Training in progress, epoch 1

model-00001-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:65e52041bd73f87ccd2279ceb343fa1178be07265fc0f9c155849efc50a35f74
+oid sha256:ae1d920d0fd202f3eea3cf1c44f97ff83505b194637a31e09d905c0888ab8225
 size 4877660776

model-00002-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:953f71529ef4d489035b0616426038e83b67f30490c9ed2a944e374d181cf8bc
+oid sha256:df567f3b98e2b209b28e0f9472a0eb723def38607f16f9ac91d55b02c3f096a7
 size 4932751008

model-00003-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:13ae8d7fd39273049ed6ddde3a437687351ad1814947b6f041b875dbd89aa0ae
+oid sha256:72c8e48efd48354b6296873a60e503bbc4b404a2af79757f8a688c58d7e930e6
 size 4330865200

model-00004-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0581dc9af5f2690c1d95c3393481797eba7bf0bbcd7258ba14b81bae30935d20
+oid sha256:f56c17b9c0f95de767836a79c198dc5d270d7e4892b3f574fa65c75b8879ac8f
 size 1089994880

trainer_log.jsonl

@@ -199,3 +199,194 @@
 {"current_steps": 199, "total_steps": 573, "loss": 0.5943, "lr": 8.261712271830564e-06, "epoch": 1.0400696864111498, "percentage": 34.73, "elapsed_time": "0:51:19", "remaining_time": "1:36:27"}
 {"current_steps": 200, "total_steps": 573, "loss": 0.5491, "lr": 8.23853435315295e-06, "epoch": 1.0452961672473868, "percentage": 34.9, "elapsed_time": "0:51:32", "remaining_time": "1:36:07"}
 {"current_steps": 201, "total_steps": 573, "loss": 0.5604, "lr": 8.215235921877403e-06, "epoch": 1.0505226480836236, "percentage": 35.08, "elapsed_time": "0:51:49", "remaining_time": "1:35:54"}
+{"current_steps": 202, "total_steps": 573, "loss": 0.6434, "lr": 8.191817844986998e-06, "epoch": 1.0557491289198606, "percentage": 35.25, "elapsed_time": "0:52:01", "remaining_time": "1:35:33"}
+{"current_steps": 203, "total_steps": 573, "loss": 0.5594, "lr": 8.168280993917078e-06, "epoch": 1.0609756097560976, "percentage": 35.43, "elapsed_time": "0:52:17", "remaining_time": "1:35:18"}
+{"current_steps": 204, "total_steps": 573, "loss": 0.5686, "lr": 8.144626244522812e-06, "epoch": 1.0662020905923344, "percentage": 35.6, "elapsed_time": "0:52:31", "remaining_time": "1:34:59"}
+{"current_steps": 205, "total_steps": 573, "loss": 0.6803, "lr": 8.120854477046621e-06, "epoch": 1.0714285714285714, "percentage": 35.78, "elapsed_time": "0:52:46", "remaining_time": "1:34:43"}
+{"current_steps": 206, "total_steps": 573, "loss": 0.5824, "lr": 8.096966576085406e-06, "epoch": 1.0766550522648084, "percentage": 35.95, "elapsed_time": "0:52:59", "remaining_time": "1:34:24"}
+{"current_steps": 207, "total_steps": 573, "loss": 0.6028, "lr": 8.072963430557636e-06, "epoch": 1.0818815331010452, "percentage": 36.13, "elapsed_time": "0:53:14", "remaining_time": "1:34:07"}
+{"current_steps": 208, "total_steps": 573, "loss": 0.5777, "lr": 8.048845933670274e-06, "epoch": 1.0871080139372822, "percentage": 36.3, "elapsed_time": "0:53:29", "remaining_time": "1:33:52"}
+{"current_steps": 209, "total_steps": 573, "loss": 0.5577, "lr": 8.024614982885527e-06, "epoch": 1.0923344947735192, "percentage": 36.47, "elapsed_time": "0:53:45", "remaining_time": "1:33:37"}
+{"current_steps": 210, "total_steps": 573, "loss": 0.6143, "lr": 8.00027147988747e-06, "epoch": 1.0975609756097562, "percentage": 36.65, "elapsed_time": "0:53:59", "remaining_time": "1:33:19"}
+{"current_steps": 211, "total_steps": 573, "loss": 0.596, "lr": 7.975816330548466e-06, "epoch": 1.102787456445993, "percentage": 36.82, "elapsed_time": "0:54:12", "remaining_time": "1:32:59"}
+{"current_steps": 212, "total_steps": 573, "loss": 0.6494, "lr": 7.951250444895485e-06, "epoch": 1.10801393728223, "percentage": 37.0, "elapsed_time": "0:54:27", "remaining_time": "1:32:43"}
+{"current_steps": 213, "total_steps": 573, "loss": 0.5217, "lr": 7.92657473707621e-06, "epoch": 1.113240418118467, "percentage": 37.17, "elapsed_time": "0:54:44", "remaining_time": "1:32:31"}
+{"current_steps": 214, "total_steps": 573, "loss": 0.6941, "lr": 7.901790125325049e-06, "epoch": 1.1184668989547037, "percentage": 37.35, "elapsed_time": "0:55:01", "remaining_time": "1:32:19"}
+{"current_steps": 215, "total_steps": 573, "loss": 0.6076, "lr": 7.876897531928943e-06, "epoch": 1.1236933797909407, "percentage": 37.52, "elapsed_time": "0:55:15", "remaining_time": "1:32:00"}
+{"current_steps": 216, "total_steps": 573, "loss": 0.5852, "lr": 7.851897883193057e-06, "epoch": 1.1289198606271778, "percentage": 37.7, "elapsed_time": "0:55:30", "remaining_time": "1:31:44"}
+{"current_steps": 217, "total_steps": 573, "loss": 0.6033, "lr": 7.82679210940631e-06, "epoch": 1.1341463414634148, "percentage": 37.87, "elapsed_time": "0:55:44", "remaining_time": "1:31:27"}
+{"current_steps": 218, "total_steps": 573, "loss": 0.5587, "lr": 7.801581144806752e-06, "epoch": 1.1393728222996515, "percentage": 38.05, "elapsed_time": "0:55:59", "remaining_time": "1:31:10"}
