Training in progress, epoch 2

model-00001-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:26522f3bf0c9fd589821d95585a036944b7a0265912e81ee77c6d53598bb2cf2
+oid sha256:ae5230dd045ca93352c98ee529e019e430109ad39fae087f5e0e5ec779c2c7a9
 size 4903351912

model-00002-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4909691d17c9915141f391443cabe116b575dd085092641a21f282fb138c5b7c
+oid sha256:f23eb496c7fff3072bac8221ff676c744b8f6b7f7c0bdefda75675b824d881ef
 size 4947570872

model-00003-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5bd62e662b330f270c6efa5f74a2fc24cddb4e746d6c491c0967fe7942f8a9d7
+oid sha256:c3e3e6677a0633e0369b02d6d8d4df304a5a7464a8e01d9f4b8ba4b86c0a6ae2
 size 4962221464

model-00004-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2b906f2f1faf2efb2f66ef1f5142bd4a15db8efc56735c5f87a7a7691419ccec
+oid sha256:dd7aa48fff33bfd6d95daccc528e1d997adb06bfcf96907eef63ab3184e14846
 size 3670322200

trainer_log.jsonl

@@ -177,3 +177,91 @@
 {"current_steps": 1760, "total_steps": 2658, "loss": 0.5268, "learning_rate": 5e-06, "epoch": 1.9858956276445698, "percentage": 66.22, "elapsed_time": "22:17:50", "remaining_time": "11:22:36"}
 {"current_steps": 1770, "total_steps": 2658, "loss": 0.5215, "learning_rate": 5e-06, "epoch": 1.997179125528914, "percentage": 66.59, "elapsed_time": "22:25:22", "remaining_time": "11:14:58"}
 {"current_steps": 1772, "total_steps": 2658, "eval_loss": 0.5913873910903931, "epoch": 1.9994358251057829, "percentage": 66.67, "elapsed_time": "22:38:52", "remaining_time": "11:19:26"}
+{"current_steps": 1780, "total_steps": 2658, "loss": 0.5208, "learning_rate": 5e-06, "epoch": 2.008462623413258, "percentage": 66.97, "elapsed_time": "22:45:59", "remaining_time": "11:13:47"}
+{"current_steps": 1790, "total_steps": 2658, "loss": 0.4524, "learning_rate": 5e-06, "epoch": 2.0197461212976022, "percentage": 67.34, "elapsed_time": "22:53:30", "remaining_time": "11:06:02"}
+{"current_steps": 1800, "total_steps": 2658, "loss": 0.4428, "learning_rate": 5e-06, "epoch": 2.0310296191819464, "percentage": 67.72, "elapsed_time": "23:01:02", "remaining_time": "10:58:17"}
+{"current_steps": 1810, "total_steps": 2658, "loss": 0.4497, "learning_rate": 5e-06, "epoch": 2.0423131170662905, "percentage": 68.1, "elapsed_time": "23:08:33", "remaining_time": "10:50:33"}
+{"current_steps": 1820, "total_steps": 2658, "loss": 0.4501, "learning_rate": 5e-06, "epoch": 2.0535966149506346, "percentage": 68.47, "elapsed_time": "23:16:05", "remaining_time": "10:42:49"}
+{"current_steps": 1830, "total_steps": 2658, "loss": 0.4366, "learning_rate": 5e-06, "epoch": 2.0648801128349787, "percentage": 68.85, "elapsed_time": "23:23:37", "remaining_time": "10:35:04"}
+{"current_steps": 1840, "total_steps": 2658, "loss": 0.45, "learning_rate": 5e-06, "epoch": 2.076163610719323, "percentage": 69.22, "elapsed_time": "23:31:09", "remaining_time": "10:27:21"}
