{
  "best_metric": 0.8415273271774395,
  "best_model_checkpoint": "results_retain/facebook/hubert-base-ls960/42/checkpoint-30000",
  "epoch": 69.20415224913495,
  "eval_steps": 1000,
  "global_step": 30000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.1534025374855825,
      "grad_norm": 2.9732823371887207,
      "learning_rate": 8.333333333333333e-05,
      "loss": 3.9827,
      "step": 500
    },
    {
      "epoch": 2.306805074971165,
      "grad_norm": 3.6868040561676025,
      "learning_rate": 0.00016666666666666666,
      "loss": 3.3559,
      "step": 1000
    },
    {
      "epoch": 2.306805074971165,
      "eval_accuracy": 0.33953082106313953,
      "eval_f1_macro": 0.12928496278744922,
      "eval_loss": 2.5730652809143066,
      "eval_runtime": 35.334,
      "eval_samples_per_second": 226.807,
      "eval_steps_per_second": 7.104,
      "step": 1000
    },
    {
      "epoch": 3.4602076124567476,
      "grad_norm": 7.014188766479492,
      "learning_rate": 0.00025,
      "loss": 2.1949,
      "step": 1500
    },
    {
      "epoch": 4.61361014994233,
      "grad_norm": 6.496627330780029,
      "learning_rate": 0.0003333333333333333,
      "loss": 1.6389,
      "step": 2000
    },
    {
      "epoch": 4.61361014994233,
      "eval_accuracy": 0.6416271524831545,
      "eval_f1_macro": 0.4299797469182877,
      "eval_loss": 1.4779495000839233,
      "eval_runtime": 35.3694,
      "eval_samples_per_second": 226.58,
      "eval_steps_per_second": 7.097,
      "step": 2000
    },
    {
      "epoch": 5.767012687427912,
      "grad_norm": 7.616945266723633,
      "learning_rate": 0.0004166666666666667,
      "loss": 1.4162,
      "step": 2500
    },
    {
      "epoch": 6.920415224913495,
      "grad_norm": 8.488947868347168,
      "learning_rate": 0.0005,
      "loss": 1.3587,
      "step": 3000
    },
    {
      "epoch": 6.920415224913495,
      "eval_accuracy": 0.6574744197654105,
      "eval_f1_macro": 0.4595491108495356,
      "eval_loss": 1.427338719367981,
      "eval_runtime": 72.7464,
      "eval_samples_per_second": 110.164,
      "eval_steps_per_second": 3.45,
      "step": 3000
    },
    {
      "epoch": 8.073817762399077,
      "grad_norm": 8.156586647033691,
      "learning_rate": 0.0004907407407407408,
      "loss": 1.3027,
      "step": 3500
    },
    {
      "epoch": 9.22722029988466,
      "grad_norm": 6.934875965118408,
      "learning_rate": 0.00048148148148148144,
      "loss": 1.1695,
      "step": 4000
    },
    {
      "epoch": 9.22722029988466,
      "eval_accuracy": 0.6961567257299726,
      "eval_f1_macro": 0.5353451961331095,
      "eval_loss": 1.2435524463653564,
      "eval_runtime": 70.2906,
      "eval_samples_per_second": 114.012,
      "eval_steps_per_second": 3.571,
      "step": 4000
    },
    {
      "epoch": 10.380622837370241,
      "grad_norm": 7.151013374328613,
      "learning_rate": 0.00047222222222222224,
      "loss": 1.0695,
      "step": 4500
    },
    {
      "epoch": 11.534025374855824,
      "grad_norm": 7.771185874938965,
      "learning_rate": 0.000462962962962963,
      "loss": 0.9787,
      "step": 5000
    },
    {
      "epoch": 11.534025374855824,
      "eval_accuracy": 0.6901672073870726,
      "eval_f1_macro": 0.5134333803516367,
      "eval_loss": 1.3313419818878174,
      "eval_runtime": 71.0472,
      "eval_samples_per_second": 112.798,
      "eval_steps_per_second": 3.533,
      "step": 5000
    },
    {
      "epoch": 12.687427912341407,
      "grad_norm": 6.754736423492432,
      "learning_rate": 0.0004537037037037037,
      "loss": 0.9056,
      "step": 5500
    },
    {
      "epoch": 13.84083044982699,
      "grad_norm": 6.322958946228027,
      "learning_rate": 0.0004444444444444444,
      "loss": 0.836,
      "step": 6000
    },
    {
      "epoch": 13.84083044982699,
      "eval_accuracy": 0.7138757174943848,
      "eval_f1_macro": 0.5729793747297807,
      "eval_loss": 1.2415224313735962,
      "eval_runtime": 73.4693,
