| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 20.0, | |
| "eval_steps": 500, | |
| "global_step": 689340, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00013054830287206266, | |
| "loss": 18.218, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0002610966057441253, | |
| "loss": 4.1185, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0002990742124651621, | |
| "loss": 3.2662, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00029775542680299976, | |
| "loss": 3.0539, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0002964366411408375, | |
| "loss": 2.9401, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0002951178554786752, | |
| "loss": 2.8804, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00029379906981651295, | |
| "loss": 2.8248, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0002924802841543507, | |
| "loss": 2.7855, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0002911614984921884, | |
| "loss": 2.7762, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0002898427128300261, | |
| "loss": 2.7404, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0002885239271678638, | |
| "loss": 2.7029, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00028720514150570154, | |
| "loss": 2.6612, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0002858863558435392, | |
| "loss": 2.6141, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.000284567570181377, | |
| "loss": 2.5986, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.00028324878451921467, | |
| "loss": 2.5744, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0002819299988570524, | |
| "loss": 2.5785, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00028061121319489013, | |
| "loss": 2.5643, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.0002792924275327278, | |
| "loss": 2.569, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00027797364187056553, | |
| "loss": 2.5455, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00027665485620840326, | |
| "loss": 2.5417, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.000275336070546241, | |
| "loss": 2.5095, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.0002740172848840787, | |
| "loss": 2.5052, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.00027269849922191645, | |
| "loss": 2.5086, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.0002713797135597541, | |
| "loss": 2.4391, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.00027006092789759185, | |
| "loss": 2.4283, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.0002687421422354296, | |
| "loss": 2.4079, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 0.0002674233565732673, | |
| "loss": 2.4065, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.00026610457091110504, | |
| "loss": 2.3893, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 0.00026478578524894277, | |
| "loss": 2.4166, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 0.00026346699958678044, | |
| "loss": 2.3907, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.00026214821392461817, | |
| "loss": 2.3829, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 0.0002608294282624559, | |
| "loss": 2.3767, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 0.00025951064260029363, | |
| "loss": 2.3518, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 0.00025819185693813136, | |
| "loss": 2.3734, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 0.0002568730712759691, | |
| "loss": 2.3126, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 0.00025555428561380676, | |
| "loss": 2.2802, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 0.0002542354999516445, | |
| "loss": 2.2739, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 0.0002529167142894822, | |
| "loss": 2.2588, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 0.00025159792862731995, | |