+{"current_steps": 219, "total_steps": 573, "loss": 0.618, "lr": 7.7762659275468e-06, "epoch": 1.1445993031358885, "percentage": 38.22, "elapsed_time": "0:56:17", "remaining_time": "1:30:58"}
+{"current_steps": 220, "total_steps": 573, "loss": 0.6282, "lr": 7.750847399658336e-06, "epoch": 1.1498257839721253, "percentage": 38.39, "elapsed_time": "0:56:31", "remaining_time": "1:30:41"}
+{"current_steps": 221, "total_steps": 573, "loss": 0.5451, "lr": 7.725326507017644e-06, "epoch": 1.1550522648083623, "percentage": 38.57, "elapsed_time": "0:56:44", "remaining_time": "1:30:22"}
+{"current_steps": 222, "total_steps": 573, "loss": 0.5972, "lr": 7.699704199310204e-06, "epoch": 1.1602787456445993, "percentage": 38.74, "elapsed_time": "0:57:00", "remaining_time": "1:30:07"}
+{"current_steps": 223, "total_steps": 573, "loss": 0.6114, "lr": 7.673981429995372e-06, "epoch": 1.1655052264808363, "percentage": 38.92, "elapsed_time": "0:57:11", "remaining_time": "1:29:46"}
+{"current_steps": 224, "total_steps": 573, "loss": 0.6268, "lr": 7.648159156270884e-06, "epoch": 1.170731707317073, "percentage": 39.09, "elapsed_time": "0:57:28", "remaining_time": "1:29:33"}
+{"current_steps": 225, "total_steps": 573, "loss": 0.6141, "lr": 7.622238339037248e-06, "epoch": 1.17595818815331, "percentage": 39.27, "elapsed_time": "0:57:43", "remaining_time": "1:29:17"}
+{"current_steps": 226, "total_steps": 573, "loss": 0.5655, "lr": 7.596219942861971e-06, "epoch": 1.181184668989547, "percentage": 39.44, "elapsed_time": "0:57:55", "remaining_time": "1:28:55"}
+{"current_steps": 227, "total_steps": 573, "loss": 0.5983, "lr": 7.570104935943685e-06, "epoch": 1.1864111498257839, "percentage": 39.62, "elapsed_time": "0:58:07", "remaining_time": "1:28:35"}
+{"current_steps": 228, "total_steps": 573, "loss": 0.6072, "lr": 7.5438942900761035e-06, "epoch": 1.1916376306620209, "percentage": 39.79, "elapsed_time": "0:58:22", "remaining_time": "1:28:20"}
+{"current_steps": 229, "total_steps": 573, "loss": 0.6306, "lr": 7.517588980611864e-06, "epoch": 1.1968641114982579, "percentage": 39.97, "elapsed_time": "0:58:36", "remaining_time": "1:28:02"}
+{"current_steps": 230, "total_steps": 573, "loss": 0.523, "lr": 7.491189986426236e-06, "epoch": 1.202090592334495, "percentage": 40.14, "elapsed_time": "0:58:51", "remaining_time": "1:27:47"}
+{"current_steps": 231, "total_steps": 573, "loss": 0.6248, "lr": 7.464698289880689e-06, "epoch": 1.2073170731707317, "percentage": 40.31, "elapsed_time": "0:59:04", "remaining_time": "1:27:28"}
+{"current_steps": 232, "total_steps": 573, "loss": 0.5892, "lr": 7.438114876786344e-06, "epoch": 1.2125435540069687, "percentage": 40.49, "elapsed_time": "0:59:20", "remaining_time": "1:27:12"}
+{"current_steps": 233, "total_steps": 573, "loss": 0.604, "lr": 7.411440736367281e-06, "epoch": 1.2177700348432055, "percentage": 40.66, "elapsed_time": "0:59:37", "remaining_time": "1:27:00"}
+{"current_steps": 234, "total_steps": 573, "loss": 0.5938, "lr": 7.384676861223738e-06, "epoch": 1.2229965156794425, "percentage": 40.84, "elapsed_time": "0:59:50", "remaining_time": "1:26:41"}
+{"current_steps": 235, "total_steps": 573, "loss": 0.5606, "lr": 7.3578242472951635e-06, "epoch": 1.2282229965156795, "percentage": 41.01, "elapsed_time": "1:00:06", "remaining_time": "1:26:26"}
+{"current_steps": 236, "total_steps": 573, "loss": 0.6321, "lr": 7.330883893823164e-06, "epoch": 1.2334494773519165, "percentage": 41.19, "elapsed_time": "1:00:21", "remaining_time": "1:26:11"}
+{"current_steps": 237, "total_steps": 573, "loss": 0.5772, "lr": 7.303856803314313e-06, "epoch": 1.2386759581881532, "percentage": 41.36, "elapsed_time": "1:00:36", "remaining_time": "1:25:55"}
+{"current_steps": 238, "total_steps": 573, "loss": 0.6249, "lr": 7.276743981502856e-06, "epoch": 1.2439024390243902, "percentage": 41.54, "elapsed_time": "1:00:50", "remaining_time": "1:25:38"}
+{"current_steps": 239, "total_steps": 573, "loss": 0.5742, "lr": 7.249546437313273e-06, "epoch": 1.2491289198606272, "percentage": 41.71, "elapsed_time": "1:01:04", "remaining_time": "1:25:21"}
+{"current_steps": 240, "total_steps": 573, "loss": 0.5931, "lr": 7.22226518282274e-06, "epoch": 1.254355400696864, "percentage": 41.88, "elapsed_time": "1:01:21", "remaining_time": "1:25:07"}
+{"current_steps": 241, "total_steps": 573, "loss": 0.6026, "lr": 7.194901233223471e-06, "epoch": 1.259581881533101, "percentage": 42.06, "elapsed_time": "1:01:37", "remaining_time": "1:24:52"}
+{"current_steps": 242, "total_steps": 573, "loss": 0.5973, "lr": 7.167455606784936e-06, "epoch": 1.264808362369338, "percentage": 42.23, "elapsed_time": "1:01:47", "remaining_time": "1:24:31"}
+{"current_steps": 243, "total_steps": 573, "loss": 0.639, "lr": 7.139929324815965e-06, "epoch": 1.270034843205575, "percentage": 42.41, "elapsed_time": "1:02:04", "remaining_time": "1:24:18"}
