{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.08,
  "eval_steps": 500,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "batch_size": 4,
      "epoch": 0,
      "step": 0,
      "tokens_per_device": 1084
    },
    {
      "epoch": 0,
      "loss_ce": 4.763055324554443,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 4.763055324554443,
      "step": 0
    },
    {
      "epoch": 0.0004,
      "grad_norm": 77.51766967773438,
      "learning_rate": 0.0,
      "loss": 4.4497,
      "step": 1
    },
    {
      "batch_size": 4,
      "epoch": 0.0004,
      "step": 1,
      "tokens_per_device": 3740
    },
    {
      "epoch": 0.0004,
      "loss_ce": 4.487368583679199,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 4.487368583679199,
      "step": 1
    },
    {
      "epoch": 0.0008,
      "grad_norm": 66.39104461669922,
      "learning_rate": 1.3333333333333336e-07,
      "loss": 4.1485,
      "step": 2
    },
    {
      "batch_size": 4,
      "epoch": 0.0008,
      "step": 2,
      "tokens_per_device": 4204
    },
    {
      "epoch": 0.0008,
      "loss_ce": 5.921818733215332,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 5.921818733215332,
      "step": 2
    },
    {
      "epoch": 0.0012,
      "grad_norm": 92.08312225341797,
      "learning_rate": 2.666666666666667e-07,
      "loss": 5.3103,
      "step": 3
    },
    {
      "batch_size": 1,
      "epoch": 0.0012,
      "step": 3,
      "tokens_per_device": 5079
    },
    {
      "epoch": 0.0012,
      "loss_ce": 4.208111763000488,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 4.208111763000488,
      "step": 3
    },
    {
      "epoch": 0.0016,
      "grad_norm": 72.99446868896484,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 3.696,
      "step": 4
    },
    {
      "batch_size": 1,
      "epoch": 0.0016,
      "step": 4,
      "tokens_per_device": 4880
    },
    {
      "epoch": 0.0016,
      "loss_ce": 2.4610111713409424,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 2.4610111713409424,
      "step": 4
    },
    {
      "epoch": 0.002,
      "grad_norm": 58.486141204833984,
      "learning_rate": 5.333333333333335e-07,
      "loss": 3.6879,
      "step": 5
    },
    {
      "batch_size": 4,
      "epoch": 0.002,
      "step": 5,
      "tokens_per_device": 6980
    },
    {
      "epoch": 0.002,
      "loss_ce": 4.51284646987915,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 4.51284646987915,
      "step": 5
    },
    {
      "epoch": 0.0024,
      "grad_norm": 71.5829086303711,
      "learning_rate": 6.666666666666667e-07,
      "loss": 4.0241,
      "step": 6
    },
    {
      "batch_size": 1,
      "epoch": 0.0024,
      "step": 6,
      "tokens_per_device": 5148
    },
    {
      "epoch": 0.0024,
      "loss_ce": 3.349120855331421,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 3.349120855331421,
      "step": 6
    },
    {
      "epoch": 0.0028,
      "grad_norm": 73.27996063232422,
      "learning_rate": 8.000000000000001e-07,
      "loss": 3.8778,
      "step": 7
    },
    {
      "batch_size": 4,
      "epoch": 0.0028,
      "step": 7,
      "tokens_per_device": 4184
    },
    {
      "epoch": 0.0028,
      "loss_ce": 4.473206520080566,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 4.473206520080566,
      "step": 7
    },
    {
      "epoch": 0.0032,
      "grad_norm": 79.25920104980469,
      "learning_rate": 9.333333333333334e-07,
      "loss": 4.4395,
      "step": 8
    },
    {
      "batch_size": 1,
      "epoch": 0.0032,
      "step": 8,
      "tokens_per_device": 4982
    },
    {
      "epoch": 0.0032,
      "loss_ce": 4.184725284576416,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 4.184725284576416,
      "step": 8
    },
    {
      "epoch": 0.0036,
      "grad_norm": 81.21875762939453,
      "learning_rate": 1.066666666666667e-06,
      "loss": 3.6885,
      "step": 9
    },
    {
      "batch_size": 4,
      "epoch": 0.0036,
      "step": 9,
      "tokens_per_device": 5652
    },
    {
      "epoch": 0.0036,
      "loss_ce": 2.0520293712615967,
      "loss_lvr": 0.7983388304710388,
      "loss_mode_switch": 0.0,
      "loss_total": 2.1318633556365967,
      "step": 9
    },
    {
      "epoch": 0.004,
      "grad_norm": 38.65834045410156,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 2.1094,
      "step": 10
    },
    {
      "batch_size": 4,
      "epoch": 0.004,
      "step": 10,
      "tokens_per_device": 4276
    },
    {
      "epoch": 0.004,
      "loss_ce": 2.0300471782684326,
      "loss_lvr": 1.0279818773269653,
      "loss_mode_switch": 0.0,
      "loss_total": 2.132845401763916,
      "step": 10
    },
    {
      "epoch": 0.0044,
      "grad_norm": 20.48471450805664,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 1.1263,
      "step": 11
    },
    {
      "batch_size": 4,
      "epoch": 0.0044,
      "step": 11,
      "tokens_per_device": 10644
    },
    {
      "epoch": 0.0044,
      "loss_ce": 3.342816114425659,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 3.342816114425659,
      "step": 11
    },
    {
      "epoch": 0.0048,
      "grad_norm": 68.77659606933594,
      "learning_rate": 1.4666666666666669e-06,
      "loss": 3.455,
      "step": 12
    },
    {
      "batch_size": 4,
      "epoch": 0.0048,
      "step": 12,
      "tokens_per_device": 4184
    },
    {
      "epoch": 0.0048,
      "loss_ce": 2.5340793132781982,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 2.5340793132781982,
      "step": 12
    },
    {
      "epoch": 0.0052,
      "grad_norm": 70.3922348022461,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 2.5678,
      "step": 13
    },
    {
      "batch_size": 4,
      "epoch": 0.0052,
      "step": 13,
      "tokens_per_device": 1376
    },
    {
      "epoch": 0.0052,
      "loss_ce": 1.3518562316894531,
      "loss_lvr": 0.9177327156066895,
      "loss_mode_switch": 0.0,
      "loss_total": 1.443629503250122,
      "step": 13
    },
    {
      "epoch": 0.0056,
      "grad_norm": 14.044103622436523,
      "learning_rate": 1.7333333333333336e-06,
      "loss": 0.7403,
      "step": 14
    },
    {
      "batch_size": 4,
      "epoch": 0.0056,
      "step": 14,
      "tokens_per_device": 10464
    },
    {
      "epoch": 0.0056,
      "loss_ce": 1.6084449291229248,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 1.6084449291229248,
      "step": 14
    },
    {
      "epoch": 0.006,
      "grad_norm": 59.4080924987793,
      "learning_rate": 1.8666666666666669e-06,
      "loss": 2.0368,
      "step": 15
    },
    {
      "batch_size": 4,
      "epoch": 0.006,
      "step": 15,
      "tokens_per_device": 8448
    },
    {
      "epoch": 0.006,
      "loss_ce": 1.2514013051986694,
      "loss_lvr": 0.978214681148529,
      "loss_mode_switch": 0.0,
      "loss_total": 1.3492227792739868,
      "step": 15
    },
    {
      "epoch": 0.0064,
      "grad_norm": 47.312156677246094,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.8061,
      "step": 16
    },
    {
      "batch_size": 4,
      "epoch": 0.0064,
      "step": 16,
      "tokens_per_device": 5664
    },
    {
      "epoch": 0.0064,
      "loss_ce": 1.6180309057235718,
      "loss_lvr": 1.1346142292022705,
      "loss_mode_switch": 0.0,
      "loss_total": 1.731492280960083,
      "step": 16
    },
    {
      "epoch": 0.0068,
      "grad_norm": 22.662246704101562,
      "learning_rate": 2.133333333333334e-06,
      "loss": 1.6662,
      "step": 17
    },
    {
      "batch_size": 4,
      "epoch": 0.0068,
      "step": 17,
      "tokens_per_device": 5580
    },
    {
      "epoch": 0.0068,
      "loss_ce": 1.115349292755127,
      "loss_lvr": 1.1785991191864014,
      "loss_mode_switch": 0.0,
      "loss_total": 1.233209252357483,
      "step": 17
    },
    {
      "epoch": 0.0072,
      "grad_norm": 35.913291931152344,
      "learning_rate": 2.266666666666667e-06,
      "loss": 1.1544,
      "step": 18
    },
    {
      "batch_size": 4,
      "epoch": 0.0072,
      "step": 18,
      "tokens_per_device": 4232
    },
    {
      "epoch": 0.0072,
      "loss_ce": 0.6951131820678711,
      "loss_lvr": 1.0979421138763428,
      "loss_mode_switch": 0.0,
      "loss_total": 0.8049073815345764,
      "step": 18
    },
    {
      "epoch": 0.0076,
      "grad_norm": 27.182315826416016,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.9491,
      "step": 19
    },
    {
      "batch_size": 4,
      "epoch": 0.0076,
      "step": 19,
      "tokens_per_device": 1616
    },
    {
      "epoch": 0.0076,
      "loss_ce": 0.9380652904510498,
      "loss_lvr": 0.9828035235404968,
      "loss_mode_switch": 0.0,
      "loss_total": 1.0363456010818481,
      "step": 19
    },
    {
      "epoch": 0.008,
      "grad_norm": 35.5374755859375,
      "learning_rate": 2.5333333333333338e-06,
      "loss": 1.0123,
      "step": 20
    },
    {
      "batch_size": 1,
      "epoch": 0.008,
      "step": 20,
      "tokens_per_device": 4863
    },
    {
      "epoch": 0.008,
      "loss_ce": 0.8586581349372864,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.8586581349372864,
      "step": 20
    },
    {
      "epoch": 0.0084,
      "grad_norm": 44.61960220336914,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.7506,
      "step": 21
    },
    {
      "batch_size": 1,
      "epoch": 0.0084,
      "step": 21,
      "tokens_per_device": 4731
    },
    {
      "epoch": 0.0084,
      "loss_ce": 0.31246820092201233,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.31246820092201233,
      "step": 21
    },
    {
      "epoch": 0.0088,
      "grad_norm": 34.55044937133789,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.417,
      "step": 22
    },
    {
      "batch_size": 4,
      "epoch": 0.0088,
      "step": 22,
      "tokens_per_device": 1424
    },
    {
      "epoch": 0.0088,
      "loss_ce": 0.7496281862258911,
      "loss_lvr": 0.9418494701385498,
      "loss_mode_switch": 0.0,
      "loss_total": 0.8438131213188171,
      "step": 22
    },
    {
      "epoch": 0.0092,
      "grad_norm": 14.349714279174805,
      "learning_rate": 2.9333333333333338e-06,
      "loss": 0.4611,
      "step": 23
    },
    {
      "batch_size": 1,
      "epoch": 0.0092,
      "step": 23,
      "tokens_per_device": 5131
    },
    {
      "epoch": 0.0092,
      "loss_ce": 0.5592010021209717,
      "loss_lvr": 0.7256385684013367,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6317648887634277,
      "step": 23
    },
    {
      "epoch": 0.0096,
      "grad_norm": 26.70309066772461,
      "learning_rate": 3.066666666666667e-06,
      "loss": 0.6382,
      "step": 24
    },
    {
      "batch_size": 1,
      "epoch": 0.0096,
      "step": 24,
      "tokens_per_device": 4771
    },
    {
      "epoch": 0.0096,
      "loss_ce": 0.8043491244316101,
      "loss_lvr": 0.2601466178894043,
      "loss_mode_switch": 0.0,
      "loss_total": 0.8303638100624084,
      "step": 24
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.766989707946777,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.4386,
      "step": 25
    },
    {
      "batch_size": 4,
      "epoch": 0.01,
      "step": 25,
      "tokens_per_device": 3784
    },
    {
      "epoch": 0.01,
      "loss_ce": 0.6543468236923218,
      "loss_lvr": 1.2464392185211182,
      "loss_mode_switch": 0.0,
      "loss_total": 0.7789907455444336,