      "eval_samples_per_second": 109.08,
      "eval_steps_per_second": 3.416,
      "step": 6000
    },
    {
      "epoch": 14.994232987312571,
      "grad_norm": 7.2826619148254395,
      "learning_rate": 0.0004351851851851852,
      "loss": 0.7867,
      "step": 6500
    },
    {
      "epoch": 16.147635524798154,
      "grad_norm": 5.9969482421875,
      "learning_rate": 0.00042592592592592595,
      "loss": 0.7135,
      "step": 7000
    },
    {
      "epoch": 16.147635524798154,
      "eval_accuracy": 0.7389568255552783,
      "eval_f1_macro": 0.5793519253003285,
      "eval_loss": 1.1902633905410767,
      "eval_runtime": 73.1754,
      "eval_samples_per_second": 109.518,
      "eval_steps_per_second": 3.43,
      "step": 7000
    },
    {
      "epoch": 17.301038062283737,
      "grad_norm": 6.838934421539307,
      "learning_rate": 0.0004166666666666667,
      "loss": 0.6719,
      "step": 7500
    },
    {
      "epoch": 18.45444059976932,
      "grad_norm": 7.784801006317139,
      "learning_rate": 0.0004074074074074074,
      "loss": 0.6009,
      "step": 8000
    },
    {
      "epoch": 18.45444059976932,
      "eval_accuracy": 0.7414524581981532,
      "eval_f1_macro": 0.6081990369390977,
      "eval_loss": 1.2159614562988281,
      "eval_runtime": 55.5881,
      "eval_samples_per_second": 144.168,
      "eval_steps_per_second": 4.515,
      "step": 8000
    },
    {
      "epoch": 19.607843137254903,
      "grad_norm": 6.171660423278809,
      "learning_rate": 0.0003981481481481481,
      "loss": 0.5756,
      "step": 8500
    },
    {
      "epoch": 20.761245674740483,
      "grad_norm": 7.959474563598633,
      "learning_rate": 0.0003888888888888889,
      "loss": 0.5355,
      "step": 9000
    },
    {
      "epoch": 20.761245674740483,
      "eval_accuracy": 0.7543049663089593,
      "eval_f1_macro": 0.5947199785680519,
      "eval_loss": 1.1460059881210327,
      "eval_runtime": 55.0321,
      "eval_samples_per_second": 145.624,
      "eval_steps_per_second": 4.561,
      "step": 9000
    },
    {
      "epoch": 21.914648212226066,
      "grad_norm": 7.287164211273193,
      "learning_rate": 0.00037962962962962966,
      "loss": 0.5046,
      "step": 9500
    },
    {
      "epoch": 23.06805074971165,
      "grad_norm": 6.368403434753418,
      "learning_rate": 0.00037037037037037035,
      "loss": 0.4737,
      "step": 10000
    },
    {
      "epoch": 23.06805074971165,
      "eval_accuracy": 0.759920139755428,
      "eval_f1_macro": 0.6214937740706044,
      "eval_loss": 1.1644535064697266,
      "eval_runtime": 54.1187,
      "eval_samples_per_second": 148.082,
      "eval_steps_per_second": 4.638,
      "step": 10000
    },
    {
      "epoch": 24.22145328719723,
      "grad_norm": 9.294144630432129,
      "learning_rate": 0.0003611111111111111,
      "loss": 0.4349,
      "step": 10500
    },
    {
      "epoch": 25.374855824682815,
      "grad_norm": 7.4235310554504395,
      "learning_rate": 0.0003518518518518519,
      "loss": 0.4352,
      "step": 11000
    },
    {
      "epoch": 25.374855824682815,
      "eval_accuracy": 0.7545545295732469,
      "eval_f1_macro": 0.5917892398903293,
      "eval_loss": 1.213472843170166,
      "eval_runtime": 54.2317,
      "eval_samples_per_second": 147.773,
      "eval_steps_per_second": 4.628,
      "step": 11000
    },
    {
      "epoch": 26.528258362168398,
      "grad_norm": 5.937560558319092,
      "learning_rate": 0.00034259259259259263,
      "loss": 0.4017,
      "step": 11500
    },
    {
      "epoch": 27.68166089965398,
      "grad_norm": 6.036593914031982,
      "learning_rate": 0.0003333333333333333,
      "loss": 0.3652,
      "step": 12000
    },
    {
      "epoch": 27.68166089965398,
      "eval_accuracy": 0.7732717743948091,
      "eval_f1_macro": 0.6375373960767734,
      "eval_loss": 1.1644330024719238,
      "eval_runtime": 75.4416,
      "eval_samples_per_second": 106.228,
      "eval_steps_per_second": 3.327,
