Upload trainer_log.jsonl with huggingface_hub
Browse files- trainer_log.jsonl +81 -0
trainer_log.jsonl
ADDED
|
@@ -0,0 +1,81 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"current_steps": 10, "total_steps": 799, "loss": 2.39, "learning_rate": 6.25e-08, "epoch": 0.012507817385866166, "percentage": 1.25, "elapsed_time": "0:00:13", "remaining_time": "0:18:05"}
|
| 2 |
+
{"current_steps": 20, "total_steps": 799, "loss": 2.4206, "learning_rate": 1.25e-07, "epoch": 0.025015634771732333, "percentage": 2.5, "elapsed_time": "0:00:20", "remaining_time": "0:13:20"}
|
| 3 |
+
{"current_steps": 30, "total_steps": 799, "loss": 2.3735, "learning_rate": 1.875e-07, "epoch": 0.0375234521575985, "percentage": 3.75, "elapsed_time": "0:00:26", "remaining_time": "0:11:07"}
|
| 4 |
+
{"current_steps": 40, "total_steps": 799, "loss": 2.3416, "learning_rate": 2.5e-07, "epoch": 0.050031269543464665, "percentage": 5.01, "elapsed_time": "0:00:32", "remaining_time": "0:10:11"}
|
| 5 |
+
{"current_steps": 50, "total_steps": 799, "loss": 2.3512, "learning_rate": 3.1249999999999997e-07, "epoch": 0.06253908692933083, "percentage": 6.26, "elapsed_time": "0:00:39", "remaining_time": "0:09:45"}
|
| 6 |
+
{"current_steps": 60, "total_steps": 799, "loss": 2.2364, "learning_rate": 3.75e-07, "epoch": 0.075046904315197, "percentage": 7.51, "elapsed_time": "0:00:45", "remaining_time": "0:09:26"}
|
| 7 |
+
{"current_steps": 70, "total_steps": 799, "loss": 2.2577, "learning_rate": 4.375e-07, "epoch": 0.08755472170106317, "percentage": 8.76, "elapsed_time": "0:00:52", "remaining_time": "0:09:05"}
|
| 8 |
+
{"current_steps": 80, "total_steps": 799, "loss": 2.2443, "learning_rate": 5e-07, "epoch": 0.10006253908692933, "percentage": 10.01, "elapsed_time": "0:00:59", "remaining_time": "0:08:57"}
|
| 9 |
+
{"current_steps": 90, "total_steps": 799, "loss": 2.0968, "learning_rate": 4.997613931640598e-07, "epoch": 0.1125703564727955, "percentage": 11.26, "elapsed_time": "0:01:05", "remaining_time": "0:08:34"}
|
| 10 |
+
{"current_steps": 100, "total_steps": 799, "loss": 2.0786, "learning_rate": 4.990460281220164e-07, "epoch": 0.12507817385866166, "percentage": 12.52, "elapsed_time": "0:01:12", "remaining_time": "0:08:24"}
|
| 11 |
+
{"current_steps": 110, "total_steps": 799, "loss": 1.9348, "learning_rate": 4.978552704017834e-07, "epoch": 0.13758599124452783, "percentage": 13.77, "elapsed_time": "0:01:18", "remaining_time": "0:08:12"}
|
| 12 |
+
{"current_steps": 120, "total_steps": 799, "loss": 1.9247, "learning_rate": 4.961913929868171e-07, "epoch": 0.150093808630394, "percentage": 15.02, "elapsed_time": "0:01:24", "remaining_time": "0:07:59"}
|
| 13 |
+
{"current_steps": 130, "total_steps": 799, "loss": 1.9058, "learning_rate": 4.940575719773205e-07, "epoch": 0.16260162601626016, "percentage": 16.27, "elapsed_time": "0:01:30", "remaining_time": "0:07:46"}
|
| 14 |
+