      "step": 25
    },
    {
      "epoch": 0.0104,
      "grad_norm": 22.12204360961914,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.9474,
      "step": 26
    },
    {
      "batch_size": 4,
      "epoch": 0.0104,
      "step": 26,
      "tokens_per_device": 4260
    },
    {
      "epoch": 0.0104,
      "loss_ce": 0.27501338720321655,
      "loss_lvr": 1.6786158084869385,
      "loss_mode_switch": 0.0,
      "loss_total": 0.4428749680519104,
      "step": 26
    },
    {
      "epoch": 0.0108,
      "grad_norm": 24.201011657714844,
      "learning_rate": 3.4666666666666672e-06,
      "loss": 0.5322,
      "step": 27
    },
    {
      "batch_size": 1,
      "epoch": 0.0108,
      "step": 27,
      "tokens_per_device": 5129
    },
    {
      "epoch": 0.0108,
      "loss_ce": 0.034078195691108704,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.034078195691108704,
      "step": 27
    },
    {
      "epoch": 0.0112,
      "grad_norm": 10.610871315002441,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.5331,
      "step": 28
    },
    {
      "batch_size": 4,
      "epoch": 0.0112,
      "step": 28,
      "tokens_per_device": 4172
    },
    {
      "epoch": 0.0112,
      "loss_ce": 0.5044699311256409,
      "loss_lvr": 1.3902287483215332,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6434928178787231,
      "step": 28
    },
    {
      "epoch": 0.0116,
      "grad_norm": 15.143689155578613,
      "learning_rate": 3.7333333333333337e-06,
      "loss": 0.7138,
      "step": 29
    },
    {
      "batch_size": 1,
      "epoch": 0.0116,
      "step": 29,
      "tokens_per_device": 4816
    },
    {
      "epoch": 0.0116,
      "loss_ce": 0.2971441447734833,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2971441447734833,
      "step": 29
    },
    {
      "epoch": 0.012,
      "grad_norm": 22.0164737701416,
      "learning_rate": 3.866666666666667e-06,
      "loss": 0.4831,
      "step": 30
    },
    {
      "batch_size": 4,
      "epoch": 0.012,
      "step": 30,
      "tokens_per_device": 1384
    },
    {
      "epoch": 0.012,
      "loss_ce": 0.5713188052177429,
      "loss_lvr": 1.1278611421585083,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6841049194335938,
      "step": 30
    },
    {
      "epoch": 0.0124,
      "grad_norm": 14.98511028289795,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.7855,
      "step": 31
    },
    {
      "batch_size": 4,
      "epoch": 0.0124,
      "step": 31,
      "tokens_per_device": 12444
    },
    {
      "epoch": 0.0124,
      "loss_ce": 0.397776335477829,
      "loss_lvr": 1.2703458070755005,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5248109102249146,
      "step": 31
    },
    {
      "epoch": 0.0128,
      "grad_norm": 12.142884254455566,
      "learning_rate": 4.133333333333333e-06,
      "loss": 0.5184,
      "step": 32
    },
    {
      "batch_size": 4,
      "epoch": 0.0128,
      "step": 32,
      "tokens_per_device": 3732
    },
    {
      "epoch": 0.0128,
      "loss_ce": 0.39775240421295166,
      "loss_lvr": 1.0891906023025513,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5066714882850647,
      "step": 32
    },
    {
      "epoch": 0.0132,
      "grad_norm": 14.759183883666992,
      "learning_rate": 4.266666666666668e-06,
      "loss": 0.6992,
      "step": 33
    },
    {
      "batch_size": 4,
      "epoch": 0.0132,
      "step": 33,
      "tokens_per_device": 5292
    },
    {
      "epoch": 0.0132,
      "loss_ce": 0.15975739061832428,
      "loss_lvr": 0.8407593369483948,
      "loss_mode_switch": 0.0,
      "loss_total": 0.24383333325386047,
      "step": 33
    },
    {
      "epoch": 0.0136,
      "grad_norm": 6.823642253875732,
      "learning_rate": 4.4e-06,
      "loss": 0.1493,
      "step": 34
    },
    {
      "batch_size": 4,
      "epoch": 0.0136,
      "step": 34,
      "tokens_per_device": 6888
    },
    {
      "epoch": 0.0136,
      "loss_ce": 0.1757616102695465,
      "loss_lvr": 0.8124101161956787,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2570026218891144,
      "step": 34
    },
    {
      "epoch": 0.014,
      "grad_norm": 12.734367370605469,
      "learning_rate": 4.533333333333334e-06,
      "loss": 0.3171,
      "step": 35
    },
    {
      "batch_size": 4,
      "epoch": 0.014,
      "step": 35,
      "tokens_per_device": 4748
    },
    {
      "epoch": 0.014,
      "loss_ce": 0.1213318407535553,
      "loss_lvr": 0.9361860752105713,
      "loss_mode_switch": 0.0,
      "loss_total": 0.21495044231414795,
      "step": 35
    },
    {
      "epoch": 0.0144,
      "grad_norm": 17.364501953125,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.3528,
      "step": 36
    },
    {
      "batch_size": 4,
      "epoch": 0.0144,
      "step": 36,
      "tokens_per_device": 3356
    },
    {
      "epoch": 0.0144,
      "loss_ce": 0.2532780170440674,
      "loss_lvr": 1.1753183603286743,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3708098530769348,
      "step": 36
    },
    {
      "epoch": 0.0148,
      "grad_norm": 9.573479652404785,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.6146,
      "step": 37
    },
    {
      "batch_size": 1,
      "epoch": 0.0148,
      "step": 37,
      "tokens_per_device": 4878
    },
    {
      "epoch": 0.0148,
      "loss_ce": 0.2578398883342743,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2578398883342743,
      "step": 37
    },
    {
      "epoch": 0.0152,
      "grad_norm": 22.129724502563477,
      "learning_rate": 4.933333333333334e-06,
      "loss": 0.2532,
      "step": 38
    },
    {
      "batch_size": 1,
      "epoch": 0.0152,
      "step": 38,
      "tokens_per_device": 4872
    },
    {
      "epoch": 0.0152,
      "loss_ce": 0.055063921958208084,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.055063921958208084,
      "step": 38
    },
    {
      "epoch": 0.0156,
      "grad_norm": 7.345645427703857,
      "learning_rate": 5.0666666666666676e-06,
      "loss": 0.1407,
      "step": 39
    },
    {
      "batch_size": 1,
      "epoch": 0.0156,
      "step": 39,
      "tokens_per_device": 5096
    },
    {
      "epoch": 0.0156,
      "loss_ce": 0.014654356986284256,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.014654356986284256,
      "step": 39
    },
    {
      "epoch": 0.016,
      "grad_norm": 3.5893378257751465,
      "learning_rate": 5.2e-06,
      "loss": 0.0262,
      "step": 40
    },
    {
      "batch_size": 4,
      "epoch": 0.016,
      "step": 40,
      "tokens_per_device": 3340
    },
    {
      "epoch": 0.016,
      "loss_ce": 0.19635461270809174,
      "loss_lvr": 1.3198816776275635,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3283427953720093,
      "step": 40
    },
    {
      "epoch": 0.0164,
      "grad_norm": 11.235848426818848,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.4488,
      "step": 41
    },
    {
      "batch_size": 1,
      "epoch": 0.0164,
      "step": 41,
      "tokens_per_device": 4853
    },
    {
      "epoch": 0.0164,
      "loss_ce": 0.10926812887191772,
      "loss_lvr": 0.2216251939535141,
      "loss_mode_switch": 0.0,
      "loss_total": 0.13143065571784973,
      "step": 41
    },
    {
      "epoch": 0.0168,
      "grad_norm": 10.715582847595215,
      "learning_rate": 5.466666666666667e-06,
      "loss": 0.2114,
      "step": 42
    },
    {
      "batch_size": 4,
      "epoch": 0.0168,
      "step": 42,
      "tokens_per_device": 3760
    },
    {
      "epoch": 0.0168,
      "loss_ce": 0.39787325263023376,
      "loss_lvr": 1.2983083724975586,
      "loss_mode_switch": 0.0,
      "loss_total": 0.527704119682312,
      "step": 42
    },
    {
      "epoch": 0.0172,
      "grad_norm": 10.726153373718262,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.5693,
      "step": 43
    },
    {
      "batch_size": 1,
      "epoch": 0.0172,
      "step": 43,
      "tokens_per_device": 4869
    },
    {
      "epoch": 0.0172,
      "loss_ce": 0.131699800491333,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.131699800491333,
      "step": 43
    },
    {
      "epoch": 0.0176,
      "grad_norm": 27.599788665771484,
      "learning_rate": 5.733333333333334e-06,
      "loss": 0.1442,
      "step": 44
    },
    {
      "batch_size": 1,
      "epoch": 0.0176,
      "step": 44,
      "tokens_per_device": 4862
    },
    {
      "epoch": 0.0176,
      "loss_ce": 0.15274938941001892,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.15274938941001892,
      "step": 44
    },
    {
      "epoch": 0.018,
      "grad_norm": 12.139158248901367,
      "learning_rate": 5.8666666666666675e-06,
      "loss": 0.3985,
      "step": 45
    },
    {
      "batch_size": 1,
      "epoch": 0.018,
      "step": 45,
      "tokens_per_device": 5120
    },
    {
      "epoch": 0.018,
      "loss_ce": 0.19653871655464172,
      "loss_lvr": 0.6258155703544617,
      "loss_mode_switch": 0.0,
      "loss_total": 0.25912028551101685,
      "step": 45
    },
    {
      "epoch": 0.0184,
      "grad_norm": 15.46597957611084,
      "learning_rate": 6e-06,
      "loss": 0.3434,
      "step": 46
    },
    {
      "batch_size": 4,
      "epoch": 0.0184,
      "step": 46,
      "tokens_per_device": 11888
    },
    {
      "epoch": 0.0184,
      "loss_ce": 0.3609757125377655,
      "loss_lvr": 0.9595096111297607,
      "loss_mode_switch": 0.0,
      "loss_total": 0.4569266736507416,
      "step": 46
    },
    {
      "epoch": 0.0188,
      "grad_norm": 11.660000801086426,
      "learning_rate": 6.133333333333334e-06,
      "loss": 0.3949,
      "step": 47
    },
    {
      "batch_size": 1,
      "epoch": 0.0188,
      "step": 47,
      "tokens_per_device": 4866
    },
    {
      "epoch": 0.0188,
      "loss_ce": 0.11310536414384842,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.11310536414384842,
      "step": 47
    },
    {
      "epoch": 0.0192,
      "grad_norm": 10.13582992553711,
      "learning_rate": 6.266666666666668e-06,
      "loss": 0.2024,
      "step": 48
    },
    {
      "batch_size": 4,
      "epoch": 0.0192,
      "step": 48,
      "tokens_per_device": 3800
    },
    {
      "epoch": 0.0192,
      "loss_ce": 0.273736834526062,
      "loss_lvr": 0.9867757558822632,
      "loss_mode_switch": 0.0,
      "loss_total": 0.37241441011428833,
      "step": 48
    },
    {
      "epoch": 0.0196,
      "grad_norm": 12.182291984558105,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.5655,
      "step": 49
    },
    {
      "batch_size": 1,
      "epoch": 0.0196,
      "step": 49,
      "tokens_per_device": 4859
    },
    {
      "epoch": 0.0196,
      "loss_ce": 0.12497997283935547,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.12497997283935547,
      "step": 49
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.839667320251465,
      "learning_rate": 6.533333333333334e-06,
      "loss": 0.2457,
      "step": 50
    },
    {
      "batch_size": 4,
      "epoch": 0.02,
      "step": 50,
      "tokens_per_device": 3712
    },
    {
      "epoch": 0.02,
      "loss_ce": 0.5009855031967163,
      "loss_lvr": 1.0258513689041138,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6035706400871277,
      "step": 50
    },
    {
      "epoch": 0.0204,
      "grad_norm": 11.8434419631958,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.4575,
      "step": 51
    },
    {
