Training in progress, epoch 3

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ce228c40dbecdd83156d0f2a95ef943cdcc53a62aecff4ff6cfef48d2dfee7c3
+oid sha256:ac3a7293a6971fca6384d7e6feae6e242da2181f4856dcafe018e177837313d9
 size 4949453792

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1be1a3ef0c4aa97e6a94605fb01f26b4daaa57473825ab24ff3f4d4e24bb84fe
+oid sha256:0bd549f9eab04f1d7d279e4a6143a4c14570df3d0065cfe204836da6c8ea2101
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5c059aa20058537cefa965664ab4b4a7d88b0a17ba973e97d356b5a6ec26b140
+oid sha256:060941567516432e4d078278410892605b1472c36ec0d9bd593ef4126268a57a
 size 4546807800

trainer_log.jsonl

@@ -100,3 +100,52 @@
 {"current_steps": 980, "total_steps": 1470, "eval_loss": 0.5511366724967957, "epoch": 2.0, "percentage": 66.67, "elapsed_time": "1:39:27", "remaining_time": "0:49:43"}
 {"current_steps": 990, "total_steps": 1470, "loss": 0.3994, "lr": 1.689915580201614e-06, "epoch": 2.020408163265306, "percentage": 67.35, "elapsed_time": "1:41:35", "remaining_time": "0:49:15"}
 {"current_steps": 1000, "total_steps": 1470, "loss": 0.4039, "lr": 1.6455253728912765e-06, "epoch": 2.0408163265306123, "percentage": 68.03, "elapsed_time": "1:42:35", "remaining_time": "0:48:13"}
+{"current_steps": 1010, "total_steps": 1470, "loss": 0.3966, "lr": 1.6016944932753123e-06, "epoch": 2.061224489795918, "percentage": 68.71, "elapsed_time": "1:43:35", "remaining_time": "0:47:10"}
+{"current_steps": 1020, "total_steps": 1470, "loss": 0.3956, "lr": 1.558445138174033e-06, "epoch": 2.0816326530612246, "percentage": 69.39, "elapsed_time": "1:44:35", "remaining_time": "0:46:08"}
+{"current_steps": 1030, "total_steps": 1470, "loss": 0.3993, "lr": 1.5157992099122896e-06, "epoch": 2.1020408163265305, "percentage": 70.07, "elapsed_time": "1:45:35", "remaining_time": "0:45:06"}
+{"current_steps": 1040, "total_steps": 1470, "loss": 0.3961, "lr": 1.4737783052277017e-06, "epoch": 2.122448979591837, "percentage": 70.75, "elapsed_time": "1:46:35", "remaining_time": "0:44:04"}
+{"current_steps": 1050, "total_steps": 1470, "loss": 0.3947, "lr": 1.432403704333643e-06, "epoch": 2.142857142857143, "percentage": 71.43, "elapsed_time": "1:47:35", "remaining_time": "0:43:02"}
+{"current_steps": 1060, "total_steps": 1470, "loss": 0.3969, "lr": 1.3916963601425293e-06, "epoch": 2.163265306122449, "percentage": 72.11, "elapsed_time": "1:48:36", "remaining_time": "0:42:00"}
+{"current_steps": 1070, "total_steps": 1470, "loss": 0.393, "lr": 1.3516768876548404e-06, "epoch": 2.183673469387755, "percentage": 72.79, "elapsed_time": "1:49:36", "remaining_time": "0:40:58"}
+{"current_steps": 1080, "total_steps": 1470, "loss": 0.3951, "lr": 1.3123655535192873e-06, "epoch": 2.204081632653061, "percentage": 73.47, "elapsed_time": "1:50:36", "remaining_time": "0:39:56"}
+{"current_steps": 1090, "total_steps": 1470, "loss": 0.3979, "lr": 1.2737822657693763e-06, "epoch": 2.2244897959183674, "percentage": 74.15, "elapsed_time": "1:51:36", "remaining_time": "0:38:54"}
+{"current_steps": 1100, "total_steps": 1470, "loss": 0.3997, "lr": 1.235946563741594e-06, "epoch": 2.2448979591836733, "percentage": 74.83, "elapsed_time": "1:52:36", "remaining_time": "0:37:52"}
+{"current_steps": 1110, "total_steps": 1470, "loss": 0.3899, "lr": 1.1988776081802975e-06, "epoch": 2.2653061224489797, "percentage": 75.51, "elapsed_time": "1:53:36", "remaining_time": "0:36:50"}
+{"current_steps": 1120, "total_steps": 1470, "loss": 0.3966, "lr": 1.1625941715343404e-06, "epoch": 2.2857142857142856, "percentage": 76.19, "elapsed_time": "1:54:40", "remaining_time": "0:35:50"}
