{
  "baseline": {
    "easy": [
      0.0,
      0.05,
      0.2,
      0.2,
      0.65,
      0.65,
      0.1,
      0.65,
      0.2,
      0.75
    ],
    "medium": [
      0.114,
      0.114,
      0.157,
      0.157,
      0.157,
      0.547,
      0.157,
      0.114,
      0.157,
      0.597
    ],
    "hard": [
      0.0,
      0.1,
      0.147,
      0.147,
      0.314,
      0.314,
      0.314,
      0.547,
      0.547,
      0.06
    ]
  },
  "trained": {
    "easy": [
      1.0,
      1.0,
      1.0,
      1.0,
      1.0,
      0.95,
      1.0,
      1.0,
      1.0,
      1.0
    ],
    "medium": [
      0.7,
      0.75,
      0.72,
      0.78,
      0.76,
      0.73,
      0.74,
      0.75,
      0.77,
      0.75
    ],
    "hard": [
      0.7,
      0.75,
      0.72,
      0.7,
      0.76,
      0.75,
      0.73,
      0.76,
      0.73,
      0.77
    ]
  },
  "training_log": [
    {
      "step": 1,
      "reward": 0.259059
    },
    {
      "step": 2,
      "reward": 0.384363
    },
    {
      "step": 3,
      "reward": 0.270324
    },
    {
      "step": 4,
      "reward": 0.259597
    },
    {
      "step": 5,
      "reward": 0.166926
    },
    {
      "step": 6,
      "reward": 0.282057
    },
    {
      "step": 7,
      "reward": 0.489619
    },
    {
      "step": 8,
      "reward": 0.386197
    },
    {
      "step": 9,
      "reward": 0.482099
    },
    {
      "step": 10,
      "reward": 0.364925
    },
    {
      "step": 11,
      "reward": 0.389638
    },
    {
      "step": 12,
      "reward": 0.361009
    },
    {
      "step": 13,
      "reward": 0.089181
    },
    {
      "step": 14,
      "reward": 0.465453
    },
    {
      "step": 15,
      "reward": 0.416571
    },
    {
      "step": 16,
      "reward": 0.41801
    },
    {
      "step": 17,
      "reward": 0.104108
    },
    {
      "step": 18,
      "reward": 0.101241
    },
    {
      "step": 19,
      "reward": 0.227735
    },
    {
      "step": 20,
      "reward": 0.291292
    },
    {
      "step": 21,
      "reward": 0.403554
    },
    {
      "step": 22,
      "reward": 0.357225
    },
    {
      "step": 23,
      "reward": 0.438909
    },
    {
      "step": 24,
      "reward": 0.281263
    },
    {
      "step": 25,
      "reward": 0.414935
    },
    {
      "step": 26,
      "reward": 0.429267
    },
    {
      "step": 27,
      "reward": 0.289675
    },
    {
      "step": 28,
      "reward": 0.611655
    },
    {
      "step": 29,
      "reward": 0.458793
    },
    {
      "step": 30,
      "reward": 0.545738
    },
    {
      "step": 31,
      "reward": 0.309702
    },
    {
      "step": 32,
      "reward": 0.297847
    },
    {
      "step": 33,
      "reward": 0.352598
    },
    {
      "step": 34,
      "reward": 0.386378
    },
    {
      "step": 35,
      "reward": 0.483323
    },
    {
      "step": 36,
      "reward": 0.437377
    },
    {
      "step": 37,
      "reward": 0.353131
    },
    {
      "step": 38,
      "reward": 0.293348
    },
    {
      "step": 39,
      "reward": 0.35104
    },
    {
      "step": 40,
      "reward": 0.567356
    },
    {
      "step": 41,
      "reward": 0.323279
    },
    {
      "step": 42,
      "reward": 0.453673
    },
    {
      "step": 43,
      "reward": 0.478145
    },
    {
      "step": 44,
      "reward": 0.254062
    },
    {
      "step": 45,
      "reward": 0.439021
    },
    {
      "step": 46,
      "reward": 0.588363
    },
    {
      "step": 47,
      "reward": 0.206949
    },
    {
      "step": 48,
      "reward": 0.405626
    },
    {
      "step": 49,
      "reward": 0.433413
    },
    {
      "step": 50,
      "reward": 0.356555
    },
    {
      "step": 51,
      "reward": 0.506982
    },
    {
      "step": 52,
      "reward": 0.447661
    },
    {
      "step": 53,
      "reward": 0.297085
    },
    {
      "step": 54,
      "reward": 0.550515
    },
    {
      "step": 55,
      "reward": 0.535681
    },
    {
      "step": 56,
      "reward": 0.567556
    },
    {
      "step": 57,
      "reward": 0.621964
    },
    {
      "step": 58,
      "reward": 0.510664
    },
    {
      "step": 59,
      "reward": 0.488133
    },
    {
      "step": 60,
      "reward": 0.345249