      "batch_size": 4,
      "epoch": 0.0204,
      "step": 51,
      "tokens_per_device": 1980
    },
    {
      "epoch": 0.0204,
      "loss_ce": 0.315250039100647,
      "loss_lvr": 1.145516037940979,
      "loss_mode_switch": 0.0,
      "loss_total": 0.4298016428947449,
      "step": 51
    },
    {
      "epoch": 0.0208,
      "grad_norm": 12.970431327819824,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.7132,
      "step": 52
    },
    {
      "batch_size": 1,
      "epoch": 0.0208,
      "step": 52,
      "tokens_per_device": 4864
    },
    {
      "epoch": 0.0208,
      "loss_ce": 0.07356984913349152,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.07356984913349152,
      "step": 52
    },
    {
      "epoch": 0.0212,
      "grad_norm": 12.188618659973145,
      "learning_rate": 6.9333333333333344e-06,
      "loss": 0.2907,
      "step": 53
    },
    {
      "batch_size": 4,
      "epoch": 0.0212,
      "step": 53,
      "tokens_per_device": 3324
    },
    {
      "epoch": 0.0212,
      "loss_ce": 0.5515737533569336,
      "loss_lvr": 0.9107203483581543,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6426457762718201,
      "step": 53
    },
    {
      "epoch": 0.0216,
      "grad_norm": 10.85720157623291,
      "learning_rate": 7.066666666666667e-06,
      "loss": 0.4167,
      "step": 54
    },
    {
      "batch_size": 1,
      "epoch": 0.0216,
      "step": 54,
      "tokens_per_device": 6797
    },
    {
      "epoch": 0.0216,
      "loss_ce": 0.1386120766401291,
      "loss_lvr": 0.39571821689605713,
      "loss_mode_switch": 0.0,
      "loss_total": 0.1781838983297348,
      "step": 54
    },
    {
      "epoch": 0.022,
      "grad_norm": 17.198837280273438,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.4683,
      "step": 55
    },
    {
      "batch_size": 1,
      "epoch": 0.022,
      "step": 55,
      "tokens_per_device": 5560
    },
    {
      "epoch": 0.022,
      "loss_ce": 0.17444881796836853,
      "loss_lvr": 0.4491371512413025,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2193625271320343,
      "step": 55
    },
    {
      "epoch": 0.0224,
      "grad_norm": 25.210329055786133,
      "learning_rate": 7.333333333333333e-06,
      "loss": 0.1769,
      "step": 56
    },
    {
      "batch_size": 4,
      "epoch": 0.0224,
      "step": 56,
      "tokens_per_device": 3756
    },
    {
      "epoch": 0.0224,
      "loss_ce": 0.5538091659545898,
      "loss_lvr": 0.8902169466018677,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6428308486938477,
      "step": 56
    },
    {
      "epoch": 0.0228,
      "grad_norm": 12.771076202392578,
      "learning_rate": 7.4666666666666675e-06,
      "loss": 0.5339,
      "step": 57
    },
    {
      "batch_size": 4,
      "epoch": 0.0228,
      "step": 57,
      "tokens_per_device": 4432
    },
    {
      "epoch": 0.0228,
      "loss_ce": 0.16717252135276794,
      "loss_lvr": 0.7458582520484924,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2417583465576172,
      "step": 57
    },
    {
      "epoch": 0.0232,
      "grad_norm": 18.2379207611084,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.165,
      "step": 58
    },
    {
      "batch_size": 4,
      "epoch": 0.0232,
      "step": 58,
      "tokens_per_device": 5660
    },
    {
      "epoch": 0.0232,
      "loss_ce": 0.49706515669822693,
      "loss_lvr": 1.5189189910888672,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6489570736885071,
      "step": 58
    },
    {
      "epoch": 0.0236,
      "grad_norm": 9.990143775939941,
      "learning_rate": 7.733333333333334e-06,
      "loss": 0.5212,
      "step": 59
    },
    {
      "batch_size": 1,
      "epoch": 0.0236,
      "step": 59,
      "tokens_per_device": 4869
    },
    {
      "epoch": 0.0236,
      "loss_ce": 0.9624334573745728,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.9624334573745728,
      "step": 59
    },
    {
      "epoch": 0.024,
      "grad_norm": 15.561911582946777,
      "learning_rate": 7.866666666666667e-06,
      "loss": 0.7369,
      "step": 60
    },
    {
      "batch_size": 4,
      "epoch": 0.024,
      "step": 60,
      "tokens_per_device": 4208
    },
    {
      "epoch": 0.024,
      "loss_ce": 0.44863590598106384,
      "loss_lvr": 0.9947433471679688,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5481102466583252,
      "step": 60
    },
    {
      "epoch": 0.0244,
      "grad_norm": 9.332984924316406,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.4949,
      "step": 61
    },
    {
      "batch_size": 1,
      "epoch": 0.0244,
      "step": 61,
      "tokens_per_device": 5265
    },
    {
      "epoch": 0.0244,
      "loss_ce": 0.08577773720026016,
      "loss_lvr": 0.40843820571899414,
      "loss_mode_switch": 0.0,
      "loss_total": 0.1266215592622757,
      "step": 61
    },
    {
      "epoch": 0.0248,
      "grad_norm": 24.38690948486328,
      "learning_rate": 8.133333333333334e-06,
      "loss": 0.4551,
      "step": 62
    },
    {
      "batch_size": 4,
      "epoch": 0.0248,
      "step": 62,
      "tokens_per_device": 1268
    },
    {
      "epoch": 0.0248,
      "loss_ce": 0.4151110053062439,
      "loss_lvr": 1.9994149208068848,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6150525212287903,
      "step": 62
    },
    {
      "epoch": 0.0252,
      "grad_norm": 14.122325897216797,
      "learning_rate": 8.266666666666667e-06,
      "loss": 0.6556,
      "step": 63
    },
    {
      "batch_size": 1,
      "epoch": 0.0252,
      "step": 63,
      "tokens_per_device": 4874
    },
    {
      "epoch": 0.0252,
      "loss_ce": 0.8747074007987976,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.8747074007987976,
      "step": 63
    },
    {
      "epoch": 0.0256,
      "grad_norm": 14.934876441955566,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.5183,
      "step": 64
    },
    {
      "batch_size": 1,
      "epoch": 0.0256,
      "step": 64,
      "tokens_per_device": 4866
    },
    {
      "epoch": 0.0256,
      "loss_ce": 0.05743364244699478,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.05743364244699478,
      "step": 64
    },
    {
      "epoch": 0.026,
      "grad_norm": 8.778703689575195,
      "learning_rate": 8.533333333333335e-06,
      "loss": 0.37,
      "step": 65
    },
    {
      "batch_size": 1,
      "epoch": 0.026,
      "step": 65,
      "tokens_per_device": 4869
    },
    {
      "epoch": 0.026,
      "loss_ce": 0.610998809337616,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.610998809337616,
      "step": 65
    },
    {
      "epoch": 0.0264,
      "grad_norm": 21.255098342895508,
      "learning_rate": 8.666666666666668e-06,
      "loss": 0.4807,
      "step": 66
    },
    {
      "batch_size": 1,
      "epoch": 0.0264,
      "step": 66,
      "tokens_per_device": 5082
    },
    {
      "epoch": 0.0264,
      "loss_ce": 0.5945461988449097,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5945461988449097,
      "step": 66
    },
    {
      "epoch": 0.0268,
      "grad_norm": 16.64108657836914,
      "learning_rate": 8.8e-06,
      "loss": 0.3873,
      "step": 67
    },
    {
      "batch_size": 4,
      "epoch": 0.0268,
      "step": 67,
      "tokens_per_device": 4172
    },
    {
      "epoch": 0.0268,
      "loss_ce": 0.4301687479019165,
      "loss_lvr": 0.972038984298706,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5273726582527161,
      "step": 67
    },
    {
      "epoch": 0.0272,
      "grad_norm": 10.666621208190918,
      "learning_rate": 8.933333333333333e-06,
      "loss": 0.3371,
      "step": 68
    },
    {
      "batch_size": 1,
      "epoch": 0.0272,
      "step": 68,
      "tokens_per_device": 4840
    },
    {
      "epoch": 0.0272,
      "loss_ce": 0.08642803877592087,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.08642803877592087,
      "step": 68
    },
    {
      "epoch": 0.0276,
      "grad_norm": 8.413142204284668,
      "learning_rate": 9.066666666666667e-06,
      "loss": 0.2875,
      "step": 69
    },
    {
      "batch_size": 1,
      "epoch": 0.0276,
      "step": 69,
      "tokens_per_device": 4848
    },
    {
      "epoch": 0.0276,
      "loss_ce": 0.1024501696228981,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.1024501696228981,
      "step": 69
    },
    {
      "epoch": 0.028,
      "grad_norm": 5.86508846282959,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.0946,
      "step": 70
    },
    {
      "batch_size": 4,
      "epoch": 0.028,
      "step": 70,
      "tokens_per_device": 4360
    },
    {
      "epoch": 0.028,
      "loss_ce": 0.24156023561954498,
      "loss_lvr": 1.0130563974380493,
      "loss_mode_switch": 0.0,
      "loss_total": 0.34286588430404663,
      "step": 70
    },
    {
      "epoch": 0.0284,
      "grad_norm": 7.542107582092285,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.1758,
      "step": 71
    },
    {
      "batch_size": 4,
      "epoch": 0.0284,
      "step": 71,
      "tokens_per_device": 5732
    },
    {
      "epoch": 0.0284,
      "loss_ce": 0.5188885927200317,
      "loss_lvr": 0.8251053094863892,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6013991236686707,
      "step": 71
    },
    {
      "epoch": 0.0288,
      "grad_norm": 14.808649063110352,
      "learning_rate": 9.466666666666667e-06,
      "loss": 0.6198,
      "step": 72
    },
    {
      "batch_size": 4,
      "epoch": 0.0288,
      "step": 72,
      "tokens_per_device": 15520
    },
    {
      "epoch": 0.0288,
      "loss_ce": 0.13771496713161469,
      "loss_lvr": 0.620291531085968,
      "loss_mode_switch": 0.0,
      "loss_total": 0.1997441202402115,
      "step": 72
    },
    {
      "epoch": 0.0292,
      "grad_norm": 8.100099563598633,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.1793,
      "step": 73
    },
    {
      "batch_size": 4,
      "epoch": 0.0292,
      "step": 73,
      "tokens_per_device": 1464
    },
    {
      "epoch": 0.0292,
      "loss_ce": 0.4486251473426819,
      "loss_lvr": 1.00223970413208,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5488491058349609,
      "step": 73
    },
    {
      "epoch": 0.0296,
      "grad_norm": 12.537096977233887,
      "learning_rate": 9.733333333333334e-06,
      "loss": 0.6083,
      "step": 74
    },
    {
      "batch_size": 1,
      "epoch": 0.0296,
      "step": 74,
      "tokens_per_device": 5161
    },
    {
      "epoch": 0.0296,
      "loss_ce": 0.21267053484916687,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.21267053484916687,
      "step": 74
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.044471740722656,
      "learning_rate": 9.866666666666668e-06,
      "loss": 0.2118,
      "step": 75
    },
    {
      "batch_size": 1,
      "epoch": 0.03,
      "step": 75,
      "tokens_per_device": 4996
    },
    {
      "epoch": 0.03,
      "loss_ce": 0.17967942357063293,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.17967942357063293,
      "step": 75
    },
    {
      "epoch": 0.0304,
      "grad_norm": 13.658750534057617,
      "learning_rate": 1e-05,
      "loss": 0.1082,
      "step": 76
    },
    {
      "batch_size": 4,
      "epoch": 0.0304,
      "step": 76,
      "tokens_per_device": 4144
    },
    {
      "epoch": 0.0304,
      "loss_ce": 0.4274771511554718,
      "loss_lvr": 0.8765531778335571,
      "loss_mode_switch": 0.0,
