ARC-AGI
Two years, 18 solver architectures across 9 families.
The finding
The binding constraint is object-level relational structure that is not recoverable from pixel context — not compute.
Commit counts are those unique to each branch, measured from the common ancestor. Total branch length would count shared history many times over and flatter every approach equally.
End-to-end neural
Latent-process CNN
2024-08 – 2024-10OpinionAI@main
Pixel-level embeddings with no downsampling, primitive selection over 64 operations, cross-attention between demonstrations and test input, and coupled shape/content heads.
Pixel-level embeddings with no downsampling, and selection over 64 primitive operations. The base both later OpinionAI branches fork from.
CNN-only baseline
2024-08 – 2025-09 · 54 commitsOpinionAI@cnn-only
Deliberately stripped convolutional baseline, to establish what pixel-context alone can reach.
A full codebase overhaul: content- and shape-aware encoders, a joint decoder, and shape prediction for the N-latent variant.
Vision encoder
2025-09 – 2025-10 · 152 commitsOpinion@vision
Treat the grid as an image and lean on vision-style feature extraction.
Padding, and variable grid size. PAD removed from the vocabulary, then reweighted, then supervised; grids cropped properly; then device management and multiprocessing.
Meta-learning
Meta-learning
2024-08 – 2024-10OpinionAI@meta-learning
Treat each task as an episode and learn an initialisation that adapts from the demonstration pairs alone.
Episodic adaptation from the demonstration pairs alone. Shares the base history; no commits unique to the line.
Reptile meta-learning
2025-09 – 2025-10 · 7 commitsOpinion@meta-learning
Reptile-style first-order meta-learning across tasks.
Five reformulations in seven commits: VQ-VAE, Gumbel-Softmax VQ-VAE, fewer codes, a 64-element basis, then Reptile.
RL
Reinforcement learning
2024-08 – 2024-10 · 53 commitsOpinionAI@rl
Grid construction as a sequential decision problem, with reward shaping tuned to give usable gradients rather than sparse terminal signal.
Reward shaping. Granular reward functions, rewards retuned specifically to give usable gradients, and repeated reverts to previously working prediction and evaluation code.
Bayesian
Active inference
2024-08 – 2024-11 · 2 commitsOpinionAI@active-inference
Free-energy formulation: pick the action that minimises expected surprise over the predicted grid.
Two commits on the line: a deeper network, and meta-learning combined with active inference. Dropped almost immediately.
Variational solver
2025-09 – 2026-03 · 214 commitsOpinion@main
A variational encoder distils the transformation rule from the demonstration pairs into a latent z; a FiLM-conditioned delta decoder applies z. Trained on an ELBO with a coupled constraint forcing demo reconstruction and test prediction to improve together.
Converged on the variational solver: a coupled ELBO forcing demo reconstruction and test prediction down together, under cosine annealing.
Statistical inversion
2025-09 – 2026-05Opinion@claude/arc-agi-bayesian-approach
Experiment harness treating the task as inverting a statistical generating process.
An experiment harness for treating the task as inverting a statistical generating process. Branched from the variational line.
Relational
Hypergraph GNN + meta-learning
2024-08 – 2025-06 · 43 commitsOpinionAI@gnn-meta
Grids as hypergraphs over cells and regions, message passing with HypergraphConv, wrapped in a meta-learning outer loop.
Getting MAML right, a distance-based loss, and predicting discrete grid heights and widths rather than regressing them.
Relational model
2025-09 – 2026-04 · 215 commitsOpinion@relational
Explicit relational structure over grid objects, with the coupled loss and D4 symmetry augmentation. Migrated back from OpinionAI.
The loss, again. DenseOrdinalReplicaLoss with adaptive class weighting and focal loss, mask-signal weighting, a balance penalty, bounding-box predictions, and debug logging added to see what training was actually doing.
Program synthesis
SAT / hypergraph
2024-08 – 2025-06 · 35 commitsOpinionAI@sat
Constraint framing over a hypergraph encoding of the grid, searching for an assignment consistent with every demonstration.
Hypergraph convolutions under a meta-learning loop, latent projector capacity, and penalties built around exact matches — inverse scaling and a multiplicative perfect-match penalty.
Combinator grammar (SKI)
2025-09 – 2025-10 · 4 commitsOpinion@grammar
Grammar-constrained combinator learning: select and compose formal operations from the demonstration pairs, replacing the prototype-based architecture.
Probabilistic SKI, then robustness and device fixes. Four commits on this line before it was left.
TyleDSL
2025-06TyleDSL@main
A typed reformulation of Michael Hodel's ARC-DSL, using the type system to prune the space of candidate programs.
A typed reformulation of the ARC-DSL. The type system prunes candidate programs before they are ever evaluated.
Transformer
Traditional transformer
2025-09 – 2025-10 · 137 commitsOpinion@traditional-transformer
Standard encoder-decoder over serialised grids with a dense per-cell loss.
The loss formulation, circularly. L2 replaced L1, then L1 was restored, then a dense per-cell loss replaced both.
Elastic transformer
2025-09 – 2026-01 · 270 commitsOpinion@transformer
Transformer over an ElasticConv1d front end sized to variable grid dimensions.
Attention cost first, then stability. O(n^2) attention was replaced by an O(n) adaptive CNN, a PVR refinement loop with a verifier and refiner was added, a cosine schedule with linear warmup after that — and the line ends on gradient explosion in ElasticConv1d.
Cellular automata
Neural cellular automaton
2025-09 – 2025-11 · 159 commitsOpinion@nca
Output shape learned implicitly through CA dynamics rather than predicted: seed the test input at the centre of a 30×30 canvas, evolve, and let the automaton learn where to stop.
The loss and the automaton's dynamics. L1 against L2, coverage switched to L1, automatic interpolation, demo-conditioned features, and perturbations to the evolving state.
LLM
LLM + LoRA
2025-09 – 2025-10 · 200 commitsOpinion@llm
Low-rank adaptation of a pretrained language model over serialised grids.
Termination and length. EOS fallback handling, repeated token-limit and max-length increases, evaluation fixes, and eventually a dedicated diagnosis command to see why generations were failing.
Each entry is a real branch in the Opinion, OpinionAI or TyleDSL repositories. Most of these repositories are private, so branch names are given as provenance rather than links. 'What it fought' is summarised from the commits unique to that branch — it is what the history shows each approach wrestling with, not a stated conclusion. Where it finally broke, and the benchmark numbers, are still to be written up: no branch records a solve rate.