+{"current_steps": 1130, "total_steps": 1470, "loss": 0.396, "lr": 1.1271146284503326e-06, "epoch": 2.306122448979592, "percentage": 76.87, "elapsed_time": "1:55:41", "remaining_time": "0:34:48"}
+{"current_steps": 1140, "total_steps": 1470, "loss": 0.3904, "lr": 1.0924569464673593e-06, "epoch": 2.326530612244898, "percentage": 77.55, "elapsed_time": "1:56:41", "remaining_time": "0:33:46"}
+{"current_steps": 1150, "total_steps": 1470, "loss": 0.3975, "lr": 1.0586386769178644e-06, "epoch": 2.3469387755102042, "percentage": 78.23, "elapsed_time": "1:57:41", "remaining_time": "0:32:44"}
+{"current_steps": 1160, "total_steps": 1470, "loss": 0.3967, "lr": 1.0256769460393087e-06, "epoch": 2.36734693877551, "percentage": 78.91, "elapsed_time": "1:58:42", "remaining_time": "0:31:43"}
+{"current_steps": 1170, "total_steps": 1470, "loss": 0.3996, "lr": 9.935884463011108e-07, "epoch": 2.387755102040816, "percentage": 79.59, "elapsed_time": "1:59:42", "remaining_time": "0:30:41"}
+{"current_steps": 1180, "total_steps": 1470, "loss": 0.396, "lr": 9.623894279512455e-07, "epoch": 2.4081632653061225, "percentage": 80.27, "elapsed_time": "2:00:41", "remaining_time": "0:29:39"}
+{"current_steps": 1190, "total_steps": 1470, "loss": 0.3937, "lr": 9.320956907868051e-07, "epoch": 2.4285714285714284, "percentage": 80.95, "elapsed_time": "2:01:42", "remaining_time": "0:28:38"}
+{"current_steps": 1200, "total_steps": 1470, "loss": 0.3952, "lr": 9.027225761526653e-07, "epoch": 2.4489795918367347, "percentage": 81.63, "elapsed_time": "2:02:42", "remaining_time": "0:27:36"}
+{"current_steps": 1210, "total_steps": 1470, "loss": 0.3959, "lr": 8.742849591723315e-07, "epoch": 2.4693877551020407, "percentage": 82.31, "elapsed_time": "2:03:42", "remaining_time": "0:26:34"}
+{"current_steps": 1220, "total_steps": 1470, "loss": 0.398, "lr": 8.467972412148767e-07, "epoch": 2.489795918367347, "percentage": 82.99, "elapsed_time": "2:04:42", "remaining_time": "0:25:33"}
+{"current_steps": 1230, "total_steps": 1470, "loss": 0.3964, "lr": 8.202733426018117e-07, "epoch": 2.510204081632653, "percentage": 83.67, "elapsed_time": "2:05:43", "remaining_time": "0:24:31"}
+{"current_steps": 1240, "total_steps": 1470, "loss": 0.3927, "lr": 7.947266955575562e-07, "epoch": 2.5306122448979593, "percentage": 84.35, "elapsed_time": "2:06:43", "remaining_time": "0:23:30"}
+{"current_steps": 1250, "total_steps": 1470, "loss": 0.395, "lr": 7.701702374070962e-07, "epoch": 2.5510204081632653, "percentage": 85.03, "elapsed_time": "2:07:44", "remaining_time": "0:22:28"}
+{"current_steps": 1260, "total_steps": 1470, "loss": 0.393, "lr": 7.466164040242694e-07, "epoch": 2.571428571428571, "percentage": 85.71, "elapsed_time": "2:08:44", "remaining_time": "0:21:27"}
+{"current_steps": 1270, "total_steps": 1470, "loss": 0.3954, "lr": 7.24077123533992e-07, "epoch": 2.5918367346938775, "percentage": 86.39, "elapsed_time": "2:09:44", "remaining_time": "0:20:25"}
+{"current_steps": 1280, "total_steps": 1470, "loss": 0.3956, "lr": 7.025638102716238e-07, "epoch": 2.612244897959184, "percentage": 87.07, "elapsed_time": "2:10:45", "remaining_time": "0:19:24"}
+{"current_steps": 1290, "total_steps": 1470, "loss": 0.3976, "lr": 6.820873590025216e-07, "epoch": 2.63265306122449, "percentage": 87.76, "elapsed_time": "2:11:45", "remaining_time": "0:18:23"}
+{"current_steps": 1300, "total_steps": 1470, "loss": 0.3984, "lr": 6.626581394047174e-07, "epoch": 2.6530612244897958, "percentage": 88.44, "elapsed_time": "2:12:46", "remaining_time": "0:17:21"}