| "loss": 2.2573, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 0.0002502791429651577, | |
| "loss": 2.2541, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 0.0002489603573029954, | |
| "loss": 2.2843, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 0.0002476415716408331, | |
| "loss": 2.2548, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 0.0002463227859786708, | |
| "loss": 2.252, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 0.00024500400031650854, | |
| "loss": 2.2677, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 0.00024368521465434624, | |
| "loss": 2.2497, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.00024236642899218397, | |
| "loss": 2.2434, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 0.0002410476433300217, | |
| "loss": 2.1625, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 0.0002397288576678594, | |
| "loss": 2.1795, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 0.00023841007200569713, | |
| "loss": 2.1666, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 0.00023709128634353483, | |
| "loss": 2.1749, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 0.00023577250068137256, | |
| "loss": 2.1931, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 0.0002344537150192103, | |
| "loss": 2.1797, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 0.000233134929357048, | |
| "loss": 2.1573, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 0.00023181614369488572, | |
| "loss": 2.1811, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 0.00023049735803272345, | |
| "loss": 2.1539, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 0.00022917857237056115, | |
| "loss": 2.1657, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 0.00022785978670839888, | |
| "loss": 2.1662, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 0.0002265410010462366, | |
| "loss": 2.1237, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 0.0002252222153840743, | |
| "loss": 2.0812, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 0.00022390342972191204, | |
| "loss": 2.0917, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 0.00022258464405974977, | |
| "loss": 2.09, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 0.00022126585839758747, | |
| "loss": 2.0931, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 0.0002199470727354252, | |
| "loss": 2.0907, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 0.00021862828707326293, | |
| "loss": 2.0855, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 0.00021730950141110063, | |
| "loss": 2.0946, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 0.00021599071574893836, | |
| "loss": 2.0904, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 0.0002146719300867761, | |
| "loss": 2.0788, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 0.0002133531444246138, | |
| "loss": 2.0771, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 0.00021203435876245152, | |
| "loss": 2.0746, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 0.00021071557310028925, | |
| "loss": 2.0091, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 0.00020939678743812695, | |
| "loss": 2.0144, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 0.00020807800177596468, | |
| "loss": 2.0083, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 0.0002067592161138024, | |
| "loss": 2.0337, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 0.0002054404304516401, | |
| "loss": 2.0169, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 0.00020412164478947784, | |
| "loss": 2.0264, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 0.00020280285912731557, | |
| "loss": 2.0089, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 0.00020148407346515327, | |
| "loss": 2.0148, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 0.000200165287802991, | |