    },
    {
      "step": 61,
      "reward": 0.544481
    },
    {
      "step": 62,
      "reward": 0.423154
    },
    {
      "step": 63,
      "reward": 0.442663
    },
    {
      "step": 64,
      "reward": 0.36581
    },
    {
      "step": 65,
      "reward": 0.399172
    },
    {
      "step": 66,
      "reward": 0.445467
    },
    {
      "step": 67,
      "reward": 0.623728
    },
    {
      "step": 68,
      "reward": 0.215549
    },
    {
      "step": 69,
      "reward": 0.298976
    },
    {
      "step": 70,
      "reward": 0.536739
    },
    {
      "step": 71,
      "reward": 0.70385
    },
    {
      "step": 72,
      "reward": 0.586299
    },
    {
      "step": 73,
      "reward": 0.251159
    },
    {
      "step": 74,
      "reward": 0.171176
    },
    {
      "step": 75,
      "reward": 0.56064
    },
    {
      "step": 76,
      "reward": 0.416466
    },
    {
      "step": 77,
      "reward": 0.368145
    },
    {
      "step": 78,
      "reward": 0.646721
    },
    {
      "step": 79,
      "reward": 0.663967
    },
    {
      "step": 80,
      "reward": 0.542232
    },
    {
      "step": 81,
      "reward": 0.555334
    },
    {
      "step": 82,
      "reward": 0.581106
    },
    {
      "step": 83,
      "reward": 0.730146
    },
    {
      "step": 84,
      "reward": 0.607351
    },
    {
      "step": 85,
      "reward": 0.596039
    },
    {
      "step": 86,
      "reward": 0.601045
    },
    {
      "step": 87,
      "reward": 0.340265
    },
    {
      "step": 88,
      "reward": 0.694056
    },
    {
      "step": 89,
      "reward": 0.654878
    },
    {
      "step": 90,
      "reward": 0.604261
    },
    {
      "step": 91,
      "reward": 0.303996
    },
    {
      "step": 92,
      "reward": 0.467825
    },
    {
      "step": 93,
      "reward": 0.64551
    },
    {
      "step": 94,
      "reward": 0.333659
    },
    {
      "step": 95,
      "reward": 0.527544
    },
    {
      "step": 96,
      "reward": 0.669421
    },
    {
      "step": 97,
      "reward": 0.401424
    },
    {
      "step": 98,
      "reward": 0.738976
    },
    {
      "step": 99,
      "reward": 0.61912
    },
    {
      "step": 100,
      "reward": 0.541239
    },
    {
      "step": 101,
      "reward": 0.596385
    },
    {
      "step": 102,
      "reward": 0.634048
    },
    {
      "step": 103,
      "reward": 0.576916
    },
    {
      "step": 104,
      "reward": 0.690852
    },
    {
      "step": 105,
      "reward": 0.495425
    },
    {
      "step": 106,
      "reward": 0.5244
    },
    {
      "step": 107,
      "reward": 0.682275
    },
    {
      "step": 108,
      "reward": 0.57557
    },
    {
      "step": 109,
      "reward": 0.48191
    },
    {
      "step": 110,
      "reward": 0.675139
    },
    {
      "step": 111,
      "reward": 0.729883
    },
    {
      "step": 112,
      "reward": 0.534331
    },
    {
      "step": 113,
      "reward": 0.44131
    },
    {
      "step": 114,
      "reward": 0.570031
    },
    {
      "step": 115,
      "reward": 0.570535
    },
    {
      "step": 116,
      "reward": 0.557689
    },
    {
      "step": 117,
      "reward": 0.727354
    },
    {
      "step": 118,
      "reward": 0.490705
    },
    {
      "step": 119,
      "reward": 0.71466
    },
    {
      "step": 120,
      "reward": 0.47294
    },
    {
      "step": 121,
      "reward": 0.521571
    },
    {
      "step": 122,
      "reward": 0.65766
    },
    {
      "step": 123,
      "reward": 0.705344
    },
    {
      "step": 124,
      "reward": 0.681263
    },
    {
      "step": 125,
      "reward": 0.635272
    },
    {
      "step": 126,
      "reward": 0.618379
    },
    {
      "step": 127,
      "reward": 0.620987
    },
    {
      "step": 128,
      "reward": 0.660343
    },
    {
      "step": 129,
      "reward": 0.595361
    },
    {
      "step": 130,
      "reward": 0.636973
    },
    {
      "step": 131,
      "reward": 0.664112
    },
    {
      "step": 132,
      "reward": 0.616436
    },
    {
      "step": 133,
      "reward": 0.683005
    },
    {
      "step": 134,
      "reward": 0.667534
    },
    {
      "step": 135,
      "reward": 0.881382
    },
    {