      "step": 12000
    },
    {
      "epoch": 28.83506343713956,
      "grad_norm": 6.821892738342285,
      "learning_rate": 0.00032407407407407406,
      "loss": 0.3443,
      "step": 12500
    },
    {
      "epoch": 29.988465974625143,
      "grad_norm": 4.1507463455200195,
      "learning_rate": 0.0003148148148148148,
      "loss": 0.3246,
      "step": 13000
    },
    {
      "epoch": 29.988465974625143,
      "eval_accuracy": 0.7776391315198403,
      "eval_f1_macro": 0.6181477901694947,
      "eval_loss": 1.143282175064087,
      "eval_runtime": 75.3981,
      "eval_samples_per_second": 106.289,
      "eval_steps_per_second": 3.329,
      "step": 13000
    },
    {
      "epoch": 31.141868512110726,
      "grad_norm": 7.311563491821289,
      "learning_rate": 0.0003055555555555556,
      "loss": 0.3082,
      "step": 13500
    },
    {
      "epoch": 32.29527104959631,
      "grad_norm": 2.214399576187134,
      "learning_rate": 0.0002962962962962963,
      "loss": 0.2876,
      "step": 14000
    },
    {
      "epoch": 32.29527104959631,
      "eval_accuracy": 0.7700274519590716,
      "eval_f1_macro": 0.6278465966595438,
      "eval_loss": 1.2212963104248047,
      "eval_runtime": 73.2438,
      "eval_samples_per_second": 109.415,
      "eval_steps_per_second": 3.427,
      "step": 14000
    },
    {
      "epoch": 33.44867358708189,
      "grad_norm": 5.876758575439453,
      "learning_rate": 0.00028703703703703703,
      "loss": 0.2722,
      "step": 14500
    },
    {
      "epoch": 34.602076124567475,
      "grad_norm": 3.34192156791687,
      "learning_rate": 0.0002777777777777778,
      "loss": 0.2539,
      "step": 15000
    },
    {
      "epoch": 34.602076124567475,
      "eval_accuracy": 0.7858747192413277,
      "eval_f1_macro": 0.6310309906248334,
      "eval_loss": 1.1600251197814941,
      "eval_runtime": 55.43,
      "eval_samples_per_second": 144.579,
      "eval_steps_per_second": 4.528,
      "step": 15000
    },
    {
      "epoch": 35.75547866205306,
      "grad_norm": 4.611924648284912,
      "learning_rate": 0.0002685185185185186,
      "loss": 0.2428,
      "step": 15500
    },
    {
      "epoch": 36.90888119953864,
      "grad_norm": 3.3283474445343018,
      "learning_rate": 0.00025925925925925926,
      "loss": 0.2322,
      "step": 16000
    },
    {
      "epoch": 36.90888119953864,
      "eval_accuracy": 0.7816321437484403,
      "eval_f1_macro": 0.6319203799590871,
      "eval_loss": 1.135780930519104,
      "eval_runtime": 63.9861,
      "eval_samples_per_second": 125.246,
      "eval_steps_per_second": 3.923,
      "step": 16000
    },
    {
      "epoch": 38.062283737024224,
      "grad_norm": 6.641352653503418,
      "learning_rate": 0.00025,
      "loss": 0.2146,
      "step": 16500
    },
    {
      "epoch": 39.21568627450981,
      "grad_norm": 6.247890949249268,
      "learning_rate": 0.00024074074074074072,
      "loss": 0.2003,
      "step": 17000
    },
    {
      "epoch": 39.21568627450981,
      "eval_accuracy": 0.7962315947092587,
      "eval_f1_macro": 0.6542244286445125,
      "eval_loss": 1.150564432144165,
      "eval_runtime": 69.2204,
      "eval_samples_per_second": 115.775,
      "eval_steps_per_second": 3.626,
      "step": 17000
    },
    {
      "epoch": 40.36908881199539,
      "grad_norm": 4.363713264465332,
      "learning_rate": 0.0002314814814814815,
      "loss": 0.1947,
      "step": 17500
    },
    {
      "epoch": 41.522491349480966,
      "grad_norm": 3.4260287284851074,
      "learning_rate": 0.0002222222222222222,
      "loss": 0.1794,
      "step": 18000
    },
    {
      "epoch": 41.522491349480966,
      "eval_accuracy": 0.7979785375592713,
      "eval_f1_macro": 0.6796792585107833,
      "eval_loss": 1.1864490509033203,
      "eval_runtime": 71.1001,
      "eval_samples_per_second": 112.714,
      "eval_steps_per_second": 3.53,
      "step": 18000
    },
    {
      "epoch": 42.67589388696655,