| "loss": 2.0224, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 0.00019884650214082873, | |
| "loss": 2.0242, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 0.00019752771647866643, | |
| "loss": 2.0142, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 0.00019620893081650416, | |
| "loss": 1.9622, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 0.0001948901451543419, | |
| "loss": 1.9643, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "learning_rate": 0.0001935713594921796, | |
| "loss": 1.9589, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 0.00019225257383001732, | |
| "loss": 1.9411, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 0.000190933788167855, | |
| "loss": 1.956, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 0.00018961500250569275, | |
| "loss": 1.9596, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 0.00018829621684353048, | |
| "loss": 1.9373, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 0.00018697743118136815, | |
| "loss": 1.9532, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 0.00018565864551920588, | |
| "loss": 1.9669, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 0.00018433985985704364, | |
| "loss": 1.957, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 0.0001830210741948813, | |
| "loss": 1.977, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 0.00018170228853271904, | |
| "loss": 1.9712, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 0.0001803835028705568, | |
| "loss": 1.8922, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 0.00017906471720839447, | |
| "loss": 1.9054, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 0.0001777459315462322, | |
| "loss": 1.8847, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "learning_rate": 0.00017642714588406993, | |
| "loss": 1.896, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 0.00017510836022190763, | |
| "loss": 1.9139, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 8.53, | |
| "learning_rate": 0.00017378957455974536, | |
| "loss": 1.9086, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 0.0001724707888975831, | |
| "loss": 1.9158, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "learning_rate": 0.0001711520032354208, | |
| "loss": 1.908, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 0.00016983321757325852, | |
| "loss": 1.9034, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "learning_rate": 0.00016851443191109625, | |
| "loss": 1.9022, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "learning_rate": 0.00016719564624893395, | |
| "loss": 1.9045, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 0.00016587686058677168, | |
| "loss": 1.8607, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 9.14, | |
| "learning_rate": 0.0001645580749246094, | |
| "loss": 1.8439, | |
| "step": 315000 | |
| }, | |
| { | |
| "epoch": 9.23, | |
| "learning_rate": 0.0001632392892624471, | |
| "loss": 1.8252, | |
| "step": 318000 | |
| }, | |
| { | |
| "epoch": 9.31, | |
| "learning_rate": 0.00016192050360028484, | |
| "loss": 1.844, | |
| "step": 321000 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 0.00016060171793812257, | |
| "loss": 1.839, | |
| "step": 324000 | |
| }, | |
| { | |
| "epoch": 9.49, | |
| "learning_rate": 0.00015928293227596027, | |
| "loss": 1.8442, | |
| "step": 327000 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 0.000157964146613798, | |
| "loss": 1.8378, | |
| "step": 330000 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 0.00015664536095163573, | |
| "loss": 1.8436, | |
| "step": 333000 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 0.00015532657528947343, | |
| "loss": 1.8399, | |
| "step": 336000 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 0.00015400778962731116, | |
| "loss": 1.8357, | |