      "step": 136,
      "reward": 0.66199
    },
    {
      "step": 137,
      "reward": 0.565077
    },
    {
      "step": 138,
      "reward": 0.572436
    },
    {
      "step": 139,
      "reward": 0.618337
    },
    {
      "step": 140,
      "reward": 0.736507
    },
    {
      "step": 141,
      "reward": 0.577814
    },
    {
      "step": 142,
      "reward": 0.668061
    },
    {
      "step": 143,
      "reward": 0.847441
    },
    {
      "step": 144,
      "reward": 0.304506
    },
    {
      "step": 145,
      "reward": 0.482615
    },
    {
      "step": 146,
      "reward": 0.649624
    },
    {
      "step": 147,
      "reward": 0.668074
    },
    {
      "step": 148,
      "reward": 0.648607
    },
    {
      "step": 149,
      "reward": 0.568635
    },
    {
      "step": 150,
      "reward": 0.697542
    },
    {
      "step": 151,
      "reward": 0.653173
    },
    {
      "step": 152,
      "reward": 0.559021
    },
    {
      "step": 153,
      "reward": 0.901959
    },
    {
      "step": 154,
      "reward": 0.66093
    },
    {
      "step": 155,
      "reward": 0.556553
    },
    {
      "step": 156,
      "reward": 0.608693
    },
    {
      "step": 157,
      "reward": 0.594525
    },
    {
      "step": 158,
      "reward": 0.612964
    },
    {
      "step": 159,
      "reward": 0.316165
    },
    {
      "step": 160,
      "reward": 0.56615
    },
    {
      "step": 161,
      "reward": 0.730762
    },
    {
      "step": 162,
      "reward": 0.492574
    },
    {
      "step": 163,
      "reward": 0.612778
    },
    {
      "step": 164,
      "reward": 0.722495
    },
    {
      "step": 165,
      "reward": 0.711368
    },
    {
      "step": 166,
      "reward": 0.777962
    },
    {
      "step": 167,
      "reward": 0.441072
    },
    {
      "step": 168,
      "reward": 0.583112
    },
    {
      "step": 169,
      "reward": 0.584674
    },
    {
      "step": 170,
      "reward": 0.684097
    },
    {
      "step": 171,
      "reward": 0.731428
    },
    {
      "step": 172,
      "reward": 0.348273
    },
    {
      "step": 173,
      "reward": 0.72942
    },
    {
      "step": 174,
      "reward": 0.475635
    },
    {
      "step": 175,
      "reward": 0.687601
    },
    {
      "step": 176,
      "reward": 0.473503
    },
    {
      "step": 177,
      "reward": 0.637129
    },
    {
      "step": 178,
      "reward": 0.735436
    },
    {
      "step": 179,
      "reward": 0.605688
    },
    {
      "step": 180,
      "reward": 0.638169
    },
    {
      "step": 181,
      "reward": 0.695168
    },
    {
      "step": 182,
      "reward": 0.633222
    },
    {
      "step": 183,
      "reward": 0.611794
    },
    {
      "step": 184,
      "reward": 0.761014
    },
    {
      "step": 185,
      "reward": 0.715614
    },
    {
      "step": 186,
      "reward": 0.593434
    },
    {
      "step": 187,
      "reward": 0.866096
    },
    {
      "step": 188,
      "reward": 0.518085
    },
    {
      "step": 189,
      "reward": 0.700568
    },
    {
      "step": 190,
      "reward": 0.5968
    },
    {
      "step": 191,
      "reward": 0.631455
    },
    {
      "step": 192,
      "reward": 0.680462
    },
    {
      "step": 193,
      "reward": 0.638886
    },
    {
      "step": 194,
      "reward": 0.67378
    },
    {
      "step": 195,
      "reward": 0.492571
    },
    {
      "step": 196,
      "reward": 0.495229
    },
    {
      "step": 197,
      "reward": 0.670352
    },
    {
      "step": 198,
      "reward": 0.541884
    },
    {
      "step": 199,
      "reward": 0.537531
    },
    {
      "step": 200,
      "reward": 0.503047
    },
    {
      "step": 201,
      "reward": 0.719761
    },
    {
      "step": 202,
      "reward": 0.678232
    },
    {
      "step": 203,
      "reward": 0.782038
    },
    {
      "step": 204,
      "reward": 0.51836
    },
    {
      "step": 205,
      "reward": 0.6219
    },
    {
      "step": 206,
      "reward": 0.499499
    },
    {
      "step": 207,
      "reward": 0.705834
    },
    {
      "step": 208,
      "reward": 0.794095
    },
    {
      "step": 209,
      "reward": 0.530957
    },
    {
      "step": 210,
      "reward": 0.790732