{"current_steps": 140, "total_steps": 799, "loss": 1.8596, "learning_rate": 4.914578805275297e-07, "epoch": 0.17510944340212634, "percentage": 17.52, "elapsed_time": "0:01:36", "remaining_time": "0:07:36"}
|
| 15 |
+
{"current_steps": 150, "total_steps": 799, "loss": 1.8314, "learning_rate": 4.883972810706548e-07, "epoch": 0.18761726078799248, "percentage": 18.77, "elapsed_time": "0:01:42", "remaining_time": "0:07:25"}
|
| 16 |
+
{"current_steps": 160, "total_steps": 799, "loss": 1.8075, "learning_rate": 4.848816158463157e-07, "epoch": 0.20012507817385866, "percentage": 20.03, "elapsed_time": "0:01:49", "remaining_time": "0:07:15"}
|
| 17 |
+
{"current_steps": 170, "total_steps": 799, "loss": 1.772, "learning_rate": 4.809175957485557e-07, "epoch": 0.21263289555972484, "percentage": 21.28, "elapsed_time": "0:01:56", "remaining_time": "0:07:10"}
|
| 18 |
+
{"current_steps": 180, "total_steps": 799, "loss": 1.7533, "learning_rate": 4.765127875157198e-07, "epoch": 0.225140712945591, "percentage": 22.53, "elapsed_time": "0:02:02", "remaining_time": "0:07:02"}
|
| 19 |
+
{"current_steps": 190, "total_steps": 799, "loss": 1.7891, "learning_rate": 4.7167559928665085e-07, "epoch": 0.23764853033145716, "percentage": 23.78, "elapsed_time": "0:02:08", "remaining_time": "0:06:51"}
|
| 20 |
+
{"current_steps": 200, "total_steps": 799, "loss": 1.8084, "learning_rate": 4.664152645507743e-07, "epoch": 0.2501563477173233, "percentage": 25.03, "elapsed_time": "0:02:15", "remaining_time": "0:06:44"}
|
| 21 |
+
{"current_steps": 210, "total_steps": 799, "loss": 1.7419, "learning_rate": 4.6074182452270883e-07, "epoch": 0.2626641651031895, "percentage": 26.28, "elapsed_time": "0:02:22", "remaining_time": "0:06:39"}
|
| 22 |
+
{"current_steps": 220, "total_steps": 799, "loss": 1.7693, "learning_rate": 4.546661089750462e-07, "epoch": 0.27517198248905567, "percentage": 27.53, "elapsed_time": "0:02:28", "remaining_time": "0:06:30"}
|
| 23 |
+
{"current_steps": 230, "total_steps": 799, "loss": 1.7335, "learning_rate": 4.4819971556588977e-07, "epoch": 0.2876797998749218, "percentage": 28.79, "elapsed_time": "0:02:33", "remaining_time": "0:06:20"}
|
| 24 |
+
{"current_steps": 240, "total_steps": 799, "loss": 1.8318, "learning_rate": 4.4135498770060986e-07, "epoch": 0.300187617260788, "percentage": 30.04, "elapsed_time": "0:02:39", "remaining_time": "0:06:12"}
|
| 25 |
+
{"current_steps": 250, "total_steps": 799, "loss": 1.7344, "learning_rate": 4.3414499097007697e-07, "epoch": 0.31269543464665417, "percentage": 31.29, "elapsed_time": "0:02:45", "remaining_time": "0:06:02"}
|
| 26 |
+
{"current_steps": 260, "total_steps": 799, "loss": 1.7508, "learning_rate": 4.265834882103472e-07, "epoch": 0.3252032520325203, "percentage": 32.54, "elapsed_time": "0:02:51", "remaining_time": "0:05:56"}
|
| 27 |
+
{"current_steps": 270, "total_steps": 799, "loss": 1.7215, "learning_rate": 4.186849132314081e-07, "epoch": 0.33771106941838647, "percentage": 33.79, "elapsed_time": "0:02:57", "remaining_time": "0:05:47"}