| "step": 339000 | |
| }, | |
| { | |
| "epoch": 9.92, | |
| "learning_rate": 0.0001526890039651489, | |
| "loss": 1.8406, | |
| "step": 342000 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 0.0001513702183029866, | |
| "loss": 1.8385, | |
| "step": 345000 | |
| }, | |
| { | |
| "epoch": 10.1, | |
| "learning_rate": 0.00015005143264082432, | |
| "loss": 1.7886, | |
| "step": 348000 | |
| }, | |
| { | |
| "epoch": 10.18, | |
| "learning_rate": 0.00014873264697866202, | |
| "loss": 1.7909, | |
| "step": 351000 | |
| }, | |
| { | |
| "epoch": 10.27, | |
| "learning_rate": 0.00014741386131649975, | |
| "loss": 1.794, | |
| "step": 354000 | |
| }, | |
| { | |
| "epoch": 10.36, | |
| "learning_rate": 0.00014609507565433748, | |
| "loss": 1.7712, | |
| "step": 357000 | |
| }, | |
| { | |
| "epoch": 10.44, | |
| "learning_rate": 0.00014477628999217518, | |
| "loss": 1.7875, | |
| "step": 360000 | |
| }, | |
| { | |
| "epoch": 10.53, | |
| "learning_rate": 0.0001434575043300129, | |
| "loss": 1.786, | |
| "step": 363000 | |
| }, | |
| { | |
| "epoch": 10.62, | |
| "learning_rate": 0.00014213871866785064, | |
| "loss": 1.7925, | |
| "step": 366000 | |
| }, | |
| { | |
| "epoch": 10.71, | |
| "learning_rate": 0.00014081993300568834, | |
| "loss": 1.7875, | |
| "step": 369000 | |
| }, | |
| { | |
| "epoch": 10.79, | |
| "learning_rate": 0.00013950114734352607, | |
| "loss": 1.7752, | |
| "step": 372000 | |
| }, | |
| { | |
| "epoch": 10.88, | |
| "learning_rate": 0.0001381823616813638, | |
| "loss": 1.791, | |
| "step": 375000 | |
| }, | |
| { | |
| "epoch": 10.97, | |
| "learning_rate": 0.0001368635760192015, | |
| "loss": 1.7857, | |
| "step": 378000 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 0.00013554479035703923, | |
| "loss": 1.7407, | |
| "step": 381000 | |
| }, | |
| { | |
| "epoch": 11.14, | |
| "learning_rate": 0.00013422600469487693, | |
| "loss": 1.7115, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 11.23, | |
| "learning_rate": 0.00013290721903271466, | |
| "loss": 1.7294, | |
| "step": 387000 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "learning_rate": 0.0001315884333705524, | |
| "loss": 1.7205, | |
| "step": 390000 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "learning_rate": 0.0001302696477083901, | |
| "loss": 1.7281, | |
| "step": 393000 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "learning_rate": 0.00012895086204622782, | |
| "loss": 1.7418, | |
| "step": 396000 | |
| }, | |
| { | |
| "epoch": 11.58, | |
| "learning_rate": 0.00012763207638406555, | |
| "loss": 1.7316, | |
| "step": 399000 | |
| }, | |
| { | |
| "epoch": 11.66, | |
| "learning_rate": 0.00012631329072190325, | |
| "loss": 1.7348, | |
| "step": 402000 | |
| }, | |
| { | |
| "epoch": 11.75, | |
| "learning_rate": 0.00012499450505974098, | |
| "loss": 1.7392, | |
| "step": 405000 | |
| }, | |
| { | |
| "epoch": 11.84, | |
| "learning_rate": 0.0001236757193975787, | |
| "loss": 1.7341, | |
| "step": 408000 | |
| }, | |
| { | |
| "epoch": 11.92, | |
| "learning_rate": 0.0001223569337354164, | |
| "loss": 1.7347, | |
| "step": 411000 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 0.00012103814807325414, | |
| "loss": 1.7196, | |
| "step": 414000 | |
| }, | |
| { | |
| "epoch": 12.1, | |
| "learning_rate": 0.00011971936241109186, | |
| "loss": 1.6614, | |
| "step": 417000 | |
| }, | |
| { | |
| "epoch": 12.19, | |
| "learning_rate": 0.00011840057674892957, | |
| "loss": 1.6671, | |
| "step": 420000 | |
| }, | |
| { | |
| "epoch": 12.27, | |
| "learning_rate": 0.0001170817910867673, | |
| "loss": 1.6665, | |
| "step": 423000 | |
| }, | |
| { | |
| "epoch": 12.36, | |
| "learning_rate": 0.00011576300542460502, | |
| "loss": 1.6775, | |
| "step": 426000 | |
| }, | |
| { | |
| "epoch": 12.45, | |
| "learning_rate": 0.00011444421976244273, | |