      "loss_total": 0.515132486820221,
      "step": 76
    },
    {
      "epoch": 0.0308,
      "grad_norm": 14.753571510314941,
      "learning_rate": 9.999995804186196e-06,
      "loss": 0.3197,
      "step": 77
    },
    {
      "batch_size": 4,
      "epoch": 0.0308,
      "step": 77,
      "tokens_per_device": 5704
    },
    {
      "epoch": 0.0308,
      "loss_ce": 0.2631130516529083,
      "loss_lvr": 0.7820257544517517,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3413156270980835,
      "step": 77
    },
    {
      "epoch": 0.0312,
      "grad_norm": 10.231792449951172,
      "learning_rate": 9.999983216751826e-06,
      "loss": 0.1983,
      "step": 78
    },
    {
      "batch_size": 4,
      "epoch": 0.0312,
      "step": 78,
      "tokens_per_device": 4568
    },
    {
      "epoch": 0.0312,
      "loss_ce": 0.3263283371925354,
      "loss_lvr": 0.6831410527229309,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3946424424648285,
      "step": 78
    },
    {
      "epoch": 0.0316,
      "grad_norm": 10.503658294677734,
      "learning_rate": 9.999962237718015e-06,
      "loss": 0.4057,
      "step": 79
    },
    {
      "batch_size": 4,
      "epoch": 0.0316,
      "step": 79,
      "tokens_per_device": 1556
    },
    {
      "epoch": 0.0316,
      "loss_ce": 0.21742594242095947,
      "loss_lvr": 1.0807017087936401,
      "loss_mode_switch": 0.0,
      "loss_total": 0.325496107339859,
      "step": 79
    },
    {
      "epoch": 0.032,
      "grad_norm": 8.889445304870605,
      "learning_rate": 9.999932867119974e-06,
      "loss": 0.4428,
      "step": 80
    },
    {
      "batch_size": 4,
      "epoch": 0.032,
      "step": 80,
      "tokens_per_device": 9148
    },
    {
      "epoch": 0.032,
      "loss_ce": 0.10908851027488708,
      "loss_lvr": 0.6354029774665833,
      "loss_mode_switch": 0.0,
      "loss_total": 0.17262881994247437,
      "step": 80
    },
    {
      "epoch": 0.0324,
      "grad_norm": 4.925052165985107,
      "learning_rate": 9.999895105006995e-06,
      "loss": 0.149,
      "step": 81
    },
    {
      "batch_size": 1,
      "epoch": 0.0324,
      "step": 81,
      "tokens_per_device": 4843
    },
    {
      "epoch": 0.0324,
      "loss_ce": 0.21545426547527313,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.21545426547527313,
      "step": 81
    },
    {
      "epoch": 0.0328,
      "grad_norm": 15.88282585144043,
      "learning_rate": 9.999848951442455e-06,
      "loss": 0.2991,
      "step": 82
    },
    {
      "batch_size": 1,
      "epoch": 0.0328,
      "step": 82,
      "tokens_per_device": 5086
    },
    {
      "epoch": 0.0328,
      "loss_ce": 0.022902607917785645,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.022902607917785645,
      "step": 82
    },
    {
      "epoch": 0.0332,
      "grad_norm": 9.770027160644531,
      "learning_rate": 9.999794406503816e-06,
      "loss": 0.1338,
      "step": 83
    },
    {
      "batch_size": 4,
      "epoch": 0.0332,
      "step": 83,
      "tokens_per_device": 5636
    },
    {
      "epoch": 0.0332,
      "loss_ce": 0.5111504793167114,
      "loss_lvr": 0.7647149562835693,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5876219868659973,
      "step": 83
    },
    {
      "epoch": 0.0336,
      "grad_norm": 8.345865249633789,
      "learning_rate": 9.999731470282621e-06,
      "loss": 0.4015,
      "step": 84
    },
    {
      "batch_size": 1,
      "epoch": 0.0336,
      "step": 84,
      "tokens_per_device": 5089
    },
    {
      "epoch": 0.0336,
      "loss_ce": 0.3383285105228424,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3383285105228424,
      "step": 84
    },
    {
      "epoch": 0.034,
      "grad_norm": 11.74326229095459,
      "learning_rate": 9.9996601428845e-06,
      "loss": 0.4784,
      "step": 85
    },
    {
      "batch_size": 4,
      "epoch": 0.034,
      "step": 85,
      "tokens_per_device": 3804
    },
    {
      "epoch": 0.034,
      "loss_ce": 0.17396529018878937,
      "loss_lvr": 1.025069236755371,
      "loss_mode_switch": 0.0,
      "loss_total": 0.27647221088409424,
      "step": 85
    },
    {
      "epoch": 0.0344,
      "grad_norm": 12.577279090881348,
      "learning_rate": 9.99958042442916e-06,
      "loss": 0.4853,
      "step": 86
    },
    {
      "batch_size": 4,
      "epoch": 0.0344,
      "step": 86,
      "tokens_per_device": 4804
    },
    {
      "epoch": 0.0344,
      "loss_ce": 0.17712929844856262,
      "loss_lvr": 0.9664345979690552,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2737727761268616,
      "step": 86
    },
    {
      "epoch": 0.0348,
      "grad_norm": 8.784632682800293,
      "learning_rate": 9.999492315050396e-06,
      "loss": 0.334,
      "step": 87
    },
    {
      "batch_size": 4,
      "epoch": 0.0348,
      "step": 87,
      "tokens_per_device": 5232
    },
    {
      "epoch": 0.0348,
      "loss_ce": 0.29060035943984985,
      "loss_lvr": 0.6381192207336426,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3544122874736786,
      "step": 87
    },
    {
      "epoch": 0.0352,
      "grad_norm": 5.510451316833496,
      "learning_rate": 9.999395814896086e-06,
      "loss": 0.1799,
      "step": 88
    },
    {
      "batch_size": 4,
      "epoch": 0.0352,
      "step": 88,
      "tokens_per_device": 1888
    },
    {
      "epoch": 0.0352,
      "loss_ce": 0.5147313475608826,
      "loss_lvr": 0.9213206171989441,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6068633794784546,
      "step": 88
    },
    {
      "epoch": 0.0356,
      "grad_norm": 8.277078628540039,
      "learning_rate": 9.999290924128186e-06,
      "loss": 0.307,
      "step": 89
    },
    {
      "batch_size": 4,
      "epoch": 0.0356,
      "step": 89,
      "tokens_per_device": 3756
    },
    {
      "epoch": 0.0356,
      "loss_ce": 0.21434064209461212,
      "loss_lvr": 0.8216838240623474,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2965090274810791,
      "step": 89
    },
    {
      "epoch": 0.036,
      "grad_norm": 7.170833587646484,
      "learning_rate": 9.999177642922736e-06,
      "loss": 0.1513,
      "step": 90
    },
    {
      "batch_size": 4,
      "epoch": 0.036,
      "step": 90,
      "tokens_per_device": 3392
    },
    {
      "epoch": 0.036,
      "loss_ce": 0.5914126038551331,
      "loss_lvr": 1.0580822229385376,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6972208023071289,
      "step": 90
    },
    {
      "epoch": 0.0364,
      "grad_norm": 17.997085571289062,
      "learning_rate": 9.999055971469864e-06,
      "loss": 0.4313,
      "step": 91
    },
    {
      "batch_size": 1,
      "epoch": 0.0364,
      "step": 91,
      "tokens_per_device": 5089
    },
    {
      "epoch": 0.0364,
      "loss_ce": 0.08707445114850998,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.08707445114850998,
      "step": 91
    },
    {
      "epoch": 0.0368,
      "grad_norm": 4.926759243011475,
      "learning_rate": 9.998925909973769e-06,
      "loss": 0.1144,
      "step": 92
    },
    {
      "batch_size": 4,
      "epoch": 0.0368,
      "step": 92,
      "tokens_per_device": 4236
    },
    {
      "epoch": 0.0368,
      "loss_ce": 0.6291149854660034,
      "loss_lvr": 1.0172175168991089,
      "loss_mode_switch": 0.0,
      "loss_total": 0.7308367490768433,
      "step": 92
    },
    {
      "epoch": 0.0372,
      "grad_norm": 18.58370018005371,
      "learning_rate": 9.99878745865274e-06,
      "loss": 0.4393,
      "step": 93
    },
    {
      "batch_size": 4,
      "epoch": 0.0372,
      "step": 93,
      "tokens_per_device": 10820
    },
    {
      "epoch": 0.0372,
      "loss_ce": 0.4302127957344055,
      "loss_lvr": 0.4827776849269867,
      "loss_mode_switch": 0.0,
      "loss_total": 0.47849056124687195,
      "step": 93
    },
    {
      "epoch": 0.0376,
      "grad_norm": 16.7265625,
      "learning_rate": 9.99864061773914e-06,
      "loss": 0.3677,
      "step": 94
    },
    {
      "batch_size": 4,
      "epoch": 0.0376,
      "step": 94,
      "tokens_per_device": 15112
    },
    {
      "epoch": 0.0376,
      "loss_ce": 0.10841912031173706,
      "loss_lvr": 0.6659187078475952,
      "loss_mode_switch": 0.0,
      "loss_total": 0.17501099407672882,
      "step": 94
    },
    {
      "epoch": 0.038,
      "grad_norm": 8.039645195007324,
      "learning_rate": 9.998485387479418e-06,
      "loss": 0.3075,
      "step": 95
    },
    {
      "batch_size": 4,
      "epoch": 0.038,
      "step": 95,
      "tokens_per_device": 4908
    },
    {
      "epoch": 0.038,
      "loss_ce": 0.36745840311050415,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.36745840311050415,
      "step": 95
    },
    {
      "epoch": 0.0384,
      "grad_norm": 15.823453903198242,
      "learning_rate": 9.998321768134101e-06,
      "loss": 0.5428,
      "step": 96
    },
    {
      "batch_size": 4,
      "epoch": 0.0384,
      "step": 96,
      "tokens_per_device": 3792
    },
    {
      "epoch": 0.0384,
      "loss_ce": 0.2773635685443878,
      "loss_lvr": 0.7412651777267456,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3514900803565979,
      "step": 96
    },
    {
      "epoch": 0.0388,
      "grad_norm": 8.493693351745605,
      "learning_rate": 9.998149759977795e-06,
      "loss": 0.3814,
      "step": 97
    },
    {
      "batch_size": 4,
      "epoch": 0.0388,
      "step": 97,
      "tokens_per_device": 1720
    },
    {
      "epoch": 0.0388,
      "loss_ce": 1.1534491777420044,
      "loss_lvr": 1.125251054763794,
      "loss_mode_switch": 0.0,
      "loss_total": 1.2659742832183838,
      "step": 97
    },
    {
      "epoch": 0.0392,
      "grad_norm": 11.638108253479004,
      "learning_rate": 9.997969363299187e-06,
      "loss": 0.8435,
      "step": 98
    },
    {
      "batch_size": 4,
      "epoch": 0.0392,
      "step": 98,
      "tokens_per_device": 11892
    },
    {
      "epoch": 0.0392,
      "loss_ce": 0.29854899644851685,
      "loss_lvr": 0.8156141042709351,
      "loss_mode_switch": 0.0,
      "loss_total": 0.38011041283607483,
      "step": 98
    },
    {
      "epoch": 0.0396,
      "grad_norm": 12.322162628173828,
      "learning_rate": 9.99778057840104e-06,
      "loss": 0.4805,
      "step": 99
    },
    {
      "batch_size": 4,
      "epoch": 0.0396,
      "step": 99,
      "tokens_per_device": 4804
    },
    {
      "epoch": 0.0396,
      "loss_ce": 0.07730291783809662,
      "loss_lvr": 0.5763208270072937,
      "loss_mode_switch": 0.0,
      "loss_total": 0.13493500649929047,
      "step": 99
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.8659305572509766,
      "learning_rate": 9.997583405600194e-06,
      "loss": 0.0844,
      "step": 100
    },
    {
      "batch_size": 4,
      "epoch": 0.04,
      "step": 100,
      "tokens_per_device": 3728
    },
    {
      "epoch": 0.04,
      "loss_ce": 0.27456116676330566,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.27456116676330566,
      "step": 100
    },
    {
      "epoch": 0.0404,
      "grad_norm": 9.431591987609863,
      "learning_rate": 9.997377845227577e-06,
      "loss": 0.3337,
      "step": 101
    },
    {
      "batch_size": 1,
      "epoch": 0.0404,
      "step": 101,
      "tokens_per_device": 4851
    },
    {
      "epoch": 0.0404,
      "loss_ce": 0.18896684050559998,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.18896684050559998,