      "grad_norm": 2.606008291244507,
      "learning_rate": 0.00021296296296296298,
      "loss": 0.1689,
      "step": 18500
    },
    {
      "epoch": 43.82929642445213,
      "grad_norm": 4.665687084197998,
      "learning_rate": 0.0002037037037037037,
      "loss": 0.1645,
      "step": 19000
    },
    {
      "epoch": 43.82929642445213,
      "eval_accuracy": 0.80059895183429,
      "eval_f1_macro": 0.6667890419585701,
      "eval_loss": 1.2014110088348389,
      "eval_runtime": 74.7112,
      "eval_samples_per_second": 107.266,
      "eval_steps_per_second": 3.36,
      "step": 19000
    },
    {
      "epoch": 44.982698961937714,
      "grad_norm": 3.0378100872039795,
      "learning_rate": 0.00019444444444444446,
      "loss": 0.1602,
      "step": 19500
    },
    {
      "epoch": 46.1361014994233,
      "grad_norm": 5.274627685546875,
      "learning_rate": 0.00018518518518518518,
      "loss": 0.144,
      "step": 20000
    },
    {
      "epoch": 46.1361014994233,
      "eval_accuracy": 0.7989767906164212,
      "eval_f1_macro": 0.6582157335341974,
      "eval_loss": 1.1411352157592773,
      "eval_runtime": 73.2065,
      "eval_samples_per_second": 109.471,
      "eval_steps_per_second": 3.429,
      "step": 20000
    },
    {
      "epoch": 47.28950403690888,
      "grad_norm": 2.336925745010376,
      "learning_rate": 0.00017592592592592595,
      "loss": 0.1368,
      "step": 20500
    },
    {
      "epoch": 48.44290657439446,
      "grad_norm": 2.7309417724609375,
      "learning_rate": 0.00016666666666666666,
      "loss": 0.1298,
      "step": 21000
    },
    {
      "epoch": 48.44290657439446,
      "eval_accuracy": 0.8064636885450461,
      "eval_f1_macro": 0.6782237618476237,
      "eval_loss": 1.1389836072921753,
      "eval_runtime": 73.3032,
      "eval_samples_per_second": 109.327,
      "eval_steps_per_second": 3.424,
      "step": 21000
    },
    {
      "epoch": 49.596309111880046,
      "grad_norm": 2.79067325592041,
      "learning_rate": 0.0001574074074074074,
      "loss": 0.1206,
      "step": 21500
    },
    {
      "epoch": 50.74971164936563,
      "grad_norm": 4.826747417449951,
      "learning_rate": 0.00014814814814814815,
      "loss": 0.1175,
      "step": 22000
    },
    {
      "epoch": 50.74971164936563,
      "eval_accuracy": 0.8068380334414774,
      "eval_f1_macro": 0.6700916407139905,
      "eval_loss": 1.2090946435928345,
      "eval_runtime": 73.8226,
      "eval_samples_per_second": 108.557,
      "eval_steps_per_second": 3.4,
      "step": 22000
    },
    {
      "epoch": 51.90311418685121,
      "grad_norm": 3.403858184814453,
      "learning_rate": 0.0001388888888888889,
      "loss": 0.1021,
      "step": 22500
    },
    {
      "epoch": 53.056516724336795,
      "grad_norm": 5.1802496910095215,
      "learning_rate": 0.00012962962962962963,
      "loss": 0.0977,
      "step": 23000
    },
    {
      "epoch": 53.056516724336795,
      "eval_accuracy": 0.8149488395308211,
      "eval_f1_macro": 0.682806558028361,
      "eval_loss": 1.1759377717971802,
      "eval_runtime": 73.8391,
      "eval_samples_per_second": 108.533,
      "eval_steps_per_second": 3.399,
      "step": 23000
    },
    {
      "epoch": 54.20991926182238,
      "grad_norm": 1.122316837310791,
      "learning_rate": 0.00012037037037037036,
      "loss": 0.0912,
      "step": 23500
    },
    {
      "epoch": 55.36332179930796,
      "grad_norm": 1.1100833415985107,
      "learning_rate": 0.0001111111111111111,
      "loss": 0.0823,
      "step": 24000
    },
    {
      "epoch": 55.36332179930796,
      "eval_accuracy": 0.8166957823808335,
      "eval_f1_macro": 0.7045678569443168,
      "eval_loss": 1.2304565906524658,
      "eval_runtime": 74.7299,
      "eval_samples_per_second": 107.24,
      "eval_steps_per_second": 3.359,
      "step": 24000
    },
    {
      "epoch": 56.516724336793544,
      "grad_norm": 4.60992956161499,
      "learning_rate": 0.00010185185185185185,