+{"current_steps": 1850, "total_steps": 2658, "loss": 0.4487, "learning_rate": 5e-06, "epoch": 2.087447108603667, "percentage": 69.6, "elapsed_time": "23:38:41", "remaining_time": "10:19:37"}
+{"current_steps": 1860, "total_steps": 2658, "loss": 0.4524, "learning_rate": 5e-06, "epoch": 2.098730606488011, "percentage": 69.98, "elapsed_time": "23:46:12", "remaining_time": "10:11:53"}
+{"current_steps": 1870, "total_steps": 2658, "loss": 0.4554, "learning_rate": 5e-06, "epoch": 2.1100141043723553, "percentage": 70.35, "elapsed_time": "23:53:44", "remaining_time": "10:04:09"}
+{"current_steps": 1880, "total_steps": 2658, "loss": 0.4545, "learning_rate": 5e-06, "epoch": 2.1212976022566994, "percentage": 70.73, "elapsed_time": "1 day, 0:01:15", "remaining_time": "9:56:26"}
+{"current_steps": 1890, "total_steps": 2658, "loss": 0.4576, "learning_rate": 5e-06, "epoch": 2.1325811001410435, "percentage": 71.11, "elapsed_time": "1 day, 0:08:47", "remaining_time": "9:48:42"}
+{"current_steps": 1900, "total_steps": 2658, "loss": 0.456, "learning_rate": 5e-06, "epoch": 2.143864598025388, "percentage": 71.48, "elapsed_time": "1 day, 0:16:19", "remaining_time": "9:40:59"}
+{"current_steps": 1910, "total_steps": 2658, "loss": 0.4555, "learning_rate": 5e-06, "epoch": 2.155148095909732, "percentage": 71.86, "elapsed_time": "1 day, 0:23:51", "remaining_time": "9:33:16"}
+{"current_steps": 1920, "total_steps": 2658, "loss": 0.4516, "learning_rate": 5e-06, "epoch": 2.1664315937940763, "percentage": 72.23, "elapsed_time": "1 day, 0:31:23", "remaining_time": "9:25:33"}
+{"current_steps": 1930, "total_steps": 2658, "loss": 0.4554, "learning_rate": 5e-06, "epoch": 2.1777150916784205, "percentage": 72.61, "elapsed_time": "1 day, 0:38:55", "remaining_time": "9:17:51"}
+{"current_steps": 1940, "total_steps": 2658, "loss": 0.4501, "learning_rate": 5e-06, "epoch": 2.1889985895627646, "percentage": 72.99, "elapsed_time": "1 day, 0:46:26", "remaining_time": "9:10:08"}
+{"current_steps": 1950, "total_steps": 2658, "loss": 0.4439, "learning_rate": 5e-06, "epoch": 2.2002820874471087, "percentage": 73.36, "elapsed_time": "1 day, 0:53:58", "remaining_time": "9:02:25"}
+{"current_steps": 1960, "total_steps": 2658, "loss": 0.4514, "learning_rate": 5e-06, "epoch": 2.211565585331453, "percentage": 73.74, "elapsed_time": "1 day, 1:01:30", "remaining_time": "8:54:43"}
+{"current_steps": 1970, "total_steps": 2658, "loss": 0.4443, "learning_rate": 5e-06, "epoch": 2.222849083215797, "percentage": 74.12, "elapsed_time": "1 day, 1:09:01", "remaining_time": "8:47:00"}
+{"current_steps": 1980, "total_steps": 2658, "loss": 0.4508, "learning_rate": 5e-06, "epoch": 2.234132581100141, "percentage": 74.49, "elapsed_time": "1 day, 1:16:33", "remaining_time": "8:39:18"}
+{"current_steps": 1990, "total_steps": 2658, "loss": 0.4533, "learning_rate": 5e-06, "epoch": 2.2454160789844853, "percentage": 74.87, "elapsed_time": "1 day, 1:24:05", "remaining_time": "8:31:36"}
+{"current_steps": 2000, "total_steps": 2658, "loss": 0.4591, "learning_rate": 5e-06, "epoch": 2.2566995768688294, "percentage": 75.24, "elapsed_time": "1 day, 1:31:36", "remaining_time": "8:23:53"}