      "step": 101
    },
    {
      "epoch": 0.0408,
      "grad_norm": 10.035370826721191,
      "learning_rate": 9.997163897628175e-06,
      "loss": 0.2826,
      "step": 102
    },
    {
      "batch_size": 1,
      "epoch": 0.0408,
      "step": 102,
      "tokens_per_device": 5151
    },
    {
      "epoch": 0.0408,
      "loss_ce": 0.021510761231184006,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.021510761231184006,
      "step": 102
    },
    {
      "epoch": 0.0412,
      "grad_norm": 3.322061061859131,
      "learning_rate": 9.996941563161071e-06,
      "loss": 0.015,
      "step": 103
    },
    {
      "batch_size": 1,
      "epoch": 0.0412,
      "step": 103,
      "tokens_per_device": 4829
    },
    {
      "epoch": 0.0412,
      "loss_ce": 0.19415193796157837,
      "loss_lvr": 0.3519080579280853,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2293427437543869,
      "step": 103
    },
    {
      "epoch": 0.0416,
      "grad_norm": 12.623085021972656,
      "learning_rate": 9.996710842199412e-06,
      "loss": 0.2061,
      "step": 104
    },
    {
      "batch_size": 4,
      "epoch": 0.0416,
      "step": 104,
      "tokens_per_device": 1288
    },
    {
      "epoch": 0.0416,
      "loss_ce": 0.25354939699172974,
      "loss_lvr": 1.0436383485794067,
      "loss_mode_switch": 0.0,
      "loss_total": 0.35791322588920593,
      "step": 104
    },
    {
      "epoch": 0.042,
      "grad_norm": 10.081673622131348,
      "learning_rate": 9.996471735130422e-06,
      "loss": 0.2579,
      "step": 105
    },
    {
      "batch_size": 4,
      "epoch": 0.042,
      "step": 105,
      "tokens_per_device": 4156
    },
    {
      "epoch": 0.042,
      "loss_ce": 0.1395236700773239,
      "loss_lvr": 0.9172446131706238,
      "loss_mode_switch": 0.0,
      "loss_total": 0.231248140335083,
      "step": 105
    },
    {
      "epoch": 0.0424,
      "grad_norm": 17.169279098510742,
      "learning_rate": 9.9962242423554e-06,
      "loss": 0.315,
      "step": 106
    },
    {
      "batch_size": 4,
      "epoch": 0.0424,
      "step": 106,
      "tokens_per_device": 1268
    },
    {
      "epoch": 0.0424,
      "loss_ce": 0.49787381291389465,
      "loss_lvr": 0.7612568140029907,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5739995241165161,
      "step": 106
    },
    {
      "epoch": 0.0428,
      "grad_norm": 13.927685737609863,
      "learning_rate": 9.995968364289719e-06,
      "loss": 0.706,
      "step": 107
    },
    {
      "batch_size": 4,
      "epoch": 0.0428,
      "step": 107,
      "tokens_per_device": 4192
    },
    {
      "epoch": 0.0428,
      "loss_ce": 0.22593143582344055,
      "loss_lvr": 1.0203602313995361,
      "loss_mode_switch": 0.0,
      "loss_total": 0.32796746492385864,
      "step": 107
    },
    {
      "epoch": 0.0432,
      "grad_norm": 10.084362983703613,
      "learning_rate": 9.99570410136283e-06,
      "loss": 0.3517,
      "step": 108
    },
    {
      "batch_size": 4,
      "epoch": 0.0432,
      "step": 108,
      "tokens_per_device": 4568
    },
    {
      "epoch": 0.0432,
      "loss_ce": 0.16896148025989532,
      "loss_lvr": 1.1289175748825073,
      "loss_mode_switch": 0.0,
      "loss_total": 0.28185322880744934,
      "step": 108
    },
    {
      "epoch": 0.0436,
      "grad_norm": 7.834078311920166,
      "learning_rate": 9.995431454018246e-06,
      "loss": 0.3224,
      "step": 109
    },
    {
      "batch_size": 1,
      "epoch": 0.0436,
      "step": 109,
      "tokens_per_device": 4615
    },
    {
      "epoch": 0.0436,
      "loss_ce": 0.2572348713874817,
      "loss_lvr": 0.43113699555397034,
      "loss_mode_switch": 0.0,
      "loss_total": 0.30034857988357544,
      "step": 109
    },
    {
      "epoch": 0.044,
      "grad_norm": 14.421305656433105,
      "learning_rate": 9.995150422713561e-06,
      "loss": 0.6087,
      "step": 110
    },
    {
      "batch_size": 4,
      "epoch": 0.044,
      "step": 110,
      "tokens_per_device": 4240
    },
    {
      "epoch": 0.044,
      "loss_ce": 0.46053189039230347,
      "loss_lvr": 0.9952828288078308,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5600601434707642,
      "step": 110
    },
    {
      "epoch": 0.0444,
      "grad_norm": 8.947388648986816,
      "learning_rate": 9.99486100792044e-06,
      "loss": 0.4588,
      "step": 111
    },
    {
      "batch_size": 4,
      "epoch": 0.0444,
      "step": 111,
      "tokens_per_device": 15916
    },
    {
      "epoch": 0.0444,
      "loss_ce": 0.39660412073135376,
      "loss_lvr": 0.9810446500778198,
      "loss_mode_switch": 0.0,
      "loss_total": 0.4947085976600647,
      "step": 111
    },
    {
      "epoch": 0.0448,
      "grad_norm": 9.45101261138916,
      "learning_rate": 9.99456321012461e-06,
      "loss": 0.4017,
      "step": 112
    },
    {
      "batch_size": 4,
      "epoch": 0.0448,
      "step": 112,
      "tokens_per_device": 7484
    },
    {
      "epoch": 0.0448,
      "loss_ce": 0.2330116480588913,
      "loss_lvr": 0.5845057368278503,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2914622128009796,
      "step": 112
    },
    {
      "epoch": 0.0452,
      "grad_norm": 7.8349995613098145,
      "learning_rate": 9.994257029825876e-06,
      "loss": 0.3651,
      "step": 113
    },
    {
      "batch_size": 4,
      "epoch": 0.0452,
      "step": 113,
      "tokens_per_device": 1188
    },
    {
      "epoch": 0.0452,
      "loss_ce": 0.38543546199798584,
      "loss_lvr": 1.0379724502563477,
      "loss_mode_switch": 0.0,
      "loss_total": 0.48923271894454956,
      "step": 113
    },
    {
      "epoch": 0.0456,
      "grad_norm": 8.869186401367188,
      "learning_rate": 9.993942467538107e-06,
      "loss": 0.4248,
      "step": 114
    },
    {
      "batch_size": 4,
      "epoch": 0.0456,
      "step": 114,
      "tokens_per_device": 14452
    },
    {
      "epoch": 0.0456,
      "loss_ce": 0.1478455513715744,
      "loss_lvr": 1.1511600017547607,
      "loss_mode_switch": 0.0,
      "loss_total": 0.26296156644821167,
      "step": 114
    },
    {
      "epoch": 0.046,
      "grad_norm": 11.535511016845703,
      "learning_rate": 9.993619523789241e-06,
      "loss": 0.5769,
      "step": 115
    },
    {
      "batch_size": 1,
      "epoch": 0.046,
      "step": 115,
      "tokens_per_device": 4891
    },
    {
      "epoch": 0.046,
      "loss_ce": 0.05704773962497711,
      "loss_lvr": 0.6744264960289001,
      "loss_mode_switch": 0.0,
      "loss_total": 0.12449038773775101,
      "step": 115
    },
    {
      "epoch": 0.0464,
      "grad_norm": 12.578680992126465,
      "learning_rate": 9.993288199121283e-06,
      "loss": 0.1306,
      "step": 116
    },
    {
      "batch_size": 4,
      "epoch": 0.0464,
      "step": 116,
      "tokens_per_device": 6424
    },
    {
      "epoch": 0.0464,
      "loss_ce": 0.4048420786857605,
      "loss_lvr": 0.45272547006607056,
      "loss_mode_switch": 0.0,
      "loss_total": 0.4501146376132965,
      "step": 116
    },
    {
      "epoch": 0.0468,
      "grad_norm": 15.339097023010254,
      "learning_rate": 9.992948494090303e-06,
      "loss": 0.295,
      "step": 117
    },
    {
      "batch_size": 4,
      "epoch": 0.0468,
      "step": 117,
      "tokens_per_device": 3740
    },
    {
      "epoch": 0.0468,
      "loss_ce": 0.19220618903636932,
      "loss_lvr": 0.8141260147094727,
      "loss_mode_switch": 0.0,
      "loss_total": 0.27361878752708435,
      "step": 117
    },
    {
      "epoch": 0.0472,
      "grad_norm": 11.80385971069336,
      "learning_rate": 9.992600409266437e-06,
      "loss": 0.3208,
      "step": 118
    },
    {
      "batch_size": 4,
      "epoch": 0.0472,
      "step": 118,
      "tokens_per_device": 3788
    },
    {
      "epoch": 0.0472,
      "loss_ce": 0.5344772934913635,
      "loss_lvr": 1.2575476169586182,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6602320671081543,
      "step": 118
    },
    {
      "epoch": 0.0476,
      "grad_norm": 11.772353172302246,
      "learning_rate": 9.992243945233886e-06,
      "loss": 0.3652,
      "step": 119
    },
    {
      "batch_size": 4,
      "epoch": 0.0476,
      "step": 119,
      "tokens_per_device": 12984
    },
    {
      "epoch": 0.0476,
      "loss_ce": 0.13681243360042572,
      "loss_lvr": 0.8885440826416016,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2256668508052826,
      "step": 119
    },
    {
      "epoch": 0.048,
      "grad_norm": 8.710790634155273,
      "learning_rate": 9.991879102590912e-06,
      "loss": 0.3333,
      "step": 120
    },
    {
      "batch_size": 4,
      "epoch": 0.048,
      "step": 120,
      "tokens_per_device": 1512
    },
    {
      "epoch": 0.048,
      "loss_ce": 0.7194777727127075,
      "loss_lvr": 0.7629538774490356,
      "loss_mode_switch": 0.0,
      "loss_total": 0.7957731485366821,
      "step": 120
    },
    {
      "epoch": 0.0484,
      "grad_norm": 10.309539794921875,
      "learning_rate": 9.991505881949837e-06,
      "loss": 0.5502,
      "step": 121
    },
    {
      "batch_size": 1,
      "epoch": 0.0484,
      "step": 121,
      "tokens_per_device": 4685
    },
    {
      "epoch": 0.0484,
      "loss_ce": 0.12873028218746185,
      "loss_lvr": 0.4614042043685913,
      "loss_mode_switch": 0.0,
      "loss_total": 0.17487069964408875,
      "step": 121
    },
    {
      "epoch": 0.0488,
      "grad_norm": 15.273765563964844,
      "learning_rate": 9.991124283937049e-06,
      "loss": 0.2919,
      "step": 122
    },
    {
      "batch_size": 1,
      "epoch": 0.0488,
      "step": 122,
      "tokens_per_device": 5078
    },
    {
      "epoch": 0.0488,
      "loss_ce": 0.009771143086254597,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.009771143086254597,
      "step": 122
    },
    {
      "epoch": 0.0492,
      "grad_norm": 4.2728753089904785,
      "learning_rate": 9.990734309192995e-06,
      "loss": 0.1243,
      "step": 123
    },
    {
      "batch_size": 4,
      "epoch": 0.0492,
      "step": 123,
      "tokens_per_device": 1176
    },
    {
      "epoch": 0.0492,
      "loss_ce": 0.6897622346878052,
      "loss_lvr": 1.1501744985580444,
      "loss_mode_switch": 0.0,
      "loss_total": 0.8047797083854675,
      "step": 123
    },
    {
      "epoch": 0.0496,
      "grad_norm": 12.110310554504395,
      "learning_rate": 9.990335958372178e-06,
      "loss": 0.7424,
      "step": 124
    },
    {
      "batch_size": 1,
      "epoch": 0.0496,
      "step": 124,
      "tokens_per_device": 5065
    },
    {
      "epoch": 0.0496,
      "loss_ce": 0.0690104216337204,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.0690104216337204,
      "step": 124
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.628517150878906,
      "learning_rate": 9.989929232143159e-06,
      "loss": 0.3023,
      "step": 125
    },
    {
      "batch_size": 1,
      "epoch": 0.05,
      "step": 125,
      "tokens_per_device": 5167
    },
    {
      "epoch": 0.05,
      "loss_ce": 0.02000514417886734,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.02000514417886734,
      "step": 125
    },
    {
      "epoch": 0.0504,
      "grad_norm": 5.298424243927002,
      "learning_rate": 9.98951413118856e-06,
      "loss": 0.1502,
      "step": 126
    },
    {
      "batch_size": 1,