      "loss": 0.0873,
      "step": 24500
    },
    {
      "epoch": 57.67012687427912,
      "grad_norm": 5.945472240447998,
      "learning_rate": 9.259259259259259e-05,
      "loss": 0.0767,
      "step": 25000
    },
    {
      "epoch": 57.67012687427912,
      "eval_accuracy": 0.8238083354130272,
      "eval_f1_macro": 0.6889311414034964,
      "eval_loss": 1.231188416481018,
      "eval_runtime": 72.9471,
      "eval_samples_per_second": 109.86,
      "eval_steps_per_second": 3.441,
      "step": 25000
    },
    {
      "epoch": 58.8235294117647,
      "grad_norm": 6.51999044418335,
      "learning_rate": 8.333333333333333e-05,
      "loss": 0.0667,
      "step": 25500
    },
    {
      "epoch": 59.976931949250286,
      "grad_norm": 3.6006715297698975,
      "learning_rate": 7.407407407407407e-05,
      "loss": 0.066,
      "step": 26000
    },
    {
      "epoch": 59.976931949250286,
      "eval_accuracy": 0.8235587721487397,
      "eval_f1_macro": 0.7127282615425515,
      "eval_loss": 1.212782621383667,
      "eval_runtime": 70.4845,
      "eval_samples_per_second": 113.699,
      "eval_steps_per_second": 3.561,
      "step": 26000
    },
    {
      "epoch": 61.13033448673587,
      "grad_norm": 1.102469563484192,
      "learning_rate": 6.481481481481482e-05,
      "loss": 0.0601,
      "step": 26500
    },
    {
      "epoch": 62.28373702422145,
      "grad_norm": 3.476552724838257,
      "learning_rate": 5.555555555555555e-05,
      "loss": 0.0493,
      "step": 27000
    },
    {
      "epoch": 62.28373702422145,
      "eval_accuracy": 0.8310456700773646,
      "eval_f1_macro": 0.7115209260308665,
      "eval_loss": 1.15741765499115,
      "eval_runtime": 74.8395,
      "eval_samples_per_second": 107.083,
      "eval_steps_per_second": 3.354,
      "step": 27000
    },
    {
      "epoch": 63.437139561707035,
      "grad_norm": 0.26378124952316284,
      "learning_rate": 4.6296296296296294e-05,
      "loss": 0.0527,
      "step": 27500
    },
    {
      "epoch": 64.59054209919262,
      "grad_norm": 1.6174193620681763,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.0479,
      "step": 28000
    },
    {
      "epoch": 64.59054209919262,
      "eval_accuracy": 0.836785625155977,
      "eval_f1_macro": 0.7171903480510493,
      "eval_loss": 1.1416091918945312,
      "eval_runtime": 72.9274,
      "eval_samples_per_second": 109.89,
      "eval_steps_per_second": 3.442,
      "step": 28000
    },
    {
      "epoch": 65.7439446366782,
      "grad_norm": 1.4499250650405884,
      "learning_rate": 2.7777777777777776e-05,
      "loss": 0.0453,
      "step": 28500
    },
    {
      "epoch": 66.89734717416378,
      "grad_norm": 3.988093614578247,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.0389,
      "step": 29000
    },
    {
      "epoch": 66.89734717416378,
      "eval_accuracy": 0.8370351884202646,
      "eval_f1_macro": 0.7212408780468642,
      "eval_loss": 1.1253269910812378,
      "eval_runtime": 53.9632,
      "eval_samples_per_second": 148.509,
      "eval_steps_per_second": 4.651,
      "step": 29000
    },
    {
      "epoch": 68.05074971164936,
      "grad_norm": 2.909609317779541,
      "learning_rate": 9.259259259259259e-06,
      "loss": 0.0433,
      "step": 29500
    },
    {
      "epoch": 69.20415224913495,
      "grad_norm": 2.1668026447296143,
      "learning_rate": 0.0,
      "loss": 0.0343,
      "step": 30000
    },
    {
      "epoch": 69.20415224913495,
      "eval_accuracy": 0.8415273271774395,
      "eval_f1_macro": 0.7162084790077747,
      "eval_loss": 1.1328068971633911,
      "eval_runtime": 73.3281,
      "eval_samples_per_second": 109.29,
      "eval_steps_per_second": 3.423,
      "step": 30000
    }
  ],
  "logging_steps": 500,
  "max_steps": 30000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 70,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.7427529644770302e+20,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}