+{"current_steps": 244, "total_steps": 573, "loss": 0.5975, "lr": 7.112323411626756e-06, "epoch": 1.2752613240418118, "percentage": 42.58, "elapsed_time": "1:02:19", "remaining_time": "1:24:02"}
+{"current_steps": 245, "total_steps": 573, "loss": 0.5932, "lr": 7.084638894490744e-06, "epoch": 1.2804878048780488, "percentage": 42.76, "elapsed_time": "1:02:34", "remaining_time": "1:23:46"}
+{"current_steps": 246, "total_steps": 573, "loss": 0.6211, "lr": 7.056876803606384e-06, "epoch": 1.2857142857142856, "percentage": 42.93, "elapsed_time": "1:02:47", "remaining_time": "1:23:28"}
+{"current_steps": 247, "total_steps": 573, "loss": 0.5816, "lr": 7.029038172058809e-06, "epoch": 1.2909407665505226, "percentage": 43.11, "elapsed_time": "1:03:05", "remaining_time": "1:23:16"}
+{"current_steps": 248, "total_steps": 573, "loss": 0.5793, "lr": 7.00112403578139e-06, "epoch": 1.2961672473867596, "percentage": 43.28, "elapsed_time": "1:03:24", "remaining_time": "1:23:05"}
+{"current_steps": 249, "total_steps": 573, "loss": 0.5969, "lr": 6.9731354335171885e-06, "epoch": 1.3013937282229966, "percentage": 43.46, "elapsed_time": "1:03:38", "remaining_time": "1:22:48"}
+{"current_steps": 250, "total_steps": 573, "loss": 0.6263, "lr": 6.945073406780296e-06, "epoch": 1.3066202090592334, "percentage": 43.63, "elapsed_time": "1:03:50", "remaining_time": "1:22:28"}
+{"current_steps": 251, "total_steps": 573, "loss": 0.6258, "lr": 6.916938999817085e-06, "epoch": 1.3118466898954704, "percentage": 43.8, "elapsed_time": "1:04:05", "remaining_time": "1:22:13"}
+{"current_steps": 252, "total_steps": 573, "loss": 0.5697, "lr": 6.888733259567343e-06, "epoch": 1.3170731707317074, "percentage": 43.98, "elapsed_time": "1:04:19", "remaining_time": "1:21:56"}
+{"current_steps": 253, "total_steps": 573, "loss": 0.6192, "lr": 6.860457235625322e-06, "epoch": 1.3222996515679442, "percentage": 44.15, "elapsed_time": "1:04:35", "remaining_time": "1:21:41"}
+{"current_steps": 254, "total_steps": 573, "loss": 0.5595, "lr": 6.832111980200672e-06, "epoch": 1.3275261324041812, "percentage": 44.33, "elapsed_time": "1:04:53", "remaining_time": "1:21:29"}
+{"current_steps": 255, "total_steps": 573, "loss": 0.6001, "lr": 6.803698548079294e-06, "epoch": 1.3327526132404182, "percentage": 44.5, "elapsed_time": "1:05:08", "remaining_time": "1:21:14"}
+{"current_steps": 256, "total_steps": 573, "loss": 0.5915, "lr": 6.775217996584082e-06, "epoch": 1.3379790940766552, "percentage": 44.68, "elapsed_time": "1:05:23", "remaining_time": "1:20:58"}
+{"current_steps": 257, "total_steps": 573, "loss": 0.5953, "lr": 6.746671385535586e-06, "epoch": 1.343205574912892, "percentage": 44.85, "elapsed_time": "1:05:41", "remaining_time": "1:20:46"}
+{"current_steps": 258, "total_steps": 573, "loss": 0.6388, "lr": 6.7180597772125665e-06, "epoch": 1.348432055749129, "percentage": 45.03, "elapsed_time": "1:05:57", "remaining_time": "1:20:31"}
+{"current_steps": 259, "total_steps": 573, "loss": 0.5838, "lr": 6.689384236312465e-06, "epoch": 1.3536585365853657, "percentage": 45.2, "elapsed_time": "1:06:12", "remaining_time": "1:20:16"}
+{"current_steps": 260, "total_steps": 573, "loss": 0.6318, "lr": 6.660645829911794e-06, "epoch": 1.3588850174216027, "percentage": 45.38, "elapsed_time": "1:06:30", "remaining_time": "1:20:03"}
+{"current_steps": 261, "total_steps": 573, "loss": 0.5425, "lr": 6.631845627426418e-06, "epoch": 1.3641114982578397, "percentage": 45.55, "elapsed_time": "1:06:46", "remaining_time": "1:19:49"}
+{"current_steps": 262, "total_steps": 573, "loss": 0.5305, "lr": 6.602984700571758e-06, "epoch": 1.3693379790940767, "percentage": 45.72, "elapsed_time": "1:07:01", "remaining_time": "1:19:33"}
+{"current_steps": 263, "total_steps": 573, "loss": 0.6405, "lr": 6.574064123322925e-06, "epoch": 1.3745644599303135, "percentage": 45.9, "elapsed_time": "1:07:18", "remaining_time": "1:19:20"}
+{"current_steps": 264, "total_steps": 573, "loss": 0.5802, "lr": 6.545084971874738e-06, "epoch": 1.3797909407665505, "percentage": 46.07, "elapsed_time": "1:07:36", "remaining_time": "1:19:07"}
+{"current_steps": 265, "total_steps": 573, "loss": 0.6166, "lr": 6.516048324601685e-06, "epoch": 1.3850174216027875, "percentage": 46.25, "elapsed_time": "1:07:52", "remaining_time": "1:18:53"}
+{"current_steps": 266, "total_steps": 573, "loss": 0.6049, "lr": 6.486955262017794e-06, "epoch": 1.3902439024390243, "percentage": 46.42, "elapsed_time": "1:08:07", "remaining_time": "1:18:37"}
+{"current_steps": 267, "total_steps": 573, "loss": 0.615, "lr": 6.457806866736424e-06, "epoch": 1.3954703832752613, "percentage": 46.6, "elapsed_time": "1:08:21", "remaining_time": "1:18:20"}
+{"current_steps": 268, "total_steps": 573, "loss": 0.522, "lr": 6.42860422342998e-06, "epoch": 1.4006968641114983, "percentage": 46.77, "elapsed_time": "1:08:30", "remaining_time": "1:17:58"}