| "loss": 1.6646, | |
| "step": 429000 | |
| }, | |
| { | |
| "epoch": 12.53, | |
| "learning_rate": 0.00011312543410028046, | |
| "loss": 1.6779, | |
| "step": 432000 | |
| }, | |
| { | |
| "epoch": 12.62, | |
| "learning_rate": 0.00011180664843811818, | |
| "loss": 1.6802, | |
| "step": 435000 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "learning_rate": 0.00011048786277595589, | |
| "loss": 1.6759, | |
| "step": 438000 | |
| }, | |
| { | |
| "epoch": 12.79, | |
| "learning_rate": 0.0001091690771137936, | |
| "loss": 1.6729, | |
| "step": 441000 | |
| }, | |
| { | |
| "epoch": 12.88, | |
| "learning_rate": 0.00010785029145163134, | |
| "loss": 1.6769, | |
| "step": 444000 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 0.00010653150578946905, | |
| "loss": 1.6721, | |
| "step": 447000 | |
| }, | |
| { | |
| "epoch": 13.06, | |
| "learning_rate": 0.00010521272012730677, | |
| "loss": 1.6302, | |
| "step": 450000 | |
| }, | |
| { | |
| "epoch": 13.14, | |
| "learning_rate": 0.0001038939344651445, | |
| "loss": 1.6112, | |
| "step": 453000 | |
| }, | |
| { | |
| "epoch": 13.23, | |
| "learning_rate": 0.00010257514880298221, | |
| "loss": 1.5991, | |
| "step": 456000 | |
| }, | |
| { | |
| "epoch": 13.32, | |
| "learning_rate": 0.00010125636314081991, | |
| "loss": 1.6221, | |
| "step": 459000 | |
| }, | |
| { | |
| "epoch": 13.4, | |
| "learning_rate": 9.993757747865765e-05, | |
| "loss": 1.6168, | |
| "step": 462000 | |
| }, | |
| { | |
| "epoch": 13.49, | |
| "learning_rate": 9.861879181649536e-05, | |
| "loss": 1.6166, | |
| "step": 465000 | |
| }, | |
| { | |
| "epoch": 13.58, | |
| "learning_rate": 9.730000615433307e-05, | |
| "loss": 1.6234, | |
| "step": 468000 | |
| }, | |
| { | |
| "epoch": 13.67, | |
| "learning_rate": 9.59812204921708e-05, | |
| "loss": 1.6245, | |
| "step": 471000 | |
| }, | |
| { | |
| "epoch": 13.75, | |
| "learning_rate": 9.466243483000852e-05, | |
| "loss": 1.618, | |
| "step": 474000 | |
| }, | |
| { | |
| "epoch": 13.84, | |
| "learning_rate": 9.334364916784623e-05, | |
| "loss": 1.6341, | |
| "step": 477000 | |
| }, | |
| { | |
| "epoch": 13.93, | |
| "learning_rate": 9.202486350568396e-05, | |
| "loss": 1.6353, | |
| "step": 480000 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 9.070607784352168e-05, | |
| "loss": 1.6104, | |
| "step": 483000 | |
| }, | |
| { | |
| "epoch": 14.1, | |
| "learning_rate": 8.938729218135939e-05, | |
| "loss": 1.562, | |
| "step": 486000 | |
| }, | |
| { | |
| "epoch": 14.19, | |
| "learning_rate": 8.806850651919711e-05, | |
| "loss": 1.5654, | |
| "step": 489000 | |
| }, | |
| { | |
| "epoch": 14.27, | |
| "learning_rate": 8.674972085703484e-05, | |
| "loss": 1.564, | |
| "step": 492000 | |
| }, | |
| { | |
| "epoch": 14.36, | |
| "learning_rate": 8.543093519487255e-05, | |
| "loss": 1.5606, | |
| "step": 495000 | |
| }, | |
| { | |
| "epoch": 14.45, | |
| "learning_rate": 8.411214953271027e-05, | |
| "loss": 1.5764, | |
| "step": 498000 | |
| }, | |
| { | |
| "epoch": 14.54, | |
| "learning_rate": 8.2793363870548e-05, | |
| "loss": 1.5718, | |
| "step": 501000 | |
| }, | |
| { | |
| "epoch": 14.62, | |
| "learning_rate": 8.147457820838571e-05, | |
| "loss": 1.5768, | |
| "step": 504000 | |
| }, | |
| { | |
| "epoch": 14.71, | |
| "learning_rate": 8.015579254622343e-05, | |
| "loss": 1.5686, | |
| "step": 507000 | |
| }, | |
| { | |
| "epoch": 14.8, | |
| "learning_rate": 7.883700688406116e-05, | |
| "loss": 1.5753, | |
| "step": 510000 | |
| }, | |
| { | |
| "epoch": 14.88, | |
| "learning_rate": 7.751822122189887e-05, | |
| "loss": 1.5588, | |
| "step": 513000 | |
| }, | |
| { | |
| "epoch": 14.97, | |
| "learning_rate": 7.619943555973659e-05, | |
| "loss": 1.5588, | |
| "step": 516000 | |
| }, | |
| { | |
| "epoch": 15.06, | |
| "learning_rate": 7.48806498975743e-05, | |
| "loss": 1.5413, | |