|
| 28 |
+
{"current_steps": 280, "total_steps": 799, "loss": 1.752, "learning_rate": 4.1046434326513303e-07, "epoch": 0.35021888680425267, "percentage": 35.04, "elapsed_time": "0:03:03", "remaining_time": "0:05:40"}
|
| 29 |
+
{"current_steps": 290, "total_steps": 799, "loss": 1.7647, "learning_rate": 4.0193747018503614e-07, "epoch": 0.3627267041901188, "percentage": 36.3, "elapsed_time": "0:03:09", "remaining_time": "0:05:32"}
|
| 30 |
+
{"current_steps": 300, "total_steps": 799, "loss": 1.7102, "learning_rate": 3.9312057055276636e-07, "epoch": 0.37523452157598497, "percentage": 37.55, "elapsed_time": "0:03:15", "remaining_time": "0:05:25"}
|
| 31 |
+
{"current_steps": 310, "total_steps": 799, "loss": 1.7913, "learning_rate": 3.8403047454851615e-07, "epoch": 0.3877423389618512, "percentage": 38.8, "elapsed_time": "0:03:21", "remaining_time": "0:05:18"}
|
| 32 |
+
{"current_steps": 320, "total_steps": 799, "loss": 1.757, "learning_rate": 3.7468453384465326e-07, "epoch": 0.4002501563477173, "percentage": 40.05, "elapsed_time": "0:03:27", "remaining_time": "0:05:10"}
|
| 33 |
+
{"current_steps": 330, "total_steps": 799, "loss": 1.7136, "learning_rate": 3.6510058848389946e-07, "epoch": 0.41275797373358347, "percentage": 41.3, "elapsed_time": "0:03:33", "remaining_time": "0:05:03"}
|
| 34 |
+
{"current_steps": 340, "total_steps": 799, "loss": 1.7576, "learning_rate": 3.5529693282528167e-07, "epoch": 0.4252657911194497, "percentage": 42.55, "elapsed_time": "0:03:40", "remaining_time": "0:04:57"}
|
| 35 |
+
{"current_steps": 350, "total_steps": 799, "loss": 1.7853, "learning_rate": 3.4529228062285864e-07, "epoch": 0.4377736085053158, "percentage": 43.8, "elapsed_time": "0:03:46", "remaining_time": "0:04:51"}
|
| 36 |
+
{"current_steps": 360, "total_steps": 799, "loss": 1.7072, "learning_rate": 3.351057293038841e-07, "epoch": 0.450281425891182, "percentage": 45.06, "elapsed_time": "0:03:53", "remaining_time": "0:04:44"}
|
| 37 |
+
{"current_steps": 370, "total_steps": 799, "loss": 1.6807, "learning_rate": 3.247567235145928e-07, "epoch": 0.4627892432770482, "percentage": 46.31, "elapsed_time": "0:03:59", "remaining_time": "0:04:38"}
|
| 38 |
+
{"current_steps": 380, "total_steps": 799, "loss": 1.6653, "learning_rate": 3.1426501800319697e-07, "epoch": 0.47529706066291433, "percentage": 47.56, "elapsed_time": "0:04:06", "remaining_time": "0:04:31"}
|
| 39 |
+
{"current_steps": 390, "total_steps": 799, "loss": 1.669, "learning_rate": 3.03650639910942e-07, "epoch": 0.4878048780487805, "percentage": 48.81, "elapsed_time": "0:04:12", "remaining_time": "0:04:24"}
|
| 40 |
+
{"current_steps": 400, "total_steps": 799, "loss": 1.6675, "learning_rate": 2.9293385054320446e-07, "epoch": 0.5003126954346466, "percentage": 50.06, "elapsed_time": "0:04:20", "remaining_time": "0:04:19"}
|
| 41 |
+