+{"current_steps": 2010, "total_steps": 2658, "loss": 0.4583, "learning_rate": 5e-06, "epoch": 2.2679830747531735, "percentage": 75.62, "elapsed_time": "1 day, 1:39:08", "remaining_time": "8:16:11"}
+{"current_steps": 2020, "total_steps": 2658, "loss": 0.4665, "learning_rate": 5e-06, "epoch": 2.2792665726375176, "percentage": 76.0, "elapsed_time": "1 day, 1:46:39", "remaining_time": "8:08:30"}
+{"current_steps": 2030, "total_steps": 2658, "loss": 0.4626, "learning_rate": 5e-06, "epoch": 2.2905500705218618, "percentage": 76.37, "elapsed_time": "1 day, 1:54:11", "remaining_time": "8:00:48"}
+{"current_steps": 2040, "total_steps": 2658, "loss": 0.4529, "learning_rate": 5e-06, "epoch": 2.301833568406206, "percentage": 76.75, "elapsed_time": "1 day, 2:01:43", "remaining_time": "7:53:06"}
+{"current_steps": 2050, "total_steps": 2658, "loss": 0.4533, "learning_rate": 5e-06, "epoch": 2.31311706629055, "percentage": 77.13, "elapsed_time": "1 day, 2:09:15", "remaining_time": "7:45:25"}
+{"current_steps": 2060, "total_steps": 2658, "loss": 0.4694, "learning_rate": 5e-06, "epoch": 2.324400564174894, "percentage": 77.5, "elapsed_time": "1 day, 2:16:47", "remaining_time": "7:37:43"}
+{"current_steps": 2070, "total_steps": 2658, "loss": 0.4571, "learning_rate": 5e-06, "epoch": 2.3356840620592383, "percentage": 77.88, "elapsed_time": "1 day, 2:24:18", "remaining_time": "7:30:02"}
+{"current_steps": 2080, "total_steps": 2658, "loss": 0.4637, "learning_rate": 5e-06, "epoch": 2.3469675599435824, "percentage": 78.25, "elapsed_time": "1 day, 2:31:50", "remaining_time": "7:22:20"}
+{"current_steps": 2090, "total_steps": 2658, "loss": 0.454, "learning_rate": 5e-06, "epoch": 2.3582510578279265, "percentage": 78.63, "elapsed_time": "1 day, 2:39:22", "remaining_time": "7:14:39"}
+{"current_steps": 2100, "total_steps": 2658, "loss": 0.4621, "learning_rate": 5e-06, "epoch": 2.3695345557122707, "percentage": 79.01, "elapsed_time": "1 day, 2:46:54", "remaining_time": "7:06:58"}
+{"current_steps": 2110, "total_steps": 2658, "loss": 0.4616, "learning_rate": 5e-06, "epoch": 2.380818053596615, "percentage": 79.38, "elapsed_time": "1 day, 2:54:25", "remaining_time": "6:59:17"}
+{"current_steps": 2120, "total_steps": 2658, "loss": 0.4544, "learning_rate": 5e-06, "epoch": 2.392101551480959, "percentage": 79.76, "elapsed_time": "1 day, 3:01:57", "remaining_time": "6:51:36"}
+{"current_steps": 2130, "total_steps": 2658, "loss": 0.4612, "learning_rate": 5e-06, "epoch": 2.403385049365303, "percentage": 80.14, "elapsed_time": "1 day, 3:09:28", "remaining_time": "6:43:55"}
+{"current_steps": 2140, "total_steps": 2658, "loss": 0.4567, "learning_rate": 5e-06, "epoch": 2.414668547249647, "percentage": 80.51, "elapsed_time": "1 day, 3:17:00", "remaining_time": "6:36:14"}
+{"current_steps": 2150, "total_steps": 2658, "loss": 0.4624, "learning_rate": 5e-06, "epoch": 2.4259520451339913, "percentage": 80.89, "elapsed_time": "1 day, 3:24:32", "remaining_time": "6:28:34"}