| "step": 519000 | |
| }, | |
| { | |
| "epoch": 15.14, | |
| "learning_rate": 7.356186423541203e-05, | |
| "loss": 1.5102, | |
| "step": 522000 | |
| }, | |
| { | |
| "epoch": 15.23, | |
| "learning_rate": 7.224307857324975e-05, | |
| "loss": 1.5183, | |
| "step": 525000 | |
| }, | |
| { | |
| "epoch": 15.32, | |
| "learning_rate": 7.092429291108746e-05, | |
| "loss": 1.5267, | |
| "step": 528000 | |
| }, | |
| { | |
| "epoch": 15.41, | |
| "learning_rate": 6.960550724892519e-05, | |
| "loss": 1.5185, | |
| "step": 531000 | |
| }, | |
| { | |
| "epoch": 15.49, | |
| "learning_rate": 6.828672158676289e-05, | |
| "loss": 1.5195, | |
| "step": 534000 | |
| }, | |
| { | |
| "epoch": 15.58, | |
| "learning_rate": 6.696793592460062e-05, | |
| "loss": 1.5154, | |
| "step": 537000 | |
| }, | |
| { | |
| "epoch": 15.67, | |
| "learning_rate": 6.564915026243834e-05, | |
| "loss": 1.5312, | |
| "step": 540000 | |
| }, | |
| { | |
| "epoch": 15.75, | |
| "learning_rate": 6.433036460027605e-05, | |
| "loss": 1.5191, | |
| "step": 543000 | |
| }, | |
| { | |
| "epoch": 15.84, | |
| "learning_rate": 6.301157893811378e-05, | |
| "loss": 1.5209, | |
| "step": 546000 | |
| }, | |
| { | |
| "epoch": 15.93, | |
| "learning_rate": 6.16927932759515e-05, | |
| "loss": 1.5145, | |
| "step": 549000 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 6.037400761378922e-05, | |
| "loss": 1.5147, | |
| "step": 552000 | |
| }, | |
| { | |
| "epoch": 16.1, | |
| "learning_rate": 5.905522195162694e-05, | |
| "loss": 1.47, | |
| "step": 555000 | |
| }, | |
| { | |
| "epoch": 16.19, | |
| "learning_rate": 5.773643628946465e-05, | |
| "loss": 1.4766, | |
| "step": 558000 | |
| }, | |
| { | |
| "epoch": 16.28, | |
| "learning_rate": 5.641765062730237e-05, | |
| "loss": 1.4688, | |
| "step": 561000 | |
| }, | |
| { | |
| "epoch": 16.36, | |
| "learning_rate": 5.5098864965140094e-05, | |
| "loss": 1.4727, | |
| "step": 564000 | |
| }, | |
| { | |
| "epoch": 16.45, | |
| "learning_rate": 5.378007930297781e-05, | |
| "loss": 1.4673, | |
| "step": 567000 | |
| }, | |
| { | |
| "epoch": 16.54, | |
| "learning_rate": 5.246129364081553e-05, | |
| "loss": 1.4728, | |
| "step": 570000 | |
| }, | |
| { | |
| "epoch": 16.62, | |
| "learning_rate": 5.1142507978653254e-05, | |
| "loss": 1.4751, | |
| "step": 573000 | |
| }, | |
| { | |
| "epoch": 16.71, | |
| "learning_rate": 4.982372231649097e-05, | |
| "loss": 1.4652, | |
| "step": 576000 | |
| }, | |
| { | |
| "epoch": 16.8, | |
| "learning_rate": 4.850493665432869e-05, | |
| "loss": 1.4783, | |
| "step": 579000 | |
| }, | |
| { | |
| "epoch": 16.89, | |
| "learning_rate": 4.718615099216641e-05, | |
| "loss": 1.4678, | |
| "step": 582000 | |
| }, | |
| { | |
| "epoch": 16.97, | |
| "learning_rate": 4.586736533000413e-05, | |
| "loss": 1.4717, | |
| "step": 585000 | |
| }, | |
| { | |
| "epoch": 17.06, | |
| "learning_rate": 4.454857966784185e-05, | |
| "loss": 1.4422, | |
| "step": 588000 | |
| }, | |
| { | |
| "epoch": 17.15, | |
| "learning_rate": 4.322979400567956e-05, | |
| "loss": 1.4209, | |
| "step": 591000 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "learning_rate": 4.191100834351728e-05, | |
| "loss": 1.4398, | |
| "step": 594000 | |
| }, | |
| { | |
| "epoch": 17.32, | |
| "learning_rate": 4.0592222681355005e-05, | |
| "loss": 1.4331, | |
| "step": 597000 | |
| }, | |
| { | |
| "epoch": 17.41, | |
| "learning_rate": 3.927343701919272e-05, | |
| "loss": 1.4253, | |
| "step": 600000 | |
| }, | |
| { | |
| "epoch": 17.49, | |
| "learning_rate": 3.795465135703044e-05, | |
| "loss": 1.4247, | |
| "step": 603000 | |
| }, | |
| { | |
| "epoch": 17.58, | |
| "learning_rate": 3.6635865694868164e-05, | |
| "loss": 1.432, | |
| "step": 606000 | |
| }, | |
| { | |
| "epoch": 17.67, | |
| "learning_rate": 3.531708003270588e-05, | |