      "epoch": 0.0504,
      "step": 126,
      "tokens_per_device": 5013
    },
    {
      "epoch": 0.0504,
      "loss_ce": 0.2773449122905731,
      "loss_lvr": 0.22950449585914612,
      "loss_mode_switch": 0.0,
      "loss_total": 0.300295352935791,
      "step": 126
    },
    {
      "epoch": 0.0508,
      "grad_norm": 10.078692436218262,
      "learning_rate": 9.989090656205052e-06,
      "loss": 0.2208,
      "step": 127
    },
    {
      "batch_size": 4,
      "epoch": 0.0508,
      "step": 127,
      "tokens_per_device": 5964
    },
    {
      "epoch": 0.0508,
      "loss_ce": 0.25345107913017273,
      "loss_lvr": 0.8941274285316467,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3428638279438019,
      "step": 127
    },
    {
      "epoch": 0.0512,
      "grad_norm": 11.44184398651123,
      "learning_rate": 9.988658807903369e-06,
      "loss": 0.3896,
      "step": 128
    },
    {
      "batch_size": 1,
      "epoch": 0.0512,
      "step": 128,
      "tokens_per_device": 4739
    },
    {
      "epoch": 0.0512,
      "loss_ce": 0.06816263496875763,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.06816263496875763,
      "step": 128
    },
    {
      "epoch": 0.0516,
      "grad_norm": 7.125197887420654,
      "learning_rate": 9.988218587008287e-06,
      "loss": 0.1286,
      "step": 129
    },
    {
      "batch_size": 4,
      "epoch": 0.0516,
      "step": 129,
      "tokens_per_device": 4208
    },
    {
      "epoch": 0.0516,
      "loss_ce": 0.09152233600616455,
      "loss_lvr": 0.6366158723831177,
      "loss_mode_switch": 0.0,
      "loss_total": 0.15518392622470856,
      "step": 129
    },
    {
      "epoch": 0.052,
      "grad_norm": 9.01850414276123,
      "learning_rate": 9.987769994258645e-06,
      "loss": 0.1049,
      "step": 130
    },
    {
      "batch_size": 1,
      "epoch": 0.052,
      "step": 130,
      "tokens_per_device": 5010
    },
    {
      "epoch": 0.052,
      "loss_ce": 0.2112990766763687,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2112990766763687,
      "step": 130
    },
    {
      "epoch": 0.0524,
      "grad_norm": 16.64312744140625,
      "learning_rate": 9.987313030407325e-06,
      "loss": 0.4614,
      "step": 131
    },
    {
      "batch_size": 1,
      "epoch": 0.0524,
      "step": 131,
      "tokens_per_device": 5187
    },
    {
      "epoch": 0.0524,
      "loss_ce": 0.02873559296131134,
      "loss_lvr": 0.19450512528419495,
      "loss_mode_switch": 0.0,
      "loss_total": 0.048186108469963074,
      "step": 131
    },
    {
      "epoch": 0.0528,
      "grad_norm": 6.522212982177734,
      "learning_rate": 9.98684769622126e-06,
      "loss": 0.0902,
      "step": 132
    },
    {
      "batch_size": 4,
      "epoch": 0.0528,
      "step": 132,
      "tokens_per_device": 4144
    },
    {
      "epoch": 0.0528,
      "loss_ce": 0.5453018546104431,
      "loss_lvr": 1.1683987379074097,
      "loss_mode_switch": 0.0,
      "loss_total": 0.662141740322113,
      "step": 132
    },
    {
      "epoch": 0.0532,
      "grad_norm": 10.550134658813477,
      "learning_rate": 9.986373992481434e-06,
      "loss": 0.515,
      "step": 133
    },
    {
      "batch_size": 4,
      "epoch": 0.0532,
      "step": 133,
      "tokens_per_device": 5696
    },
    {
      "epoch": 0.0532,
      "loss_ce": 0.4195650517940521,
      "loss_lvr": 0.978010892868042,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5173661708831787,
      "step": 133
    },
    {
      "epoch": 0.0536,
      "grad_norm": 13.687209129333496,
      "learning_rate": 9.985891919982878e-06,
      "loss": 0.5553,
      "step": 134
    },
    {
      "batch_size": 4,
      "epoch": 0.0536,
      "step": 134,
      "tokens_per_device": 5200
    },
    {
      "epoch": 0.0536,
      "loss_ce": 0.34458819031715393,
      "loss_lvr": 1.0119065046310425,
      "loss_mode_switch": 0.0,
      "loss_total": 0.44577884674072266,
      "step": 134
    },
    {
      "epoch": 0.054,
      "grad_norm": 8.673235893249512,
      "learning_rate": 9.985401479534664e-06,
      "loss": 0.3215,
      "step": 135
    },
    {
      "batch_size": 4,
      "epoch": 0.054,
      "step": 135,
      "tokens_per_device": 4168
    },
    {
      "epoch": 0.054,
      "loss_ce": 0.2771547734737396,
      "loss_lvr": 1.0045875310897827,
      "loss_mode_switch": 0.0,
      "loss_total": 0.37761354446411133,
      "step": 135
    },
    {
      "epoch": 0.0544,
      "grad_norm": 6.75148344039917,
      "learning_rate": 9.984902671959911e-06,
      "loss": 0.1981,
      "step": 136
    },
    {
      "batch_size": 1,
      "epoch": 0.0544,
      "step": 136,
      "tokens_per_device": 4733
    },
    {
      "epoch": 0.0544,
      "loss_ce": 0.07755197584629059,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.07755197584629059,
      "step": 136
    },
    {
      "epoch": 0.0548,
      "grad_norm": 7.315802574157715,
      "learning_rate": 9.98439549809578e-06,
      "loss": 0.0466,
      "step": 137
    },
    {
      "batch_size": 1,
      "epoch": 0.0548,
      "step": 137,
      "tokens_per_device": 4840
    },
    {
      "epoch": 0.0548,
      "loss_ce": 0.021204866468906403,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.021204866468906403,
      "step": 137
    },
    {
      "epoch": 0.0552,
      "grad_norm": 2.0620532035827637,
      "learning_rate": 9.983879958793476e-06,
      "loss": 0.0852,
      "step": 138
    },
    {
      "batch_size": 4,
      "epoch": 0.0552,
      "step": 138,
      "tokens_per_device": 1356
    },
    {
      "epoch": 0.0552,
      "loss_ce": 0.3249052166938782,
      "loss_lvr": 0.7319856286048889,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3981037735939026,
      "step": 138
    },
    {
      "epoch": 0.0556,
      "grad_norm": 11.380744934082031,
      "learning_rate": 9.983356054918238e-06,
      "loss": 0.495,
      "step": 139
    },
    {
      "batch_size": 4,
      "epoch": 0.0556,
      "step": 139,
      "tokens_per_device": 3560
    },
    {
      "epoch": 0.0556,
      "loss_ce": 0.2721814215183258,
      "loss_lvr": 0.7968034148216248,
      "loss_mode_switch": 0.0,
      "loss_total": 0.35186177492141724,
      "step": 139
    },
    {
      "epoch": 0.056,
      "grad_norm": 10.390738487243652,
      "learning_rate": 9.982823787349352e-06,
      "loss": 0.4685,
      "step": 140
    },
    {
      "batch_size": 4,
      "epoch": 0.056,
      "step": 140,
      "tokens_per_device": 1308
    },
    {
      "epoch": 0.056,
      "loss_ce": 0.7361812591552734,
      "loss_lvr": 1.1738325357437134,
      "loss_mode_switch": 0.0,
      "loss_total": 0.8535645008087158,
      "step": 140
    },
    {
      "epoch": 0.0564,
      "grad_norm": 9.345442771911621,
      "learning_rate": 9.982283156980133e-06,
      "loss": 0.5405,
      "step": 141
    },
    {
      "batch_size": 4,
      "epoch": 0.0564,
      "step": 141,
      "tokens_per_device": 7804
    },
    {
      "epoch": 0.0564,
      "loss_ce": 0.267467200756073,
      "loss_lvr": 0.8713016510009766,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3545973598957062,
      "step": 141
    },
    {
      "epoch": 0.0568,
      "grad_norm": 10.474932670593262,
      "learning_rate": 9.981734164717936e-06,
      "loss": 0.2769,
      "step": 142
    },
    {
      "batch_size": 1,
      "epoch": 0.0568,
      "step": 142,
      "tokens_per_device": 4869
    },
    {
      "epoch": 0.0568,
      "loss_ce": 0.02990519441664219,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.02990519441664219,
      "step": 142
    },
    {
      "epoch": 0.0572,
      "grad_norm": 9.67310619354248,
      "learning_rate": 9.981176811484148e-06,
      "loss": 0.0804,
      "step": 143
    },
    {
      "batch_size": 4,
      "epoch": 0.0572,
      "step": 143,
      "tokens_per_device": 3716
    },
    {
      "epoch": 0.0572,
      "loss_ce": 0.657302737236023,
      "loss_lvr": 0.8478293418884277,
      "loss_mode_switch": 0.0,
      "loss_total": 0.7420856952667236,
      "step": 143
    },
    {
      "epoch": 0.0576,
      "grad_norm": 13.42434024810791,
      "learning_rate": 9.98061109821419e-06,
      "loss": 0.4686,
      "step": 144
    },
    {
      "batch_size": 4,
      "epoch": 0.0576,
      "step": 144,
      "tokens_per_device": 4168
    },
    {
      "epoch": 0.0576,
      "loss_ce": 0.10645794868469238,
      "loss_lvr": 0.787754237651825,
      "loss_mode_switch": 0.0,
      "loss_total": 0.18523338437080383,
      "step": 144
    },
    {
      "epoch": 0.058,
      "grad_norm": 4.21500301361084,
      "learning_rate": 9.980037025857511e-06,
      "loss": 0.1158,
      "step": 145
    },
    {
      "batch_size": 1,
      "epoch": 0.058,
      "step": 145,
      "tokens_per_device": 4881
    },
    {
      "epoch": 0.058,
      "loss_ce": 0.12326357513666153,
      "loss_lvr": 0.3597036898136139,
      "loss_mode_switch": 0.0,
      "loss_total": 0.1592339426279068,
      "step": 145
    },
    {
      "epoch": 0.0584,
      "grad_norm": 8.884831428527832,
      "learning_rate": 9.979454595377594e-06,
      "loss": 0.2047,
      "step": 146
    },
    {
      "batch_size": 4,
      "epoch": 0.0584,
      "step": 146,
      "tokens_per_device": 5696
    },
    {
      "epoch": 0.0584,
      "loss_ce": 0.2858547270298004,
      "loss_lvr": 0.9548780918121338,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3813425302505493,
      "step": 146
    },
    {
      "epoch": 0.0588,
      "grad_norm": 10.341806411743164,
      "learning_rate": 9.978863807751944e-06,
      "loss": 0.4002,
      "step": 147
    },
    {
      "batch_size": 4,
      "epoch": 0.0588,
      "step": 147,
      "tokens_per_device": 3796
    },
    {
      "epoch": 0.0588,
      "loss_ce": 0.1697288304567337,
      "loss_lvr": 0.8732799887657166,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2570568323135376,
      "step": 147
    },
    {
      "epoch": 0.0592,
      "grad_norm": 13.500733375549316,
      "learning_rate": 9.978264663972099e-06,
      "loss": 0.1955,
      "step": 148
    },
    {
      "batch_size": 4,
      "epoch": 0.0592,
      "step": 148,
      "tokens_per_device": 1312
    },
    {
      "epoch": 0.0592,
      "loss_ce": 0.23529602587223053,
      "loss_lvr": 0.8695405721664429,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3222500681877136,
      "step": 148
    },
    {
      "epoch": 0.0596,
      "grad_norm": 10.706506729125977,
      "learning_rate": 9.977657165043613e-06,
      "loss": 0.4042,
      "step": 149
    },
    {
      "batch_size": 4,
      "epoch": 0.0596,
      "step": 149,
      "tokens_per_device": 3744
    },
    {
      "epoch": 0.0596,
      "loss_ce": 0.40540024638175964,
      "loss_lvr": 0.7563202977180481,
      "loss_mode_switch": 0.0,
      "loss_total": 0.48103228211402893,
      "step": 149
    },
    {
      "epoch": 0.06,
      "grad_norm": 19.285926818847656,
      "learning_rate": 9.977041311986072e-06,
      "loss": 0.4437,
      "step": 150
    },
    {
      "batch_size": 4,
      "epoch": 0.06,
      "step": 150,
      "tokens_per_device": 1240
    },
    {
      "epoch": 0.06,
      "loss_ce": 0.5960803627967834,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5960803627967834,
      "step": 150
    },
    {
      "epoch": 0.0604,
      "grad_norm": 11.658673286437988,
