arcprize

This repository contains the TinyRecursiveModels checkpoints for arc v1 public eval and arc v2 public eval that were trained for the performance verification. They were trained using the code and recipe of the official TRM repository. We had to adapt the environment setup as detailed below. We provide these checkpoints for transparency and to facilitate further research. We did not contribute to the TRM reserach nor maintain the TRM code. For any questions, please reach out to the TRM maintainers. TRM writes checkpoints as `torch statedicts`. The subdirectories `arcv1public` and `arcv2public` contain the final checkpoints `step `, which can be loaded with the `loadcheckpoint` or by providing the checkpoint path as `loadcheckpoint=path/to/checkpoint`. For reference, see the `PretrainConfig` in `pretrain.py`. - ARC-AGI-1: 40%, $1.76/task - ARC-AGI-2: 6.2%, $2.10/task Tweet: https://x.com/arcprize/status/1978872651180577060 Leaderboard: https://arcprize.org/leaderboard Dataset preprocessing The repository already contains the raw data, but it needs to be preprocessed. Run the following commands to preprocess the v1 and v2 datasets to make predictions for the public eval datasets. Training To reproduce the checkpoints, run the following two training runs on a single 8:H100 node. Each run takes ~20-30h. To speed it up, instructions for multi-node training are below.