+{"current_steps": 1310, "total_steps": 1470, "loss": 0.3959, "lr": 6.442859908175084e-07, "epoch": 2.673469387755102, "percentage": 89.12, "elapsed_time": "2:13:46", "remaining_time": "0:16:20"}
+{"current_steps": 1320, "total_steps": 1470, "loss": 0.397, "lr": 6.269802172586215e-07, "epoch": 2.693877551020408, "percentage": 89.8, "elapsed_time": "2:14:46", "remaining_time": "0:15:18"}
+{"current_steps": 1330, "total_steps": 1470, "loss": 0.39, "lr": 6.107495827124764e-07, "epoch": 2.7142857142857144, "percentage": 90.48, "elapsed_time": "2:15:47", "remaining_time": "0:14:17"}
+{"current_steps": 1340, "total_steps": 1470, "loss": 0.3959, "lr": 5.956023066919313e-07, "epoch": 2.7346938775510203, "percentage": 91.16, "elapsed_time": "2:16:48", "remaining_time": "0:13:16"}
+{"current_steps": 1350, "total_steps": 1470, "loss": 0.3959, "lr": 5.815460600757599e-07, "epoch": 2.7551020408163263, "percentage": 91.84, "elapsed_time": "2:17:48", "remaining_time": "0:12:15"}
+{"current_steps": 1360, "total_steps": 1470, "loss": 0.3966, "lr": 5.685879612239649e-07, "epoch": 2.7755102040816326, "percentage": 92.52, "elapsed_time": "2:18:49", "remaining_time": "0:11:13"}
+{"current_steps": 1370, "total_steps": 1470, "loss": 0.3981, "lr": 5.567345723729062e-07, "epoch": 2.795918367346939, "percentage": 93.2, "elapsed_time": "2:19:49", "remaining_time": "0:10:12"}
+{"current_steps": 1380, "total_steps": 1470, "loss": 0.3958, "lr": 5.459918963120477e-07, "epoch": 2.816326530612245, "percentage": 93.88, "elapsed_time": "2:20:49", "remaining_time": "0:09:11"}
+{"current_steps": 1390, "total_steps": 1470, "loss": 0.3937, "lr": 5.363653733440328e-07, "epoch": 2.836734693877551, "percentage": 94.56, "elapsed_time": "2:21:50", "remaining_time": "0:08:09"}
+{"current_steps": 1400, "total_steps": 1470, "loss": 0.3955, "lr": 5.278598785296044e-07, "epoch": 2.857142857142857, "percentage": 95.24, "elapsed_time": "2:22:51", "remaining_time": "0:07:08"}
+{"current_steps": 1410, "total_steps": 1470, "loss": 0.3935, "lr": 5.2047971921878e-07, "epoch": 2.877551020408163, "percentage": 95.92, "elapsed_time": "2:23:51", "remaining_time": "0:06:07"}
+{"current_steps": 1420, "total_steps": 1470, "loss": 0.3991, "lr": 5.142286328695235e-07, "epoch": 2.8979591836734695, "percentage": 96.6, "elapsed_time": "2:24:51", "remaining_time": "0:05:06"}
+{"current_steps": 1430, "total_steps": 1470, "loss": 0.3984, "lr": 5.091097851550239e-07, "epoch": 2.9183673469387754, "percentage": 97.28, "elapsed_time": "2:25:52", "remaining_time": "0:04:04"}
+{"current_steps": 1440, "total_steps": 1470, "loss": 0.3977, "lr": 5.051257683605363e-07, "epoch": 2.938775510204082, "percentage": 97.96, "elapsed_time": "2:26:53", "remaining_time": "0:03:03"}
+{"current_steps": 1450, "total_steps": 1470, "loss": 0.3948, "lr": 5.022786000705963e-07, "epoch": 2.9591836734693877, "percentage": 98.64, "elapsed_time": "2:27:53", "remaining_time": "0:02:02"}
+{"current_steps": 1460, "total_steps": 1470, "loss": 0.3949, "lr": 5.005697221472769e-07, "epoch": 2.979591836734694, "percentage": 99.32, "elapsed_time": "2:28:53", "remaining_time": "0:01:01"}
+{"current_steps": 1470, "total_steps": 1470, "loss": 0.3994, "lr": 5e-07, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "2:29:53", "remaining_time": "0:00:00"}
+{"current_steps": 1470, "total_steps": 1470, "eval_loss": 0.5735070705413818, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "2:31:44", "remaining_time": "0:00:00"}
+{"current_steps": 1470, "total_steps": 1470, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "2:32:41", "remaining_time": "0:00:00"}