{ "mode": "marginal", "args": { "mode": "marginal", "seed": 23, "modulus": 11, "op_family": "full", "train_min_len": 1, "train_max_len": 6, "eval_lengths": "3,6,9,12", "eval_k": "0,1,2,3,6,9,12", "transition_temperature": 1.0, "dim": 128, "heads": 4, "compiler_layers": 2, "dropout": 0.05, "train_steps": 300, "batch_size": 512, "eval_batch_size": 512, "eval_examples": 1024, "lr": 0.03, "weight_decay": 0.0, "max_grad_norm": 1.0, "log_every": 25, "eval_every": 100, "output_dir": "experiments/joint_register_executor/runs/control_marginal_mod11", "checkpoint_dir": "large_artifacts/joint_register_executor/checkpoints/control_marginal_mod11" }, "env": { "python": "3.12.3", "torch": "2.8.0+cu128", "device": "cuda", "cuda_name": "NVIDIA RTX 6000 Ada Generation" }, "train": [ { "step": 1, "loss": 3.086867570877075, "pair_acc": 0.017578125, "grad_norm": 0.03142501413822174, "transition_entropy": 2.397849361101786 }, { "step": 25, "loss": 2.6942343711853027, "pair_acc": 0.3203125, "grad_norm": 0.032605767250061035, "transition_entropy": 2.3729117711385093 }, { "step": 50, "loss": 2.0430963039398193, "pair_acc": 0.62890625, "grad_norm": 0.03622467815876007, "transition_entropy": 2.2464651664098105 }, { "step": 75, "loss": 1.2070151567459106, "pair_acc": 0.96875, "grad_norm": 0.03477146849036217, "transition_entropy": 1.9115692575772603 }, { "step": 100, "loss": 0.5651628375053406, "pair_acc": 1.0, "grad_norm": 0.021922508254647255, "transition_entropy": 1.5024449129899342 }, { "step": 125, "loss": 0.2953287661075592, "pair_acc": 1.0, "grad_norm": 0.012529619969427586, "transition_entropy": 1.3138638734817505 }, { "step": 150, "loss": 0.182949498295784, "pair_acc": 1.0, "grad_norm": 0.007916738279163837, "transition_entropy": 1.2179111937681835 }, { "step": 175, "loss": 0.13767680525779724, "pair_acc": 1.0, "grad_norm": 0.006002397276461124, "transition_entropy": 1.1444205443064372 }, { "step": 200, "loss": 0.10903125256299973, "pair_acc": 1.0, "grad_norm": 0.004762078635394573, "transition_entropy": 1.0823808709780376 }, { "step": 225, "loss": 0.08092417567968369, "pair_acc": 1.0, "grad_norm": 0.0037094755098223686, "transition_entropy": 1.169189599653085 }, { "step": 250, "loss": 0.07179122418165207, "pair_acc": 1.0, "grad_norm": 0.0032499174121767282, "transition_entropy": 1.0589657028516133 }, { "step": 275, "loss": 0.05933147296309471, "pair_acc": 1.0, "grad_norm": 0.0026845361571758986, "transition_entropy": 1.0591234465440114 }, { "step": 300, "loss": 0.054570965468883514, "pair_acc": 1.0, "grad_norm": 0.0024620965123176575, "transition_entropy": 1.0475060890118282 } ], "eval": [ { "step": 100, "rows": [ { "model": "marginal", "length": 3, "k": 0, "n": 1024, "pair_accuracy": 0.0283203125, "a_accuracy": 0.20703125, "b_accuracy": 0.2197265625, "register_accuracy": 0.21337890625 }, { "model": "marginal", "length": 3, "k": 1, "n": 1024, "pair_accuracy": 0.095703125, "a_accuracy": 0.3369140625, "b_accuracy": 0.30078125, "register_accuracy": 0.31884765625 }, { "model": "marginal", "length": 3, "k": 2, "n": 1024, "pair_accuracy": 0.0556640625, "a_accuracy": 0.5341796875, "b_accuracy": 0.521484375, "register_accuracy": 0.52783203125 }, { "model": "marginal", "length": 3, "k": 3, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 3, "k": 6, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 3, "k": 9, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 3, "k": 12, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 6, "k": 0, "n": 1024, "pair_accuracy": 0.0126953125, "a_accuracy": 0.1005859375, "b_accuracy": 0.1083984375, "register_accuracy": 0.1044921875 }, { "model": "marginal", "length": 6, "k": 1, "n": 1024, "pair_accuracy": 0.0166015625, "a_accuracy": 0.109375, "b_accuracy": 