      "learning_rate": 9.97641710583307e-06,
      "loss": 0.3869,
      "step": 151
    },
    {
      "batch_size": 4,
      "epoch": 0.0604,
      "step": 151,
      "tokens_per_device": 5792
    },
    {
      "epoch": 0.0604,
      "loss_ce": 0.38365140557289124,
      "loss_lvr": 0.7615958452224731,
      "loss_mode_switch": 0.0,
      "loss_total": 0.4598110020160675,
      "step": 151
    },
    {
      "epoch": 0.0608,
      "grad_norm": 9.331374168395996,
      "learning_rate": 9.975784547632237e-06,
      "loss": 0.4133,
      "step": 152
    },
    {
      "batch_size": 1,
      "epoch": 0.0608,
      "step": 152,
      "tokens_per_device": 5105
    },
    {
      "epoch": 0.0608,
      "loss_ce": 0.01205648947507143,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.01205648947507143,
      "step": 152
    },
    {
      "epoch": 0.0612,
      "grad_norm": 5.857874393463135,
      "learning_rate": 9.975143638445205e-06,
      "loss": 0.1897,
      "step": 153
    },
    {
      "batch_size": 4,
      "epoch": 0.0612,
      "step": 153,
      "tokens_per_device": 2588
    },
    {
      "epoch": 0.0612,
      "loss_ce": 0.12604810297489166,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.12604810297489166,
      "step": 153
    },
    {
      "epoch": 0.0616,
      "grad_norm": 9.216309547424316,
      "learning_rate": 9.974494379347632e-06,
      "loss": 0.39,
      "step": 154
    },
    {
      "batch_size": 4,
      "epoch": 0.0616,
      "step": 154,
      "tokens_per_device": 5512
    },
    {
      "epoch": 0.0616,
      "loss_ce": 0.14892669022083282,
      "loss_lvr": 0.8365105390548706,
      "loss_mode_switch": 0.0,
      "loss_total": 0.23257774114608765,
      "step": 154
    },
    {
      "epoch": 0.062,
      "grad_norm": 15.17099666595459,
      "learning_rate": 9.973836771429185e-06,
      "loss": 0.2756,
      "step": 155
    },
    {
      "batch_size": 4,
      "epoch": 0.062,
      "step": 155,
      "tokens_per_device": 4212
    },
    {
      "epoch": 0.062,
      "loss_ce": 0.5514479875564575,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5514479875564575,
      "step": 155
    },
    {
      "epoch": 0.0624,
      "grad_norm": 10.67226791381836,
      "learning_rate": 9.973170815793543e-06,
      "loss": 0.5277,
      "step": 156
    },
    {
      "batch_size": 4,
      "epoch": 0.0624,
      "step": 156,
      "tokens_per_device": 5588
    },
    {
      "epoch": 0.0624,
      "loss_ce": 0.39043524861335754,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.39043524861335754,
      "step": 156
    },
    {
      "epoch": 0.0628,
      "grad_norm": 10.888937950134277,
      "learning_rate": 9.972496513558399e-06,
      "loss": 0.639,
      "step": 157
    },
    {
      "batch_size": 4,
      "epoch": 0.0628,
      "step": 157,
      "tokens_per_device": 4360
    },
    {
      "epoch": 0.0628,
      "loss_ce": 0.4401870369911194,
      "loss_lvr": 0.9035695791244507,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5305439829826355,
      "step": 157
    },
    {
      "epoch": 0.0632,
      "grad_norm": 16.659099578857422,
      "learning_rate": 9.971813865855448e-06,
      "loss": 0.5979,
      "step": 158
    },
    {
      "batch_size": 1,
      "epoch": 0.0632,
      "step": 158,
      "tokens_per_device": 6279
    },
    {
      "epoch": 0.0632,
      "loss_ce": 0.23629224300384521,
      "loss_lvr": 0.46730703115463257,
      "loss_mode_switch": 0.0,
      "loss_total": 0.283022940158844,
      "step": 158
    },
    {
      "epoch": 0.0636,
      "grad_norm": 14.724784851074219,
      "learning_rate": 9.971122873830398e-06,
      "loss": 0.5326,
      "step": 159
    },
    {
      "batch_size": 1,
      "epoch": 0.0636,
      "step": 159,
      "tokens_per_device": 5004
    },
    {
      "epoch": 0.0636,
      "loss_ce": 0.022241558879613876,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.022241558879613876,
      "step": 159
    },
    {
      "epoch": 0.064,
      "grad_norm": 7.006820201873779,
      "learning_rate": 9.970423538642959e-06,
      "loss": 0.1527,
      "step": 160
    },
    {
      "batch_size": 4,
      "epoch": 0.064,
      "step": 160,
      "tokens_per_device": 1144
    },
    {
      "epoch": 0.064,
      "loss_ce": 0.41480574011802673,
      "loss_lvr": 1.4044311046600342,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5552488565444946,
      "step": 160
    },
    {
      "epoch": 0.0644,
      "grad_norm": 10.77254867553711,
      "learning_rate": 9.969715861466839e-06,
      "loss": 0.5285,
      "step": 161
    },
    {
      "batch_size": 4,
      "epoch": 0.0644,
      "step": 161,
      "tokens_per_device": 4264
    },
    {
      "epoch": 0.0644,
      "loss_ce": 0.37468385696411133,
      "loss_lvr": 1.1608057022094727,
      "loss_mode_switch": 0.0,
      "loss_total": 0.49076443910598755,
      "step": 161
    },
    {
      "epoch": 0.0648,
      "grad_norm": 8.504761695861816,
      "learning_rate": 9.968999843489755e-06,
      "loss": 0.456,
      "step": 162
    },
    {
      "batch_size": 4,
      "epoch": 0.0648,
      "step": 162,
      "tokens_per_device": 5156
    },
    {
      "epoch": 0.0648,
      "loss_ce": 0.17189355194568634,
      "loss_lvr": 0.7765597105026245,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2495495229959488,
      "step": 162
    },
    {
      "epoch": 0.0652,
      "grad_norm": 10.866538047790527,
      "learning_rate": 9.968275485913417e-06,
      "loss": 0.3166,
      "step": 163
    },
    {
      "batch_size": 4,
      "epoch": 0.0652,
      "step": 163,
      "tokens_per_device": 1892
    },
    {
      "epoch": 0.0652,
      "loss_ce": 0.1418648213148117,
      "loss_lvr": 1.012363314628601,
      "loss_mode_switch": 0.0,
      "loss_total": 0.24310114979743958,
      "step": 163
    },
    {
      "epoch": 0.0656,
      "grad_norm": 9.645085334777832,
      "learning_rate": 9.967542789953532e-06,
      "loss": 0.338,
      "step": 164
    },
    {
      "batch_size": 4,
      "epoch": 0.0656,
      "step": 164,
      "tokens_per_device": 2960
    },
    {
      "epoch": 0.0656,
      "loss_ce": 0.48259347677230835,
      "loss_lvr": 0.8273354768753052,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5653270483016968,
      "step": 164
    },
    {
      "epoch": 0.066,
      "grad_norm": 8.953377723693848,
      "learning_rate": 9.966801756839802e-06,
      "loss": 0.4318,
      "step": 165
    },
    {
      "batch_size": 4,
      "epoch": 0.066,
      "step": 165,
      "tokens_per_device": 4212
    },
    {
      "epoch": 0.066,
      "loss_ce": 0.630429208278656,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.630429208278656,
      "step": 165
    },
    {
      "epoch": 0.0664,
      "grad_norm": 22.183073043823242,
      "learning_rate": 9.966052387815923e-06,
      "loss": 0.5865,
      "step": 166
    },
    {
      "batch_size": 4,
      "epoch": 0.0664,
      "step": 166,
      "tokens_per_device": 5660
    },
    {
      "epoch": 0.0664,
      "loss_ce": 0.2864610254764557,
      "loss_lvr": 1.0181609392166138,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3882771134376526,
      "step": 166
    },
    {
      "epoch": 0.0668,
      "grad_norm": 19.522796630859375,
      "learning_rate": 9.96529468413958e-06,
      "loss": 0.3736,
      "step": 167
    },
    {
      "batch_size": 4,
      "epoch": 0.0668,
      "step": 167,
      "tokens_per_device": 1156
    },
    {
      "epoch": 0.0668,
      "loss_ce": 0.2697365880012512,
      "loss_lvr": 1.268803596496582,
      "loss_mode_switch": 0.0,
      "loss_total": 0.39661693572998047,
      "step": 167
    },
    {
      "epoch": 0.0672,
      "grad_norm": 9.148942947387695,
      "learning_rate": 9.964528647082447e-06,
      "loss": 0.3312,
      "step": 168
    },
    {
      "batch_size": 4,
      "epoch": 0.0672,
      "step": 168,
      "tokens_per_device": 3304
    },
    {
      "epoch": 0.0672,
      "loss_ce": 0.18332062661647797,
      "loss_lvr": 0.7361084818840027,
      "loss_mode_switch": 0.0,
      "loss_total": 0.25693148374557495,
      "step": 168
    },
    {
      "epoch": 0.0676,
      "grad_norm": 7.632505416870117,
      "learning_rate": 9.96375427793018e-06,
      "loss": 0.173,
      "step": 169
    },
    {
      "batch_size": 4,
      "epoch": 0.0676,
      "step": 169,
      "tokens_per_device": 4220
    },
    {
      "epoch": 0.0676,
      "loss_ce": 0.8488138914108276,
      "loss_lvr": 0.961979866027832,
      "loss_mode_switch": 0.0,
      "loss_total": 0.9450118541717529,
      "step": 169
    },
    {
      "epoch": 0.068,
      "grad_norm": 12.118610382080078,
      "learning_rate": 9.962971577982428e-06,
      "loss": 0.7084,
      "step": 170
    },
    {
      "batch_size": 4,
      "epoch": 0.068,
      "step": 170,
      "tokens_per_device": 5652
    },
    {
      "epoch": 0.068,
      "loss_ce": 0.3035632371902466,
      "loss_lvr": 0.879654586315155,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3915286958217621,
      "step": 170
    },
    {
      "epoch": 0.0684,
      "grad_norm": 13.89213752746582,
      "learning_rate": 9.962180548552812e-06,
      "loss": 0.7509,
      "step": 171
    },
    {
      "batch_size": 1,
      "epoch": 0.0684,
      "step": 171,
      "tokens_per_device": 5011
    },
    {
      "epoch": 0.0684,
      "loss_ce": 0.05361044779419899,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.05361044779419899,
      "step": 171
    },
    {
      "epoch": 0.0688,
      "grad_norm": 6.800137519836426,
      "learning_rate": 9.96138119096894e-06,
      "loss": 0.0536,
      "step": 172
    },
    {
      "batch_size": 4,
      "epoch": 0.0688,
      "step": 172,
      "tokens_per_device": 5572
    },
    {
      "epoch": 0.0688,
      "loss_ce": 0.14795224368572235,
      "loss_lvr": 0.6949019432067871,
      "loss_mode_switch": 0.0,
      "loss_total": 0.21744243800640106,
      "step": 172
    },
    {
      "epoch": 0.0692,
      "grad_norm": 9.969058990478516,
      "learning_rate": 9.960573506572391e-06,
      "loss": 0.3052,
      "step": 173
    },
    {
      "batch_size": 4,
      "epoch": 0.0692,
      "step": 173,
      "tokens_per_device": 4172
    },
    {
      "epoch": 0.0692,
      "loss_ce": 0.5400973558425903,
      "loss_lvr": 0.9764474630355835,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6377421021461487,
      "step": 173
    },
    {
      "epoch": 0.0696,
      "grad_norm": 13.394407272338867,
      "learning_rate": 9.959757496718723e-06,
      "loss": 0.5929,
      "step": 174
    },
    {
      "batch_size": 4,
      "epoch": 0.0696,
      "step": 174,
      "tokens_per_device": 9612
    },
    {
      "epoch": 0.0696,
      "loss_ce": 0.22796152532100677,
      "loss_lvr": 0.8940377235412598,
      "loss_mode_switch": 0.0,
      "loss_total": 0.31736528873443604,
      "step": 174
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.31251049041748,
      "learning_rate": 9.958933162777468e-06,
      "loss": 0.4219,
      "step": 175
    },
    {
      "batch_size": 4,
      "epoch": 0.07,
      "step": 175,
      "tokens_per_device": 4200
    },
    {
      "epoch": 0.07,
      "loss_ce": 0.6076923608779907,
      "loss_lvr": 1.1480004787445068,
      "loss_mode_switch": 0.0,
      "loss_total": 0.7224923968315125,
      "step": 175
    },
    {