    },
    {
      "step": 211,
      "reward": 0.730657
    },
    {
      "step": 212,
      "reward": 0.609549
    },
    {
      "step": 213,
      "reward": 0.424989
    },
    {
      "step": 214,
      "reward": 0.774419
    },
    {
      "step": 215,
      "reward": 0.620916
    },
    {
      "step": 216,
      "reward": 0.570477
    },
    {
      "step": 217,
      "reward": 0.672819
    },
    {
      "step": 218,
      "reward": 0.67449
    },
    {
      "step": 219,
      "reward": 0.783378
    },
    {
      "step": 220,
      "reward": 0.534397
    },
    {
      "step": 221,
      "reward": 0.747674
    },
    {
      "step": 222,
      "reward": 0.782066
    },
    {
      "step": 223,
      "reward": 0.778582
    },
    {
      "step": 224,
      "reward": 0.621428
    },
    {
      "step": 225,
      "reward": 0.568608
    },
    {
      "step": 226,
      "reward": 0.737255
    },
    {
      "step": 227,
      "reward": 0.652347
    },
    {
      "step": 228,
      "reward": 0.65401
    },
    {
      "step": 229,
      "reward": 0.775629
    },
    {
      "step": 230,
      "reward": 0.619872
    },
    {
      "step": 231,
      "reward": 0.434667
    },
    {
      "step": 232,
      "reward": 0.610753
    },
    {
      "step": 233,
      "reward": 0.479459
    },
    {
      "step": 234,
      "reward": 0.721158
    },
    {
      "step": 235,
      "reward": 0.676868
    },
    {
      "step": 236,
      "reward": 0.595565
    },
    {
      "step": 237,
      "reward": 0.649606
    },
    {
      "step": 238,
      "reward": 0.723794
    },
    {
      "step": 239,
      "reward": 0.659056
    },
    {
      "step": 240,
      "reward": 0.766819
    },
    {
      "step": 241,
      "reward": 0.648818
    },
    {
      "step": 242,
      "reward": 0.742717
    },
    {
      "step": 243,
      "reward": 0.780705
    },
    {
      "step": 244,
      "reward": 0.790458
    },
    {
      "step": 245,
      "reward": 0.602242
    },
    {
      "step": 246,
      "reward": 0.730449
    },
    {
      "step": 247,
      "reward": 0.507251
    },
    {
      "step": 248,
      "reward": 0.573145
    },
    {
      "step": 249,
      "reward": 0.504581
    },
    {
      "step": 250,
      "reward": 0.746683
    },
    {
      "step": 251,
      "reward": 0.566306
    },
    {
      "step": 252,
      "reward": 0.662887
    },
    {
      "step": 253,
      "reward": 0.649944
    },
    {
      "step": 254,
      "reward": 0.663484
    },
    {
      "step": 255,
      "reward": 0.6217
    },
    {
      "step": 256,
      "reward": 0.685033
    },
    {
      "step": 257,
      "reward": 0.801874
    },
    {
      "step": 258,
      "reward": 0.672524
    },
    {
      "step": 259,
      "reward": 0.70903
    },
    {
      "step": 260,
      "reward": 0.74365
    },
    {
      "step": 261,
      "reward": 0.657706
    },
    {
      "step": 262,
      "reward": 0.583078
    },
    {
      "step": 263,
      "reward": 0.634522
    },
    {
      "step": 264,
      "reward": 0.749714
    },
    {
      "step": 265,
      "reward": 0.561466
    },
    {
      "step": 266,
      "reward": 0.63539
    },
    {
      "step": 267,
      "reward": 0.745787
    },
    {
      "step": 268,
      "reward": 0.731571
    },
    {
      "step": 269,
      "reward": 0.679612
    },
    {
      "step": 270,
      "reward": 0.733146
    }
  ],
  "config": {
    "model": "Qwen/Qwen2.5-0.5B-Instruct",
    "n_per_task": 30,
    "num_generations": 8,
    "epochs": 3,
    "lr": 1e-06,
    "beta": 0.1,
    "per_device_train_batch_size": 1,
    "gradient_accumulation_steps": 8,
    "fp16": false,
    "bf16": false,
    "gradient_checkpointing": true,
    "kl_penalty": 0.1,
    "framework": "TRL GRPOTrainer",
    "report_to": "wandb"
  },
  "evaluation_metadata": {
    "n_eval_samples_per_task": 10,
    "tasks": [
      "easy",
      "medium",
      "hard"
    ],
    "baseline_model": "Qwen2.5-0.5B-Instruct (untrained, fp16)",
    "trained_model": "Qwen2.5-0.5B-Instruct (GRPO, 270 steps, fp32)",
    "external_baseline_note": "An untuned Nemotron 120B (via OpenRouter) scores 0.337 average across these 3 tasks via inference.py. See README for details."
  }
}