0.126953125, "register_accuracy": 0.1181640625 }, { "model": "marginal", "length": 6, "k": 2, "n": 1024, "pair_accuracy": 0.0283203125, "a_accuracy": 0.1455078125, "b_accuracy": 0.154296875, "register_accuracy": 0.14990234375 }, { "model": "marginal", "length": 6, "k": 3, "n": 1024, "pair_accuracy": 0.0234375, "a_accuracy": 0.197265625, "b_accuracy": 0.20703125, "register_accuracy": 0.2021484375 }, { "model": "marginal", "length": 6, "k": 6, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 6, "k": 9, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 6, "k": 12, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 9, "k": 0, "n": 1024, "pair_accuracy": 0.0068359375, "a_accuracy": 0.0908203125, "b_accuracy": 0.095703125, "register_accuracy": 0.09326171875 }, { "model": "marginal", "length": 9, "k": 1, "n": 1024, "pair_accuracy": 0.0078125, "a_accuracy": 0.1005859375, "b_accuracy": 0.0888671875, "register_accuracy": 0.0947265625 }, { "model": "marginal", "length": 9, "k": 2, "n": 1024, "pair_accuracy": 0.0107421875, "a_accuracy": 0.099609375, "b_accuracy": 0.1005859375, "register_accuracy": 0.10009765625 }, { "model": "marginal", "length": 9, "k": 3, "n": 1024, "pair_accuracy": 0.013671875, "a_accuracy": 0.09765625, "b_accuracy": 0.1123046875, "register_accuracy": 0.10498046875 }, { "model": "marginal", "length": 9, "k": 6, "n": 1024, "pair_accuracy": 0.0341796875, "a_accuracy": 0.21875, "b_accuracy": 0.2158203125, "register_accuracy": 0.21728515625 }, { "model": "marginal", "length": 9, "k": 9, "n": 1024, "pair_accuracy": 0.892578125, "a_accuracy": 0.9453125, "b_accuracy": 0.939453125, "register_accuracy": 0.9423828125 }, { "model": "marginal", "length": 9, "k": 12, "n": 1024, "pair_accuracy": 0.892578125, "a_accuracy": 0.9453125, "b_accuracy": 0.939453125, "register_accuracy": 0.9423828125 }, { "model": "marginal", "length": 12, "k": 0, "n": 1024, "pair_accuracy": 0.0087890625, "a_accuracy": 0.0947265625, "b_accuracy": 0.0810546875, "register_accuracy": 0.087890625 }, { "model": "marginal", "length": 12, "k": 1, "n": 1024, "pair_accuracy": 0.009765625, "a_accuracy": 0.0947265625, "b_accuracy": 0.08203125, "register_accuracy": 0.08837890625 }, { "model": "marginal", "length": 12, "k": 2, "n": 1024, "pair_accuracy": 0.00390625, "a_accuracy": 0.1025390625, "b_accuracy": 0.076171875, "register_accuracy": 0.08935546875 }, { "model": "marginal", "length": 12, "k": 3, "n": 1024, "pair_accuracy": 0.005859375, "a_accuracy": 0.1064453125, "b_accuracy": 0.0869140625, "register_accuracy": 0.0966796875 }, { "model": "marginal", "length": 12, "k": 6, "n": 1024, "pair_accuracy": 0.0078125, "a_accuracy": 0.12109375, "b_accuracy": 0.115234375, "register_accuracy": 0.1181640625 }, { "model": "marginal", "length": 12, "k": 9, "n": 1024, "pair_accuracy": 0.0283203125, "a_accuracy": 0.2060546875, "b_accuracy": 0.2138671875, "register_accuracy": 0.2099609375 }, { "model": "marginal", "length": 12, "k": 12, "n": 1024, "pair_accuracy": 0.4873046875, "a_accuracy": 0.6513671875, "b_accuracy": 0.6767578125, "register_accuracy": 0.6640625 } ] }, { "step": 200, "rows": [ { "model": "marginal", "length": 3, "k": 0, "n": 1024, "pair_accuracy": 0.025390625, "a_accuracy": 0.234375, "b_accuracy": 0.212890625, "register_accuracy": 0.2236328125 }, { "model": "marginal", "length": 3, "k": 1, "n": 1024, "pair_accuracy": 0.109375, "a_accuracy": 0.341796875, "b_accuracy": 0.31640625, "register_accuracy": 0.3291015625 }, { "model": "marginal", "length": 3, "k": 2, "n": 1024, "pair_accuracy": 0.0537109375, "a_accuracy": 0.5263671875, "b_accuracy": 0.52734375, "register_accuracy": 0.52685546875 }, { "model": "marginal", "length": 3, "k": 3, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 3, "k": 6, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 