      "epoch": 0.0704,
      "grad_norm": 33.97924041748047,
      "learning_rate": 9.958100506132127e-06,
      "loss": 1.651,
      "step": 176
    },
    {
      "batch_size": 4,
      "epoch": 0.0704,
      "step": 176,
      "tokens_per_device": 5416
    },
    {
      "epoch": 0.0704,
      "loss_ce": 0.5607596635818481,
      "loss_lvr": 0.7510778307914734,
      "loss_mode_switch": 0.0,
      "loss_total": 0.6358674764633179,
      "step": 176
    },
    {
      "epoch": 0.0708,
      "grad_norm": 9.973333358764648,
      "learning_rate": 9.957259528180166e-06,
      "loss": 0.495,
      "step": 177
    },
    {
      "batch_size": 1,
      "epoch": 0.0708,
      "step": 177,
      "tokens_per_device": 5097
    },
    {
      "epoch": 0.0708,
      "loss_ce": 0.0372319370508194,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.0372319370508194,
      "step": 177
    },
    {
      "epoch": 0.0712,
      "grad_norm": 9.480403900146484,
      "learning_rate": 9.956410230333023e-06,
      "loss": 0.0725,
      "step": 178
    },
    {
      "batch_size": 4,
      "epoch": 0.0712,
      "step": 178,
      "tokens_per_device": 5812
    },
    {
      "epoch": 0.0712,
      "loss_ce": 0.2876702547073364,
      "loss_lvr": 0.7084423899650574,
      "loss_mode_switch": 0.0,
      "loss_total": 0.3585144877433777,
      "step": 178
    },
    {
      "epoch": 0.0716,
      "grad_norm": 8.531497955322266,
      "learning_rate": 9.955552614016093e-06,
      "loss": 0.2086,
      "step": 179
    },
    {
      "batch_size": 4,
      "epoch": 0.0716,
      "step": 179,
      "tokens_per_device": 4468
    },
    {
      "epoch": 0.0716,
      "loss_ce": 0.6258969306945801,
      "loss_lvr": 0.9620586633682251,
      "loss_mode_switch": 0.0,
      "loss_total": 0.7221028208732605,
      "step": 179
    },
    {
      "epoch": 0.072,
      "grad_norm": 11.116934776306152,
      "learning_rate": 9.954686680668737e-06,
      "loss": 0.4476,
      "step": 180
    },
    {
      "batch_size": 4,
      "epoch": 0.072,
      "step": 180,
      "tokens_per_device": 3724
    },
    {
      "epoch": 0.072,
      "loss_ce": 0.5414818525314331,
      "loss_lvr": 1.0110447406768799,
      "loss_mode_switch": 0.0,
      "loss_total": 0.642586350440979,
      "step": 180
    },
    {
      "epoch": 0.0724,
      "grad_norm": 9.602731704711914,
      "learning_rate": 9.953812431744274e-06,
      "loss": 0.5806,
      "step": 181
    },
    {
      "batch_size": 4,
      "epoch": 0.0724,
      "step": 181,
      "tokens_per_device": 3772
    },
    {
      "epoch": 0.0724,
      "loss_ce": 0.25464725494384766,
      "loss_lvr": 0.7644992470741272,
      "loss_mode_switch": 0.0,
      "loss_total": 0.33109718561172485,
      "step": 181
    },
    {
      "epoch": 0.0728,
      "grad_norm": 5.706144332885742,
      "learning_rate": 9.95292986870998e-06,
      "loss": 0.1815,
      "step": 182
    },
    {
      "batch_size": 1,
      "epoch": 0.0728,
      "step": 182,
      "tokens_per_device": 4863
    },
    {
      "epoch": 0.0728,
      "loss_ce": 0.17501778900623322,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.17501778900623322,
      "step": 182
    },
    {
      "epoch": 0.0732,
      "grad_norm": 12.050512313842773,
      "learning_rate": 9.952038993047076e-06,
      "loss": 0.1161,
      "step": 183
    },
    {
      "batch_size": 4,
      "epoch": 0.0732,
      "step": 183,
      "tokens_per_device": 4164
    },
    {
      "epoch": 0.0732,
      "loss_ce": 0.45779234170913696,
      "loss_lvr": 1.0654261112213135,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5643349289894104,
      "step": 183
    },
    {
      "epoch": 0.0736,
      "grad_norm": 10.899062156677246,
      "learning_rate": 9.951139806250747e-06,
      "loss": 0.589,
      "step": 184
    },
    {
      "batch_size": 1,
      "epoch": 0.0736,
      "step": 184,
      "tokens_per_device": 4808
    },
    {
      "epoch": 0.0736,
      "loss_ce": 0.10112921893596649,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.10112921893596649,
      "step": 184
    },
    {
      "epoch": 0.074,
      "grad_norm": 9.077938079833984,
      "learning_rate": 9.950232309830121e-06,
      "loss": 0.2827,
      "step": 185
    },
    {
      "batch_size": 4,
      "epoch": 0.074,
      "step": 185,
      "tokens_per_device": 3924
    },
    {
      "epoch": 0.074,
      "loss_ce": 0.4965969920158386,
      "loss_lvr": 0.9547463655471802,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5920716524124146,
      "step": 185
    },
    {
      "epoch": 0.0744,
      "grad_norm": 11.916533470153809,
      "learning_rate": 9.94931650530827e-06,
      "loss": 0.6775,
      "step": 186
    },
    {
      "batch_size": 1,
      "epoch": 0.0744,
      "step": 186,
      "tokens_per_device": 4918
    },
    {
      "epoch": 0.0744,
      "loss_ce": 0.1463746577501297,
      "loss_lvr": 0.21429818868637085,
      "loss_mode_switch": 0.0,
      "loss_total": 0.16780447959899902,
      "step": 186
    },
    {
      "epoch": 0.0748,
      "grad_norm": 14.6821928024292,
      "learning_rate": 9.948392394222214e-06,
      "loss": 0.1618,
      "step": 187
    },
    {
      "batch_size": 4,
      "epoch": 0.0748,
      "step": 187,
      "tokens_per_device": 4704
    },
    {
      "epoch": 0.0748,
      "loss_ce": 0.6345129013061523,
      "loss_lvr": 0.7145840525627136,
      "loss_mode_switch": 0.0,
      "loss_total": 0.7059713006019592,
      "step": 187
    },
    {
      "epoch": 0.0752,
      "grad_norm": 10.782206535339355,
      "learning_rate": 9.947459978122912e-06,
      "loss": 0.6476,
      "step": 188
    },
    {
      "batch_size": 4,
      "epoch": 0.0752,
      "step": 188,
      "tokens_per_device": 9684
    },
    {
      "epoch": 0.0752,
      "loss_ce": 0.653563380241394,
      "loss_lvr": 0.5863479971885681,
      "loss_mode_switch": 0.0,
      "loss_total": 0.7121981978416443,
      "step": 188
    },
    {
      "epoch": 0.0756,
      "grad_norm": 13.491598129272461,
      "learning_rate": 9.946519258575263e-06,
      "loss": 0.4824,
      "step": 189
    },
    {
      "batch_size": 4,
      "epoch": 0.0756,
      "step": 189,
      "tokens_per_device": 2892
    },
    {
      "epoch": 0.0756,
      "loss_ce": 0.6633128523826599,
      "loss_lvr": 0.7712944746017456,
      "loss_mode_switch": 0.0,
      "loss_total": 0.7404422760009766,
      "step": 189
    },
    {
      "epoch": 0.076,
      "grad_norm": 11.317464828491211,
      "learning_rate": 9.945570237158098e-06,
      "loss": 0.6923,
      "step": 190
    },
    {
      "batch_size": 4,
      "epoch": 0.076,
      "step": 190,
      "tokens_per_device": 5664
    },
    {
      "epoch": 0.076,
      "loss_ce": 0.17480017244815826,
      "loss_lvr": 0.9979178309440613,
      "loss_mode_switch": 0.0,
      "loss_total": 0.27459195256233215,
      "step": 190
    },
    {
      "epoch": 0.0764,
      "grad_norm": 7.234347343444824,
      "learning_rate": 9.944612915464183e-06,
      "loss": 0.3263,
      "step": 191
    },
    {
      "batch_size": 4,
      "epoch": 0.0764,
      "step": 191,
      "tokens_per_device": 4488
    },
    {
      "epoch": 0.0764,
      "loss_ce": 0.16864602267742157,
      "loss_lvr": 0.9448134899139404,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2631273865699768,
      "step": 191
    },
    {
      "epoch": 0.0768,
      "grad_norm": 7.423867225646973,
      "learning_rate": 9.943647295100219e-06,
      "loss": 0.33,
      "step": 192
    },
    {
      "batch_size": 4,
      "epoch": 0.0768,
      "step": 192,
      "tokens_per_device": 5252
    },
    {
      "epoch": 0.0768,
      "loss_ce": 0.179182231426239,
      "loss_lvr": 1.0863101482391357,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2878132462501526,
      "step": 192
    },
    {
      "epoch": 0.0772,
      "grad_norm": 8.794830322265625,
      "learning_rate": 9.94267337768683e-06,
      "loss": 0.3282,
      "step": 193
    },
    {
      "batch_size": 4,
      "epoch": 0.0772,
      "step": 193,
      "tokens_per_device": 1384
    },
    {
      "epoch": 0.0772,
      "loss_ce": 0.4221283197402954,
      "loss_lvr": 0.863936185836792,
      "loss_mode_switch": 0.0,
      "loss_total": 0.5085219144821167,
      "step": 193
    },
    {
      "epoch": 0.0776,
      "grad_norm": 9.082275390625,
      "learning_rate": 9.941691164858565e-06,
      "loss": 0.4166,
      "step": 194
    },
    {
      "batch_size": 4,
      "epoch": 0.0776,
      "step": 194,
      "tokens_per_device": 5852
    },
    {
      "epoch": 0.0776,
      "loss_ce": 0.13656996190547943,
      "loss_lvr": 0.7093530893325806,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2075052708387375,
      "step": 194
    },
    {
      "epoch": 0.078,
      "grad_norm": 6.745866298675537,
      "learning_rate": 9.940700658263897e-06,
      "loss": 0.2173,
      "step": 195
    },
    {
      "batch_size": 1,
      "epoch": 0.078,
      "step": 195,
      "tokens_per_device": 4872
    },
    {
      "epoch": 0.078,
      "loss_ce": 0.16997094452381134,
      "loss_lvr": 0.0,
      "loss_mode_switch": 0.0,
      "loss_total": 0.16997094452381134,
      "step": 195
    },
    {
      "epoch": 0.0784,
      "grad_norm": 15.79223918914795,
      "learning_rate": 9.93970185956522e-06,
      "loss": 0.187,
      "step": 196
    },
    {
      "batch_size": 4,
      "epoch": 0.0784,
      "step": 196,
      "tokens_per_device": 4228
    },
    {
      "epoch": 0.0784,
      "loss_ce": 0.10435216128826141,
      "loss_lvr": 1.3260124921798706,
      "loss_mode_switch": 0.0,
      "loss_total": 0.23695340752601624,
      "step": 196
    },
    {
      "epoch": 0.0788,
      "grad_norm": 11.199546813964844,
      "learning_rate": 9.938694770438843e-06,
      "loss": 0.2872,
      "step": 197
    },
    {
      "batch_size": 1,
      "epoch": 0.0788,
      "step": 197,
      "tokens_per_device": 5243
    },
    {
      "epoch": 0.0788,
      "loss_ce": 0.08581629395484924,
      "loss_lvr": 0.5192096829414368,
      "loss_mode_switch": 0.0,
      "loss_total": 0.13773725926876068,
      "step": 197
    },
    {
      "epoch": 0.0792,
      "grad_norm": 4.618615627288818,
      "learning_rate": 9.937679392574991e-06,
      "loss": 0.0829,
      "step": 198
    },
    {
      "batch_size": 4,
      "epoch": 0.0792,
      "step": 198,
      "tokens_per_device": 1272
    },
    {
      "epoch": 0.0792,
      "loss_ce": 0.2846511900424957,
      "loss_lvr": 1.2284373044967651,
      "loss_mode_switch": 0.0,
      "loss_total": 0.4074949324131012,
      "step": 198
    },
    {
      "epoch": 0.0796,
      "grad_norm": 14.528177261352539,
      "learning_rate": 9.936655727677795e-06,
      "loss": 0.3228,
      "step": 199
    },
    {
      "batch_size": 4,
      "epoch": 0.0796,
      "step": 199,
      "tokens_per_device": 4360
    },
    {
      "epoch": 0.0796,
      "loss_ce": 0.17171026766300201,
      "loss_lvr": 1.0165114402770996,
      "loss_mode_switch": 0.0,
      "loss_total": 0.2733614146709442,
      "step": 199
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.068370819091797,
      "learning_rate": 9.9356237774653e-06,
      "loss": 0.3675,
      "step": 200
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 8.8584263565312e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}