+{"current_steps": 269, "total_steps": 573, "loss": 0.6314, "lr": 6.399348418789545e-06, "epoch": 1.4059233449477353, "percentage": 46.95, "elapsed_time": "1:08:47", "remaining_time": "1:17:44"}
+{"current_steps": 270, "total_steps": 573, "loss": 0.5822, "lr": 6.37004054148445e-06, "epoch": 1.411149825783972, "percentage": 47.12, "elapsed_time": "1:09:00", "remaining_time": "1:17:26"}
+{"current_steps": 271, "total_steps": 573, "loss": 0.6093, "lr": 6.3406816821217554e-06, "epoch": 1.416376306620209, "percentage": 47.29, "elapsed_time": "1:09:16", "remaining_time": "1:17:12"}
+{"current_steps": 272, "total_steps": 573, "loss": 0.5832, "lr": 6.311272933205672e-06, "epoch": 1.4216027874564459, "percentage": 47.47, "elapsed_time": "1:09:29", "remaining_time": "1:16:54"}
+{"current_steps": 273, "total_steps": 573, "loss": 0.529, "lr": 6.281815389096903e-06, "epoch": 1.4268292682926829, "percentage": 47.64, "elapsed_time": "1:09:45", "remaining_time": "1:16:39"}
+{"current_steps": 274, "total_steps": 573, "loss": 0.6526, "lr": 6.2523101459719204e-06, "epoch": 1.4320557491289199, "percentage": 47.82, "elapsed_time": "1:10:01", "remaining_time": "1:16:25"}
+{"current_steps": 275, "total_steps": 573, "loss": 0.6282, "lr": 6.222758301782183e-06, "epoch": 1.4372822299651569, "percentage": 47.99, "elapsed_time": "1:10:17", "remaining_time": "1:16:09"}
+{"current_steps": 276, "total_steps": 573, "loss": 0.5903, "lr": 6.193160956213262e-06, "epoch": 1.4425087108013936, "percentage": 48.17, "elapsed_time": "1:10:29", "remaining_time": "1:15:51"}
+{"current_steps": 277, "total_steps": 573, "loss": 0.6083, "lr": 6.163519210643939e-06, "epoch": 1.4477351916376306, "percentage": 48.34, "elapsed_time": "1:10:42", "remaining_time": "1:15:33"}
+{"current_steps": 278, "total_steps": 573, "loss": 0.5271, "lr": 6.133834168105206e-06, "epoch": 1.4529616724738676, "percentage": 48.52, "elapsed_time": "1:10:58", "remaining_time": "1:15:18"}
+{"current_steps": 279, "total_steps": 573, "loss": 0.5636, "lr": 6.104106933239227e-06, "epoch": 1.4581881533101044, "percentage": 48.69, "elapsed_time": "1:11:12", "remaining_time": "1:15:02"}
+{"current_steps": 280, "total_steps": 573, "loss": 0.6205, "lr": 6.07433861225823e-06, "epoch": 1.4634146341463414, "percentage": 48.87, "elapsed_time": "1:11:29", "remaining_time": "1:14:48"}
+{"current_steps": 281, "total_steps": 573, "loss": 0.5794, "lr": 6.044530312903343e-06, "epoch": 1.4686411149825784, "percentage": 49.04, "elapsed_time": "1:11:42", "remaining_time": "1:14:31"}
+{"current_steps": 282, "total_steps": 573, "loss": 0.6167, "lr": 6.014683144403375e-06, "epoch": 1.4738675958188154, "percentage": 49.21, "elapsed_time": "1:11:55", "remaining_time": "1:14:13"}
+{"current_steps": 283, "total_steps": 573, "loss": 0.6008, "lr": 5.9847982174335314e-06, "epoch": 1.4790940766550522, "percentage": 49.39, "elapsed_time": "1:12:09", "remaining_time": "1:13:56"}
+{"current_steps": 284, "total_steps": 573, "loss": 0.6057, "lr": 5.954876644074092e-06, "epoch": 1.4843205574912892, "percentage": 49.56, "elapsed_time": "1:12:26", "remaining_time": "1:13:42"}
+{"current_steps": 285, "total_steps": 573, "loss": 0.5872, "lr": 5.924919537769025e-06, "epoch": 1.489547038327526, "percentage": 49.74, "elapsed_time": "1:12:40", "remaining_time": "1:13:26"}
+{"current_steps": 286, "total_steps": 573, "loss": 0.549, "lr": 5.894928013284551e-06, "epoch": 1.494773519163763, "percentage": 49.91, "elapsed_time": "1:12:55", "remaining_time": "1:13:10"}
+{"current_steps": 287, "total_steps": 573, "loss": 0.6325, "lr": 5.8649031866676685e-06, "epoch": 1.5, "percentage": 50.09, "elapsed_time": "1:13:11", "remaining_time": "1:12:56"}
+{"current_steps": 288, "total_steps": 573, "loss": 0.5278, "lr": 5.834846175204612e-06, "epoch": 1.505226480836237, "percentage": 50.26, "elapsed_time": "1:13:26", "remaining_time": "1:12:40"}
+{"current_steps": 289, "total_steps": 573, "loss": 0.6634, "lr": 5.804758097379281e-06, "epoch": 1.510452961672474, "percentage": 50.44, "elapsed_time": "1:13:43", "remaining_time": "1:12:26"}
+{"current_steps": 290, "total_steps": 573, "loss": 0.5773, "lr": 5.774640072831622e-06, "epoch": 1.5156794425087108, "percentage": 50.61, "elapsed_time": "1:13:57", "remaining_time": "1:12:09"}
+{"current_steps": 291, "total_steps": 573, "loss": 0.6216, "lr": 5.74449322231596e-06, "epoch": 1.5209059233449478, "percentage": 50.79, "elapsed_time": "1:14:11", "remaining_time": "1:11:54"}
+{"current_steps": 292, "total_steps": 573, "loss": 0.5697, "lr": 5.714318667659295e-06, "epoch": 1.5261324041811846, "percentage": 50.96, "elapsed_time": "1:14:26", "remaining_time": "1:11:37"}
+{"current_steps": 293, "total_steps": 573, "loss": 0.5902, "lr": 5.684117531719552e-06, "epoch": 1.5313588850174216, "percentage": 51.13, "elapsed_time": "1:14:38", "remaining_time": "1:11:20"}