3, "k": 9, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 3, "k": 12, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 6, "k": 0, "n": 1024, "pair_accuracy": 0.017578125, "a_accuracy": 0.095703125, "b_accuracy": 0.1201171875, "register_accuracy": 0.10791015625 }, { "model": "marginal", "length": 6, "k": 1, "n": 1024, "pair_accuracy": 0.0166015625, "a_accuracy": 0.1298828125, "b_accuracy": 0.1220703125, "register_accuracy": 0.1259765625 }, { "model": "marginal", "length": 6, "k": 2, "n": 1024, "pair_accuracy": 0.0205078125, "a_accuracy": 0.1708984375, "b_accuracy": 0.14453125, "register_accuracy": 0.15771484375 }, { "model": "marginal", "length": 6, "k": 3, "n": 1024, "pair_accuracy": 0.02734375, "a_accuracy": 0.21484375, "b_accuracy": 0.1962890625, "register_accuracy": 0.20556640625 }, { "model": "marginal", "length": 6, "k": 6, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 6, "k": 9, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 6, "k": 12, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 9, "k": 0, "n": 1024, "pair_accuracy": 0.0068359375, "a_accuracy": 0.0888671875, "b_accuracy": 0.0908203125, "register_accuracy": 0.08984375 }, { "model": "marginal", "length": 9, "k": 1, "n": 1024, "pair_accuracy": 0.0048828125, "a_accuracy": 0.099609375, "b_accuracy": 0.072265625, "register_accuracy": 0.0859375 }, { "model": "marginal", "length": 9, "k": 2, "n": 1024, "pair_accuracy": 0.0087890625, "a_accuracy": 0.1064453125, "b_accuracy": 0.095703125, "register_accuracy": 0.10107421875 }, { "model": "marginal", "length": 9, "k": 3, "n": 1024, "pair_accuracy": 0.013671875, "a_accuracy": 0.119140625, "b_accuracy": 0.095703125, "register_accuracy": 0.107421875 }, { "model": "marginal", "length": 9, "k": 6, "n": 1024, "pair_accuracy": 0.02734375, "a_accuracy": 0.2158203125, "b_accuracy": 0.2216796875, "register_accuracy": 0.21875 }, { "model": "marginal", "length": 9, "k": 9, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 9, "k": 12, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 12, "k": 0, "n": 1024, "pair_accuracy": 0.0068359375, "a_accuracy": 0.0888671875, "b_accuracy": 0.0927734375, "register_accuracy": 0.0908203125 }, { "model": "marginal", "length": 12, "k": 1, "n": 1024, "pair_accuracy": 0.0068359375, "a_accuracy": 0.099609375, "b_accuracy": 0.103515625, "register_accuracy": 0.1015625 }, { "model": "marginal", "length": 12, "k": 2, "n": 1024, "pair_accuracy": 0.009765625, "a_accuracy": 0.08984375, "b_accuracy": 0.099609375, "register_accuracy": 0.0947265625 }, { "model": "marginal", "length": 12, "k": 3, "n": 1024, "pair_accuracy": 0.005859375, "a_accuracy": 0.0888671875, "b_accuracy": 0.0986328125, "register_accuracy": 0.09375 }, { "model": "marginal", "length": 12, "k": 6, "n": 1024, "pair_accuracy": 0.017578125, "a_accuracy": 0.1064453125, "b_accuracy": 0.109375, "register_accuracy": 0.10791015625 }, { "model": "marginal", "length": 12, "k": 9, "n": 1024, "pair_accuracy": 0.015625, "a_accuracy": 0.22265625, "b_accuracy": 0.1943359375, "register_accuracy": 0.20849609375 }, { "model": "marginal", "length": 12, "k": 12, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 } ] }, { "step": 300, "rows": [ { "model": "marginal", "length": 3, "k": 0, "n": 1024, "pair_accuracy": 0.025390625, "a_accuracy": 0.201171875, "b_accuracy": 0.2197265625, "register_accuracy": 0.21044921875 }, { "model": "marginal", "length": 3, "k": 1, "n": 1024, "pair_accuracy": 0.0986328125, "a_accuracy": 0.314453125, "b_accuracy": 0.328125, "register_accuracy": 0.3212890625 }, { "model": "marginal", "length": 3, "k": 2, "n": 1024, "pair_accuracy": 0.044921875, "a_accuracy": 0.517578125, "b_accuracy": 0.52734375, "register_accuracy": 0.5224609375 }, { "model": "marginal", "length": 3, "k": 3, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 3, "k": 6, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 3, "k": 9, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 3, "k": 12, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 6, "k": 0, "n": 1024, "pair_accuracy": 0.0126953125, "a_accuracy": 0.1240234375, "b_accuracy": 0.1064453125, "register_accuracy": 0.115234375 }, { "model": "marginal", "length": 6, "k": 1, "n": 1024, "pair_accuracy": 0.0126953125, "a_accuracy": 0.138671875, "b_accuracy": 0.1240234375, "register_accuracy": 0.13134765625 }, { "model": "marginal", "length": 6, "k": 2, "n": 1024, "pair_accuracy": 0.0263671875, "a_accuracy": 0.1669921875, "b_accuracy": 0.16015625, "register_accuracy": 0.16357421875 }, { "model": "marginal", "length": 6, "k": 3, "n": 1024, "pair_accuracy": 0.0244140625, "a_accuracy": 0.189453125, "b_accuracy": 0.2197265625, "register_accuracy": 0.20458984375 }, { "model": "marginal", "length": 6, "k": 6, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 6, "k": 9, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 6, "k": 12, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 9, "k": 0, "n": 1024, "pair_accuracy": 0.0087890625, "a_accuracy": 0.087890625, "b_accuracy": 0.10546875, "register_accuracy": 0.0966796875 }, { "model": "marginal", "length": 9, "k": 1, "n": 1024, "pair_accuracy": 0.0126953125, "a_accuracy": 0.0966796875, "b_accuracy": 0.10546875, "register_accuracy": 0.10107421875 }, { "model": "marginal", "length": 9, "k": 2, "n": 1024, "pair_accuracy": 0.0126953125, "a_accuracy": 0.0986328125, "b_accuracy": 0.1181640625, "register_accuracy": 0.1083984375 }, { "model": "marginal", "length": 9, "k": 3, "n": 1024, "pair_accuracy": 0.013671875, "a_accuracy": 0.1123046875, "b_accuracy": 0.123046875, "register_accuracy": 0.11767578125 }, { "model": "marginal", "length": 9, "k": 6, "n": 1024, "pair_accuracy": 0.0380859375, "a_accuracy": 0.234375, "b_accuracy": 0.2529296875, "register_accuracy": 0.24365234375 }, { "model": "marginal", "length": 9, "k": 9, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 9, "k": 12, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 }, { "model": "marginal", "length": 12, "k": 0, "n": 1024, "pair_accuracy": 0.0107421875, "a_accuracy": 0.095703125, "b_accuracy": 0.095703125, "register_accuracy": 0.095703125 }, { "model": "marginal", "length": 12, "k": 1, "n": 1024, "pair_accuracy": 0.0107421875, "a_accuracy": 0.0888671875, "b_accuracy": 0.09765625, "register_accuracy": 0.09326171875 }, { "model": "marginal", "length": 12, "k": 2, "n": 1024, "pair_accuracy": 0.0087890625, "a_accuracy": 0.091796875, "b_accuracy": 0.103515625, "register_accuracy": 0.09765625 }, { "model": "marginal", "length": 12, "k": 3, "n": 1024, "pair_accuracy": 0.005859375, "a_accuracy": 0.091796875, "b_accuracy": 0.1025390625, "register_accuracy": 0.09716796875 }, { "model": "marginal", "length": 12, "k": 6, "n": 1024, "pair_accuracy": 0.0224609375, "a_accuracy": 0.119140625, "b_accuracy": 0.1240234375, "register_accuracy": 0.12158203125 }, { "model": "marginal", "length": 12, "k": 9, "n": 1024, "pair_accuracy": 0.025390625, "a_accuracy": 0.22265625, "b_accuracy": 0.2060546875, "register_accuracy": 0.21435546875 }, { "model": "marginal", "length": 12, "k": 12, "n": 1024, "pair_accuracy": 1.0, "a_accuracy": 1.0, "b_accuracy": 1.0, "register_accuracy": 1.0 } ] } ], "checkpoints": [ "large_artifacts/joint_register_executor/checkpoints/control_marginal_mod11/checkpoint_step00100.pt", "large_artifacts/joint_register_executor/checkpoints/control_marginal_mod11/checkpoint_step00200.pt", "large_artifacts/joint_register_executor/checkpoints/control_marginal_mod11/checkpoint_step00300.pt" ] }