{"current_steps": 410, "total_steps": 799, "loss": 1.6581, "learning_rate": 2.8213510669360425e-07, "epoch": 0.5128205128205128, "percentage": 51.31, "elapsed_time": "0:04:25", "remaining_time": "0:04:12"}
|
| 42 |
+
{"current_steps": 420, "total_steps": 799, "loss": 1.6945, "learning_rate": 2.712750215949578e-07, "epoch": 0.525328330206379, "percentage": 52.57, "elapsed_time": "0:04:32", "remaining_time": "0:04:05"}
|
| 43 |
+
{"current_steps": 430, "total_steps": 799, "loss": 1.7825, "learning_rate": 2.6037432557161284e-07, "epoch": 0.5378361475922452, "percentage": 53.82, "elapsed_time": "0:04:37", "remaining_time": "0:03:58"}
|
| 44 |
+
{"current_steps": 440, "total_steps": 799, "loss": 1.7377, "learning_rate": 2.494538264682706e-07, "epoch": 0.5503439649781113, "percentage": 55.07, "elapsed_time": "0:04:44", "remaining_time": "0:03:51"}
|
| 45 |
+
{"current_steps": 450, "total_steps": 799, "loss": 1.7268, "learning_rate": 2.385343699308345e-07, "epoch": 0.5628517823639775, "percentage": 56.32, "elapsed_time": "0:04:51", "remaining_time": "0:03:45"}
|
| 46 |
+
{"current_steps": 460, "total_steps": 799, "loss": 1.7919, "learning_rate": 2.2763679961510135e-07, "epoch": 0.5753595997498436, "percentage": 57.57, "elapsed_time": "0:04:57", "remaining_time": "0:03:39"}
|
| 47 |
+
{"current_steps": 470, "total_steps": 799, "loss": 1.7826, "learning_rate": 2.1678191739925102e-07, "epoch": 0.5878674171357098, "percentage": 58.82, "elapsed_time": "0:05:03", "remaining_time": "0:03:32"}
|
| 48 |
+
{"current_steps": 480, "total_steps": 799, "loss": 1.6794, "learning_rate": 2.0599044367608349e-07, "epoch": 0.600375234521576, "percentage": 60.08, "elapsed_time": "0:05:09", "remaining_time": "0:03:25"}
|
| 49 |
+
{"current_steps": 490, "total_steps": 799, "loss": 1.7104, "learning_rate": 1.9528297780080066e-07, "epoch": 0.6128830519074422, "percentage": 61.33, "elapsed_time": "0:05:14", "remaining_time": "0:03:18"}
|
| 50 |
+
{"current_steps": 500, "total_steps": 799, "loss": 1.6521, "learning_rate": 1.8467995876983001e-07, "epoch": 0.6253908692933083, "percentage": 62.58, "elapsed_time": "0:05:20", "remaining_time": "0:03:11"}
|
| 51 |
+
{"current_steps": 500, "total_steps": 799, "eval_loss": 1.9212596416473389, "epoch": 0.6253908692933083, "percentage": 62.58, "elapsed_time": "0:06:13", "remaining_time": "0:03:43"}
|
| 52 |
+
{"current_steps": 510, "total_steps": 799, "loss": 1.748, "learning_rate": 1.742016262057507e-07, "epoch": 0.6378986866791745, "percentage": 63.83, "elapsed_time": "0:06:18", "remaining_time": "0:03:34"}
|
| 53 |
+
{"current_steps": 520, "total_steps": 799, "loss": 1.702, "learning_rate": 1.6386798172279508e-07, "epoch": 0.6504065040650406, "percentage": 65.08, "elapsed_time": "0:06:24", "remaining_time": "0:03:26"}
|
| 54 |
+
{"current_steps": 530, "total_steps": 799, "loss": 1.6578, "learning_rate": 1.5369875074667365e-07, "epoch": 0.6629143214509068, "percentage": 66.33, "elapsed_time": "0:06:31", "remaining_time": "0:03:18"}
|
| 55 |
+
{"current_steps": 540, "total_steps": 799, "loss": 1.6457, "learning_rate": 1.4371334486160352e-07, "epoch": 0.6754221388367729, "percentage": 67.58, "elapsed_time": "0:06:37", "remaining_time": "0:03:10"}
|
| 56 |
+
{"current_steps": 550, "total_steps": 799, "loss": 1.6502, "learning_rate": 1.3393082475641544e-07, "epoch": 0.6879299562226392, "percentage": 68.84, "elapsed_time": "0:06:44", "remaining_time": "0:03:02"}
|
| 57 |
+
{"current_steps": 560, "total_steps": 799, "loss": 1.8363, "learning_rate": 1.2436986384046786e-07, "epoch": 0.7004377736085053, "percentage": 70.09, "elapsed_time": "0:06:50", "remaining_time": "0:02:55"}
|
| 58 |
+
{"current_steps": 570, "total_steps": 799, "loss": 1.646, "learning_rate": 1.150487125988224e-07, "epoch": 0.7129455909943715, "percentage": 71.34, "elapsed_time": "0:06:57", "remaining_time": "0:02:47"}
|
| 59 |
+
{"current_steps": 580, "total_steps": 799, "loss": 1.7004, "learning_rate": 1.0598516375471978e-07, "epoch": 0.7254534083802376, "percentage": 72.59, "elapsed_time": "0:07:03", "remaining_time": "0:02:39"}
|
| 60 |
+
{"current_steps": 590, "total_steps": 799, "loss": 1.6428, "learning_rate": 9.719651830585668e-08, "epoch": 0.7379612257661038, "percentage": 73.84, "elapsed_time": "0:07:09", "remaining_time": "0:02:31"}
|
| 61 |
+
{"current_steps": 600, "total_steps": 799, "loss": 1.6991, "learning_rate": 8.86995524992951e-08, "epoch": 0.7504690431519699, "percentage": 75.09, "elapsed_time": "0:07:15", "remaining_time": "0:02:24"}
|
| 62 |
+
{"current_steps": 610, "total_steps": 799, "loss": 1.7138, "learning_rate": 8.051048580804456e-08, "epoch": 0.7629768605378362, "percentage": 76.35, "elapsed_time": "0:07:21", "remaining_time": "0:02:16"}
|
| 63 |
+
{"current_steps": 620, "total_steps": 799, "loss": 1.6684, "learning_rate": 7.264494997044521e-08, "epoch": 0.7754846779237023, "percentage": 77.6, "elapsed_time": "0:07:29", "remaining_time": "0:02:09"}
|
| 64 |
+
{"current_steps": 630, "total_steps": 799, "loss": 1.6755, "learning_rate": 6.51179591514505e-08, "epoch": 0.7879924953095685, "percentage": 78.85, "elapsed_time": "0:07:34", "remaining_time": "0:02:02"}
|
| 65 |
+
{"current_steps": 640, "total_steps": 799, "loss": 1.6433, "learning_rate": 5.794388128276798e-08, "epoch": 0.8005003126954346, "percentage": 80.1, "elapsed_time": "0:07:41", "remaining_time": "0:01:54"}
|
| 66 |
+
{"current_steps": 650, "total_steps": 799, "loss": 1.7482, "learning_rate": 5.113641063656618e-08, "epoch": 0.8130081300813008, "percentage": 81.35, "elapsed_time": "0:07:49", "remaining_time": "0:01:47"}
|
| 67 |
+
{"current_steps": 660, "total_steps": 799, "loss": 1.7041, "learning_rate": 4.470854168509819e-08, "epoch": 0.8255159474671669, "percentage": 82.6, "elapsed_time": "0:07:54", "remaining_time": "0:01:40"}
|
| 68 |
+
{"current_steps": 670, "total_steps": 799, "loss": 1.7051, "learning_rate": 3.867254429614281e-08, "epoch": 0.8380237648530331, "percentage": 83.85, "elapsed_time": "0:08:00", "remaining_time": "0:01:32"}
|
| 69 |
+
{"current_steps": 680, "total_steps": 799, "loss": 1.6833, "learning_rate": 3.3039940311609866e-08, "epoch": 0.8505315822388994, "percentage": 85.11, "elapsed_time": "0:08:06", "remaining_time": "0:01:25"}
|
| 70 |
+
{"current_steps": 690, "total_steps": 799, "loss": 1.7021, "learning_rate": 2.782148155401806e-08, "epoch": 0.8630393996247655, "percentage": 86.36, "elapsed_time": "0:08:12", "remaining_time": "0:01:17"}
|
| 71 |
+
{"current_steps": 700, "total_steps": 799, "loss": 1.724, "learning_rate": 2.302712930282849e-08, "epoch": 0.8755472170106317, "percentage": 87.61, "elapsed_time": "0:08:19", "remaining_time": "0:01:10"}
|
| 72 |
+
{"current_steps": 710, "total_steps": 799, "loss": 1.6296, "learning_rate": 1.8666035279809533e-08, "epoch": 0.8880550343964978, "percentage": 88.86, "elapsed_time": "0:08:25", "remaining_time": "0:01:03"}
|
| 73 |
+
{"current_steps": 720, "total_steps": 799, "loss": 1.7291, "learning_rate": 1.474652417972974e-08, "epoch": 0.900562851782364, "percentage": 90.11, "elapsed_time": "0:08:31", "remaining_time": "0:00:56"}
|
| 74 |
+
{"current_steps": 730, "total_steps": 799, "loss": 1.7443, "learning_rate": 1.1276077779725335e-08, "epoch": 0.9130706691682301, "percentage": 91.36, "elapsed_time": "0:08:36", "remaining_time": "0:00:48"}
|
| 75 |
+
{"current_steps": 740, "total_steps": 799, "loss": 1.6627, "learning_rate": 8.261320657674626e-09, "epoch": 0.9255784865540964, "percentage": 92.62, "elapsed_time": "0:08:42", "remaining_time": "0:00:41"}
|
| 76 |
+
{"current_steps": 750, "total_steps": 799, "loss": 1.631, "learning_rate": 5.708007546841892e-09, "epoch": 0.9380863039399625, "percentage": 93.87, "elapsed_time": "0:08:49", "remaining_time": "0:00:34"}
|
| 77 |
+
{"current_steps": 760, "total_steps": 799, "loss": 1.7077, "learning_rate": 3.621012350927377e-09, "epoch": 0.9505941213258287, "percentage": 95.12, "elapsed_time": "0:08:55", "remaining_time": "0:00:27"}
|
| 78 |
+
{"current_steps": 770, "total_steps": 799, "loss": 1.7075, "learning_rate": 2.004318840493707e-09, "epoch": 0.9631019387116948, "percentage": 96.37, "elapsed_time": "0:09:02", "remaining_time": "0:00:20"}
|
| 79 |
+
{"current_steps": 780, "total_steps": 799, "loss": 1.6703, "learning_rate": 8.610130485265655e-10, "epoch": 0.975609756097561, "percentage": 97.62, "elapsed_time": "0:09:08", "remaining_time": "0:00:13"}
|
| 80 |
+
{"current_steps": 790, "total_steps": 799, "loss": 1.6873, "learning_rate": 1.932773796461673e-10, "epoch": 0.9881175734834271, "percentage": 98.87, "elapsed_time": "0:09:15", "remaining_time": "0:00:06"}
|
| 81 |
+
{"current_steps": 799, "total_steps": 799, "epoch": 0.9993746091307066, "percentage": 100.0, "elapsed_time": "0:09:28", "remaining_time": "0:00:00"}
|