+{"current_steps": 294, "total_steps": 573, "loss": 0.6084, "lr": 5.653890938343806e-06, "epoch": 1.5365853658536586, "percentage": 51.31, "elapsed_time": "1:14:55", "remaining_time": "1:11:05"}
+{"current_steps": 295, "total_steps": 573, "loss": 0.6074, "lr": 5.623640012326455e-06, "epoch": 1.5418118466898956, "percentage": 51.48, "elapsed_time": "1:15:11", "remaining_time": "1:10:51"}
+{"current_steps": 296, "total_steps": 573, "loss": 0.559, "lr": 5.593365879367361e-06, "epoch": 1.5470383275261324, "percentage": 51.66, "elapsed_time": "1:15:25", "remaining_time": "1:10:34"}
+{"current_steps": 297, "total_steps": 573, "loss": 0.6259, "lr": 5.56306966602997e-06, "epoch": 1.5522648083623694, "percentage": 51.83, "elapsed_time": "1:15:42", "remaining_time": "1:10:21"}
+{"current_steps": 298, "total_steps": 573, "loss": 0.6274, "lr": 5.532752499699381e-06, "epoch": 1.5574912891986061, "percentage": 52.01, "elapsed_time": "1:15:57", "remaining_time": "1:10:05"}
+{"current_steps": 299, "total_steps": 573, "loss": 0.5689, "lr": 5.502415508540401e-06, "epoch": 1.5627177700348431, "percentage": 52.18, "elapsed_time": "1:16:09", "remaining_time": "1:09:47"}
+{"current_steps": 300, "total_steps": 573, "loss": 0.615, "lr": 5.472059821455554e-06, "epoch": 1.5679442508710801, "percentage": 52.36, "elapsed_time": "1:16:26", "remaining_time": "1:09:33"}
+{"current_steps": 301, "total_steps": 573, "loss": 0.6209, "lr": 5.441686568043086e-06, "epoch": 1.5731707317073171, "percentage": 52.53, "elapsed_time": "1:16:42", "remaining_time": "1:09:18"}
+{"current_steps": 302, "total_steps": 573, "loss": 0.5778, "lr": 5.411296878554918e-06, "epoch": 1.5783972125435541, "percentage": 52.71, "elapsed_time": "1:16:56", "remaining_time": "1:09:02"}
+{"current_steps": 303, "total_steps": 573, "loss": 0.5891, "lr": 5.380891883854591e-06, "epoch": 1.583623693379791, "percentage": 52.88, "elapsed_time": "1:17:12", "remaining_time": "1:08:47"}
+{"current_steps": 304, "total_steps": 573, "loss": 0.5781, "lr": 5.3504727153751865e-06, "epoch": 1.588850174216028, "percentage": 53.05, "elapsed_time": "1:17:24", "remaining_time": "1:08:30"}
+{"current_steps": 305, "total_steps": 573, "loss": 0.6108, "lr": 5.320040505077222e-06, "epoch": 1.5940766550522647, "percentage": 53.23, "elapsed_time": "1:17:40", "remaining_time": "1:08:15"}
+{"current_steps": 306, "total_steps": 573, "loss": 0.5925, "lr": 5.289596385406527e-06, "epoch": 1.5993031358885017, "percentage": 53.4, "elapsed_time": "1:17:57", "remaining_time": "1:08:01"}
+{"current_steps": 307, "total_steps": 573, "loss": 0.5211, "lr": 5.259141489252104e-06, "epoch": 1.6045296167247387, "percentage": 53.58, "elapsed_time": "1:18:10", "remaining_time": "1:07:44"}
+{"current_steps": 308, "total_steps": 573, "loss": 0.6835, "lr": 5.228676949903974e-06, "epoch": 1.6097560975609757, "percentage": 53.75, "elapsed_time": "1:18:27", "remaining_time": "1:07:30"}
+{"current_steps": 309, "total_steps": 573, "loss": 0.5614, "lr": 5.198203901010993e-06, "epoch": 1.6149825783972127, "percentage": 53.93, "elapsed_time": "1:18:42", "remaining_time": "1:07:14"}
+{"current_steps": 310, "total_steps": 573, "loss": 0.6063, "lr": 5.167723476538683e-06, "epoch": 1.6202090592334495, "percentage": 54.1, "elapsed_time": "1:18:56", "remaining_time": "1:06:58"}
+{"current_steps": 311, "total_steps": 573, "loss": 0.5358, "lr": 5.137236810727025e-06, "epoch": 1.6254355400696863, "percentage": 54.28, "elapsed_time": "1:19:10", "remaining_time": "1:06:42"}
+{"current_steps": 312, "total_steps": 573, "loss": 0.6031, "lr": 5.106745038048251e-06, "epoch": 1.6306620209059233, "percentage": 54.45, "elapsed_time": "1:19:19", "remaining_time": "1:06:21"}
+{"current_steps": 313, "total_steps": 573, "loss": 0.5926, "lr": 5.07624929316463e-06, "epoch": 1.6358885017421603, "percentage": 54.62, "elapsed_time": "1:19:33", "remaining_time": "1:06:04"}
+{"current_steps": 314, "total_steps": 573, "loss": 0.5909, "lr": 5.045750710886248e-06, "epoch": 1.6411149825783973, "percentage": 54.8, "elapsed_time": "1:19:50", "remaining_time": "1:05:51"}
+{"current_steps": 315, "total_steps": 573, "loss": 0.5394, "lr": 5.015250426128772e-06, "epoch": 1.6463414634146343, "percentage": 54.97, "elapsed_time": "1:20:06", "remaining_time": "1:05:36"}
+{"current_steps": 316, "total_steps": 573, "loss": 0.6349, "lr": 4.984749573871228e-06, "epoch": 1.651567944250871, "percentage": 55.15, "elapsed_time": "1:20:23", "remaining_time": "1:05:22"}
+{"current_steps": 317, "total_steps": 573, "loss": 0.6271, "lr": 4.954249289113753e-06, "epoch": 1.656794425087108, "percentage": 55.32, "elapsed_time": "1:20:37", "remaining_time": "1:05:06"}