| "loss": 1.4275, | |
| "step": 609000 | |
| }, | |
| { | |
| "epoch": 17.76, | |
| "learning_rate": 3.39982943705436e-05, | |
| "loss": 1.4305, | |
| "step": 612000 | |
| }, | |
| { | |
| "epoch": 17.84, | |
| "learning_rate": 3.267950870838132e-05, | |
| "loss": 1.4237, | |
| "step": 615000 | |
| }, | |
| { | |
| "epoch": 17.93, | |
| "learning_rate": 3.136072304621904e-05, | |
| "loss": 1.4339, | |
| "step": 618000 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 3.0041937384056755e-05, | |
| "loss": 1.4083, | |
| "step": 621000 | |
| }, | |
| { | |
| "epoch": 18.1, | |
| "learning_rate": 2.8723151721894477e-05, | |
| "loss": 1.383, | |
| "step": 624000 | |
| }, | |
| { | |
| "epoch": 18.19, | |
| "learning_rate": 2.7404366059732196e-05, | |
| "loss": 1.3949, | |
| "step": 627000 | |
| }, | |
| { | |
| "epoch": 18.28, | |
| "learning_rate": 2.6085580397569915e-05, | |
| "loss": 1.3857, | |
| "step": 630000 | |
| }, | |
| { | |
| "epoch": 18.37, | |
| "learning_rate": 2.4766794735407634e-05, | |
| "loss": 1.4049, | |
| "step": 633000 | |
| }, | |
| { | |
| "epoch": 18.45, | |
| "learning_rate": 2.3448009073245356e-05, | |
| "loss": 1.3972, | |
| "step": 636000 | |
| }, | |
| { | |
| "epoch": 18.54, | |
| "learning_rate": 2.2129223411083075e-05, | |
| "loss": 1.3983, | |
| "step": 639000 | |
| }, | |
| { | |
| "epoch": 18.63, | |
| "learning_rate": 2.081043774892079e-05, | |
| "loss": 1.3907, | |
| "step": 642000 | |
| }, | |
| { | |
| "epoch": 18.71, | |
| "learning_rate": 1.949165208675851e-05, | |
| "loss": 1.3888, | |
| "step": 645000 | |
| }, | |
| { | |
| "epoch": 18.8, | |
| "learning_rate": 1.817286642459623e-05, | |
| "loss": 1.3762, | |
| "step": 648000 | |
| }, | |
| { | |
| "epoch": 18.89, | |
| "learning_rate": 1.685408076243395e-05, | |
| "loss": 1.3926, | |
| "step": 651000 | |
| }, | |
| { | |
| "epoch": 18.97, | |
| "learning_rate": 1.553529510027167e-05, | |
| "loss": 1.3874, | |
| "step": 654000 | |
| }, | |
| { | |
| "epoch": 19.06, | |
| "learning_rate": 1.4216509438109387e-05, | |
| "loss": 1.375, | |
| "step": 657000 | |
| }, | |
| { | |
| "epoch": 19.15, | |
| "learning_rate": 1.2897723775947106e-05, | |
| "loss": 1.3655, | |
| "step": 660000 | |
| }, | |
| { | |
| "epoch": 19.24, | |
| "learning_rate": 1.1578938113784827e-05, | |
| "loss": 1.3625, | |
| "step": 663000 | |
| }, | |
| { | |
| "epoch": 19.32, | |
| "learning_rate": 1.0260152451622544e-05, | |
| "loss": 1.3705, | |
| "step": 666000 | |
| }, | |
| { | |
| "epoch": 19.41, | |
| "learning_rate": 8.941366789460264e-06, | |
| "loss": 1.3672, | |
| "step": 669000 | |
| }, | |
| { | |
| "epoch": 19.5, | |
| "learning_rate": 7.622581127297983e-06, | |
| "loss": 1.3545, | |
| "step": 672000 | |
| }, | |
| { | |
| "epoch": 19.58, | |
| "learning_rate": 6.303795465135703e-06, | |
| "loss": 1.3611, | |
| "step": 675000 | |
| }, | |
| { | |
| "epoch": 19.67, | |
| "learning_rate": 4.985009802973422e-06, | |
| "loss": 1.3574, | |
| "step": 678000 | |
| }, | |
| { | |
| "epoch": 19.76, | |
| "learning_rate": 3.666224140811141e-06, | |
| "loss": 1.3673, | |
| "step": 681000 | |
| }, | |
| { | |
| "epoch": 19.85, | |
| "learning_rate": 2.34743847864886e-06, | |
| "loss": 1.3629, | |
| "step": 684000 | |
| }, | |
| { | |
| "epoch": 19.93, | |
| "learning_rate": 1.028652816486579e-06, | |
| "loss": 1.3694, | |
| "step": 687000 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "step": 689340, | |
| "total_flos": 1.193067231326117e+21, | |
| "train_loss": 1.952947931640795, | |
| "train_runtime": 385301.7422, | |
| "train_samples_per_second": 28.625, | |
| "train_steps_per_second": 1.789 | |
| } | |
| ], | |
| "logging_steps": 3000, | |
| "max_steps": 689340, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "total_flos": 1.193067231326117e+21, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |