a weak early policy still gets signal. - One small sentence of craft in the prompt, paint each petal two or three times, a big pass first and a smaller, more opaque one inside it. The rest, all-linear for the LoRA, infrastructure failures returning None instead of 0.0, stopping the judge runs at step 110, are decisions I had to make along the way because his blog does not specify them. His implementation is not published, so I cannot tell whether they match his choices or diverge from them. Everything is published | artifact | where | |---|---| | the recipe, and how to reproduce it | 02-watercolour/ | watercolour-reference-pool watercolour-env watercolour-hpsv3 · https://huggingface.co/HuggingEnvs/watercolour-grpo-hps-only watercolour-grpo-hps-only watercolour-rollouts-hps-only · https://huggingface.co/HuggingEnvs/watercolour-grpo-judge-led watercolour-grpo-judge-led watercolour-rollouts-judge-led · https://huggingface.co/HuggingEnvs/watercolour-grpo-hps-led watercolour-grpo-hps-led watercolour-rollouts-hps-led watercolour-gallery · https://huggingface.co/spaces/HuggingEnvs/watercolour-trackio-judge-led judge-led · https://huggingface.co/spaces/HuggingEnvs/watercolour-trackio-hps-led hps-led , and the CSV files in https://huggingface.co/spaces/HuggingEnvs/watercolour-trackio-hps-only hps-only results/ Paint with Code https://huggingface.co/collections/HuggingEnvs/paint-with-code-6a955b79d63f67f1631d9be6 The per-rollout numbers in this article can be recomputed from the published datasets. Nothing here depends on any Space staying switched on. The method and original idea are Surya Narreddi's. The library is Alejandro Campos Uribe's.