+{"current_steps": 318, "total_steps": 573, "loss": 0.6067, "lr": 4.923750706835371e-06, "epoch": 1.6620209059233448, "percentage": 55.5, "elapsed_time": "1:20:53", "remaining_time": "1:04:52"}
+{"current_steps": 319, "total_steps": 573, "loss": 0.5688, "lr": 4.8932549619517514e-06, "epoch": 1.6672473867595818, "percentage": 55.67, "elapsed_time": "1:21:10", "remaining_time": "1:04:37"}
+{"current_steps": 320, "total_steps": 573, "loss": 0.5993, "lr": 4.862763189272976e-06, "epoch": 1.6724738675958188, "percentage": 55.85, "elapsed_time": "1:21:25", "remaining_time": "1:04:22"}
+{"current_steps": 321, "total_steps": 573, "loss": 0.5251, "lr": 4.832276523461317e-06, "epoch": 1.6777003484320558, "percentage": 56.02, "elapsed_time": "1:21:38", "remaining_time": "1:04:05"}
+{"current_steps": 322, "total_steps": 573, "loss": 0.6171, "lr": 4.801796098989009e-06, "epoch": 1.6829268292682928, "percentage": 56.2, "elapsed_time": "1:21:53", "remaining_time": "1:03:50"}
+{"current_steps": 323, "total_steps": 573, "loss": 0.5868, "lr": 4.771323050096028e-06, "epoch": 1.6881533101045296, "percentage": 56.37, "elapsed_time": "1:22:07", "remaining_time": "1:03:34"}
+{"current_steps": 324, "total_steps": 573, "loss": 0.6094, "lr": 4.740858510747897e-06, "epoch": 1.6933797909407664, "percentage": 56.54, "elapsed_time": "1:22:22", "remaining_time": "1:03:18"}
+{"current_steps": 325, "total_steps": 573, "loss": 0.551, "lr": 4.710403614593475e-06, "epoch": 1.6986062717770034, "percentage": 56.72, "elapsed_time": "1:22:33", "remaining_time": "1:02:59"}
+{"current_steps": 326, "total_steps": 573, "loss": 0.6207, "lr": 4.679959494922779e-06, "epoch": 1.7038327526132404, "percentage": 56.89, "elapsed_time": "1:22:48", "remaining_time": "1:02:44"}
+{"current_steps": 327, "total_steps": 573, "loss": 0.5813, "lr": 4.649527284624814e-06, "epoch": 1.7090592334494774, "percentage": 57.07, "elapsed_time": "1:23:05", "remaining_time": "1:02:30"}
+{"current_steps": 328, "total_steps": 573, "loss": 0.5997, "lr": 4.619108116145411e-06, "epoch": 1.7142857142857144, "percentage": 57.24, "elapsed_time": "1:23:21", "remaining_time": "1:02:16"}
+{"current_steps": 329, "total_steps": 573, "loss": 0.6162, "lr": 4.588703121445084e-06, "epoch": 1.7195121951219512, "percentage": 57.42, "elapsed_time": "1:23:37", "remaining_time": "1:02:01"}
+{"current_steps": 330, "total_steps": 573, "loss": 0.5616, "lr": 4.558313431956914e-06, "epoch": 1.7247386759581882, "percentage": 57.59, "elapsed_time": "1:23:53", "remaining_time": "1:01:46"}
+{"current_steps": 331, "total_steps": 573, "loss": 0.6088, "lr": 4.527940178544446e-06, "epoch": 1.729965156794425, "percentage": 57.77, "elapsed_time": "1:24:08", "remaining_time": "1:01:31"}
+{"current_steps": 332, "total_steps": 573, "loss": 0.5808, "lr": 4.4975844914596015e-06, "epoch": 1.735191637630662, "percentage": 57.94, "elapsed_time": "1:24:23", "remaining_time": "1:01:15"}
+{"current_steps": 333, "total_steps": 573, "loss": 0.6187, "lr": 4.467247500300621e-06, "epoch": 1.740418118466899, "percentage": 58.12, "elapsed_time": "1:24:39", "remaining_time": "1:01:01"}
+{"current_steps": 334, "total_steps": 573, "loss": 0.5532, "lr": 4.436930333970033e-06, "epoch": 1.745644599303136, "percentage": 58.29, "elapsed_time": "1:24:53", "remaining_time": "1:00:44"}
+{"current_steps": 335, "total_steps": 573, "loss": 0.6331, "lr": 4.40663412063264e-06, "epoch": 1.750871080139373, "percentage": 58.46, "elapsed_time": "1:25:09", "remaining_time": "1:00:29"}
+{"current_steps": 336, "total_steps": 573, "loss": 0.5759, "lr": 4.376359987673547e-06, "epoch": 1.7560975609756098, "percentage": 58.64, "elapsed_time": "1:25:22", "remaining_time": "1:00:13"}
+{"current_steps": 337, "total_steps": 573, "loss": 0.5859, "lr": 4.346109061656196e-06, "epoch": 1.7613240418118465, "percentage": 58.81, "elapsed_time": "1:25:32", "remaining_time": "0:59:54"}
+{"current_steps": 338, "total_steps": 573, "loss": 0.6084, "lr": 4.31588246828045e-06, "epoch": 1.7665505226480835, "percentage": 58.99, "elapsed_time": "1:25:47", "remaining_time": "0:59:38"}
+{"current_steps": 339, "total_steps": 573, "loss": 0.5706, "lr": 4.285681332340708e-06, "epoch": 1.7717770034843205, "percentage": 59.16, "elapsed_time": "1:26:02", "remaining_time": "0:59:23"}
+{"current_steps": 340, "total_steps": 573, "loss": 0.5904, "lr": 4.255506777684041e-06, "epoch": 1.7770034843205575, "percentage": 59.34, "elapsed_time": "1:26:15", "remaining_time": "0:59:06"}
+{"current_steps": 341, "total_steps": 573, "loss": 0.5667, "lr": 4.225359927168379e-06, "epoch": 1.7822299651567945, "percentage": 59.51, "elapsed_time": "1:26:32", "remaining_time": "0:58:52"}
+{"current_steps": 342, "total_steps": 573, "loss": 0.6049, "lr": 4.195241902620721e-06, "epoch": 1.7874564459930313, "percentage": 59.69, "elapsed_time": "1:26:48", "remaining_time": "0:58:37"}
+{"current_steps": 343, "total_steps": 573, "loss": 0.6078, "lr": 4.165153824795391e-06, "epoch": 1.7926829268292683, "percentage": 59.86, "elapsed_time": "1:27:01", "remaining_time": "0:58:21"}
+{"current_steps": 344, "total_steps": 573, "loss": 0.588, "lr": 4.135096813332333e-06, "epoch": 1.797909407665505, "percentage": 60.03, "elapsed_time": "1:27:13", "remaining_time": "0:58:03"}
+{"current_steps": 345, "total_steps": 573, "loss": 0.6015, "lr": 4.105071986715449e-06, "epoch": 1.8031358885017421, "percentage": 60.21, "elapsed_time": "1:27:30", "remaining_time": "0:57:49"}
+{"current_steps": 346, "total_steps": 573, "loss": 0.5697, "lr": 4.075080462230977e-06, "epoch": 1.8083623693379791, "percentage": 60.38, "elapsed_time": "1:27:46", "remaining_time": "0:57:35"}
+{"current_steps": 347, "total_steps": 573, "loss": 0.5698, "lr": 4.04512335592591e-06, "epoch": 1.8135888501742161, "percentage": 60.56, "elapsed_time": "1:28:00", "remaining_time": "0:57:19"}
+{"current_steps": 348, "total_steps": 573, "loss": 0.5787, "lr": 4.015201782566471e-06, "epoch": 1.8188153310104531, "percentage": 60.73, "elapsed_time": "1:28:15", "remaining_time": "0:57:03"}
+{"current_steps": 349, "total_steps": 573, "loss": 0.6, "lr": 3.985316855596627e-06, "epoch": 1.82404181184669, "percentage": 60.91, "elapsed_time": "1:28:27", "remaining_time": "0:56:46"}
+{"current_steps": 350, "total_steps": 573, "loss": 0.5759, "lr": 3.955469687096657e-06, "epoch": 1.8292682926829267, "percentage": 61.08, "elapsed_time": "1:28:43", "remaining_time": "0:56:31"}
+{"current_steps": 351, "total_steps": 573, "loss": 0.6505, "lr": 3.9256613877417715e-06, "epoch": 1.8344947735191637, "percentage": 61.26, "elapsed_time": "1:28:59", "remaining_time": "0:56:16"}
+{"current_steps": 352, "total_steps": 573, "loss": 0.5581, "lr": 3.895893066760774e-06, "epoch": 1.8397212543554007, "percentage": 61.43, "elapsed_time": "1:29:13", "remaining_time": "0:56:01"}
+{"current_steps": 353, "total_steps": 573, "loss": 0.6126, "lr": 3.866165831894796e-06, "epoch": 1.8449477351916377, "percentage": 61.61, "elapsed_time": "1:29:31", "remaining_time": "0:55:47"}
+{"current_steps": 354, "total_steps": 573, "loss": 0.6353, "lr": 3.836480789356063e-06, "epoch": 1.8501742160278747, "percentage": 61.78, "elapsed_time": "1:29:46", "remaining_time": "0:55:32"}
+{"current_steps": 355, "total_steps": 573, "loss": 0.5625, "lr": 3.806839043786738e-06, "epoch": 1.8554006968641115, "percentage": 61.95, "elapsed_time": "1:30:01", "remaining_time": "0:55:16"}
+{"current_steps": 356, "total_steps": 573, "loss": 0.5839, "lr": 3.777241698217819e-06, "epoch": 1.8606271777003485, "percentage": 62.13, "elapsed_time": "1:30:15", "remaining_time": "0:55:01"}
+{"current_steps": 357, "total_steps": 573, "loss": 0.596, "lr": 3.747689854028081e-06, "epoch": 1.8658536585365852, "percentage": 62.3, "elapsed_time": "1:30:31", "remaining_time": "0:54:46"}
+{"current_steps": 358, "total_steps": 573, "loss": 0.5786, "lr": 3.7181846109031007e-06, "epoch": 1.8710801393728222, "percentage": 62.48, "elapsed_time": "1:30:47", "remaining_time": "0:54:31"}
+{"current_steps": 359, "total_steps": 573, "loss": 0.5763, "lr": 3.6887270667943285e-06, "epoch": 1.8763066202090593, "percentage": 62.65, "elapsed_time": "1:31:02", "remaining_time": "0:54:16"}
+{"current_steps": 360, "total_steps": 573, "loss": 0.6022, "lr": 3.6593183178782454e-06, "epoch": 1.8815331010452963, "percentage": 62.83, "elapsed_time": "1:31:17", "remaining_time": "0:54:00"}
+{"current_steps": 361, "total_steps": 573, "loss": 0.5949, "lr": 3.6299594585155513e-06, "epoch": 1.8867595818815333, "percentage": 63.0, "elapsed_time": "1:31:30", "remaining_time": "0:53:44"}
+{"current_steps": 362, "total_steps": 573, "loss": 0.5868, "lr": 3.600651581210457e-06, "epoch": 1.89198606271777, "percentage": 63.18, "elapsed_time": "1:31:45", "remaining_time": "0:53:28"}
+{"current_steps": 363, "total_steps": 573, "loss": 0.5522, "lr": 3.5713957765700224e-06, "epoch": 1.8972125435540068, "percentage": 63.35, "elapsed_time": "1:32:01", "remaining_time": "0:53:14"}
+{"current_steps": 364, "total_steps": 573, "loss": 0.6013, "lr": 3.542193133263576e-06, "epoch": 1.9024390243902438, "percentage": 63.53, "elapsed_time": "1:32:18", "remaining_time": "0:52:59"}
+{"current_steps": 365, "total_steps": 573, "loss": 0.5448, "lr": 3.5130447379822076e-06, "epoch": 1.9076655052264808, "percentage": 63.7, "elapsed_time": "1:32:33", "remaining_time": "0:52:44"}
+{"current_steps": 366, "total_steps": 573, "loss": 0.6007, "lr": 3.483951675398316e-06, "epoch": 1.9128919860627178, "percentage": 63.87, "elapsed_time": "1:32:48", "remaining_time": "0:52:29"}
+{"current_steps": 367, "total_steps": 573, "loss": 0.5685, "lr": 3.4549150281252635e-06, "epoch": 1.9181184668989548, "percentage": 64.05, "elapsed_time": "1:33:01", "remaining_time": "0:52:12"}
+{"current_steps": 368, "total_steps": 573, "loss": 0.5857, "lr": 3.425935876677077e-06, "epoch": 1.9233449477351916, "percentage": 64.22, "elapsed_time": "1:33:09", "remaining_time": "0:51:53"}
+{"current_steps": 369, "total_steps": 573, "loss": 0.5809, "lr": 3.397015299428242e-06, "epoch": 1.9285714285714286, "percentage": 64.4, "elapsed_time": "1:33:23", "remaining_time": "0:51:37"}
+{"current_steps": 370, "total_steps": 573, "loss": 0.6067, "lr": 3.3681543725735843e-06, "epoch": 1.9337979094076654, "percentage": 64.57, "elapsed_time": "1:33:40", "remaining_time": "0:51:23"}
+{"current_steps": 371, "total_steps": 573, "loss": 0.5398, "lr": 3.339354170088207e-06, "epoch": 1.9390243902439024, "percentage": 64.75, "elapsed_time": "1:33:51", "remaining_time": "0:51:06"}
+{"current_steps": 372, "total_steps": 573, "loss": 0.5702, "lr": 3.3106157636875356e-06, "epoch": 1.9442508710801394, "percentage": 64.92, "elapsed_time": "1:34:08", "remaining_time": "0:50:51"}
+{"current_steps": 373, "total_steps": 573, "loss": 0.5925, "lr": 3.2819402227874364e-06, "epoch": 1.9494773519163764, "percentage": 65.1, "elapsed_time": "1:34:22", "remaining_time": "0:50:36"}
+{"current_steps": 374, "total_steps": 573, "loss": 0.6521, "lr": 3.253328614464414e-06, "epoch": 1.9547038327526134, "percentage": 65.27, "elapsed_time": "1:34:38", "remaining_time": "0:50:21"}
+{"current_steps": 375, "total_steps": 573, "loss": 0.5637, "lr": 3.2247820034159182e-06, "epoch": 1.9599303135888502, "percentage": 65.45, "elapsed_time": "1:34:54", "remaining_time": "0:50:06"}
+{"current_steps": 376, "total_steps": 573, "loss": 0.6078, "lr": 3.196301451920708e-06, "epoch": 1.965156794425087, "percentage": 65.62, "elapsed_time": "1:35:10", "remaining_time": "0:49:52"}
+{"current_steps": 377, "total_steps": 573, "loss": 0.614, "lr": 3.16788801979933e-06, "epoch": 1.970383275261324, "percentage": 65.79, "elapsed_time": "1:35:28", "remaining_time": "0:49:38"}
+{"current_steps": 378, "total_steps": 573, "loss": 0.5407, "lr": 3.1395427643746802e-06, "epoch": 1.975609756097561, "percentage": 65.97, "elapsed_time": "1:35:41", "remaining_time": "0:49:21"}
+{"current_steps": 379, "total_steps": 573, "loss": 0.5792, "lr": 3.111266740432658e-06, "epoch": 1.980836236933798, "percentage": 66.14, "elapsed_time": "1:35:56", "remaining_time": "0:49:06"}
+{"current_steps": 380, "total_steps": 573, "loss": 0.5552, "lr": 3.0830610001829173e-06, "epoch": 1.986062717770035, "percentage": 66.32, "elapsed_time": "1:36:11", "remaining_time": "0:48:51"}
+{"current_steps": 381, "total_steps": 573, "loss": 0.5474, "lr": 3.0549265932197055e-06, "epoch": 1.9912891986062717, "percentage": 66.49, "elapsed_time": "1:36:28", "remaining_time": "0:48:36"}
+{"current_steps": 382, "total_steps": 573, "loss": 0.5926, "lr": 3.026864566482813e-06, "epoch": 1.9965156794425087, "percentage": 66.67, "elapsed_time": "1:36:41", "remaining_time": "0:48:20"}
+{"current_steps": 383, "total_steps": 573, "loss": 0.9583, "lr": 2.99887596421861e-06, "epoch": 2.0017421602787455, "percentage": 66.84, "elapsed_time": "1:38:44", "remaining_time": "0:48:59"}
+{"current_steps": 384, "total_steps": 573, "loss": 0.4765, "lr": 2.9709618279411922e-06, "epoch": 2.0069686411149825, "percentage": 67.02, "elapsed_time": "1:38:57", "remaining_time": "0:48:42"}
+{"current_steps": 385, "total_steps": 573, "loss": 0.5513, "lr": 2.9431231963936176e-06, "epoch": 2.0121951219512195, "percentage": 67.19, "elapsed_time": "1:39:13", "remaining_time": "0:48:27"}
+{"current_steps": 386, "total_steps": 573, "loss": 0.5568, "lr": 2.9153611055092586e-06, "epoch": 2.0174216027874565, "percentage": 67.36, "elapsed_time": "1:39:28", "remaining_time": "0:48:11"}
+{"current_steps": 387, "total_steps": 573, "loss": 0.5378, "lr": 2.8876765883732447e-06, "epoch": 2.0226480836236935, "percentage": 67.54, "elapsed_time": "1:39:45", "remaining_time": "0:47:56"}
+{"current_steps": 388, "total_steps": 573, "loss": 0.5742, "lr": 2.860070675184036e-06, "epoch": 2.0278745644599305, "percentage": 67.71, "elapsed_time": "1:39:59", "remaining_time": "0:47:40"}
+{"current_steps": 389, "total_steps": 573, "loss": 0.5343, "lr": 2.8325443932150646e-06, "epoch": 2.033101045296167, "percentage": 67.89, "elapsed_time": "1:40:14", "remaining_time": "0:47:24"}
+{"current_steps": 390, "total_steps": 573, "loss": 0.5849, "lr": 2.805098766776529e-06, "epoch": 2.038327526132404, "percentage": 68.06, "elapsed_time": "1:40:29", "remaining_time": "0:47:09"}
+{"current_steps": 391, "total_steps": 573, "loss": 0.5105, "lr": 2.7777348171772604e-06, "epoch": 2.043554006968641, "percentage": 68.24, "elapsed_time": "1:40:46", "remaining_time": "0:46:54"}
+{"current_steps": 392, "total_steps": 573, "loss": 0.5143, "lr": 2.750453562686729e-06, "epoch": 2.048780487804878, "percentage": 68.41, "elapsed_time": "1:40:58", "remaining_time": "0:46:37"}