+{"current_steps": 2160, "total_steps": 2658, "loss": 0.4621, "learning_rate": 5e-06, "epoch": 2.4372355430183354, "percentage": 81.26, "elapsed_time": "1 day, 3:32:05", "remaining_time": "6:20:53"}
+{"current_steps": 2170, "total_steps": 2658, "loss": 0.4602, "learning_rate": 5e-06, "epoch": 2.44851904090268, "percentage": 81.64, "elapsed_time": "1 day, 3:39:37", "remaining_time": "6:13:13"}
+{"current_steps": 2180, "total_steps": 2658, "loss": 0.461, "learning_rate": 5e-06, "epoch": 2.459802538787024, "percentage": 82.02, "elapsed_time": "1 day, 3:47:09", "remaining_time": "6:05:33"}
+{"current_steps": 2190, "total_steps": 2658, "loss": 0.4631, "learning_rate": 5e-06, "epoch": 2.4710860366713683, "percentage": 82.39, "elapsed_time": "1 day, 3:54:40", "remaining_time": "5:57:52"}
+{"current_steps": 2200, "total_steps": 2658, "loss": 0.4638, "learning_rate": 5e-06, "epoch": 2.4823695345557124, "percentage": 82.77, "elapsed_time": "1 day, 4:02:12", "remaining_time": "5:50:12"}
+{"current_steps": 2210, "total_steps": 2658, "loss": 0.4702, "learning_rate": 5e-06, "epoch": 2.4936530324400565, "percentage": 83.15, "elapsed_time": "1 day, 4:09:44", "remaining_time": "5:42:32"}
+{"current_steps": 2220, "total_steps": 2658, "loss": 0.4636, "learning_rate": 5e-06, "epoch": 2.5049365303244007, "percentage": 83.52, "elapsed_time": "1 day, 4:17:16", "remaining_time": "5:34:51"}
+{"current_steps": 2230, "total_steps": 2658, "loss": 0.457, "learning_rate": 5e-06, "epoch": 2.516220028208745, "percentage": 83.9, "elapsed_time": "1 day, 4:24:47", "remaining_time": "5:27:11"}
+{"current_steps": 2240, "total_steps": 2658, "loss": 0.4603, "learning_rate": 5e-06, "epoch": 2.527503526093089, "percentage": 84.27, "elapsed_time": "1 day, 4:32:19", "remaining_time": "5:19:31"}
+{"current_steps": 2250, "total_steps": 2658, "loss": 0.4598, "learning_rate": 5e-06, "epoch": 2.538787023977433, "percentage": 84.65, "elapsed_time": "1 day, 4:39:50", "remaining_time": "5:11:51"}
+{"current_steps": 2260, "total_steps": 2658, "loss": 0.4665, "learning_rate": 5e-06, "epoch": 2.550070521861777, "percentage": 85.03, "elapsed_time": "1 day, 4:47:22", "remaining_time": "5:04:12"}
+{"current_steps": 2270, "total_steps": 2658, "loss": 0.4612, "learning_rate": 5e-06, "epoch": 2.5613540197461213, "percentage": 85.4, "elapsed_time": "1 day, 4:54:54", "remaining_time": "4:56:32"}
+{"current_steps": 2280, "total_steps": 2658, "loss": 0.4607, "learning_rate": 5e-06, "epoch": 2.5726375176304654, "percentage": 85.78, "elapsed_time": "1 day, 5:02:26", "remaining_time": "4:48:52"}
+{"current_steps": 2290, "total_steps": 2658, "loss": 0.4549, "learning_rate": 5e-06, "epoch": 2.5839210155148096, "percentage": 86.16, "elapsed_time": "1 day, 5:09:58", "remaining_time": "4:41:13"}
+{"current_steps": 2300, "total_steps": 2658, "loss": 0.4694, "learning_rate": 5e-06, "epoch": 2.5952045133991537, "percentage": 86.53, "elapsed_time": "1 day, 5:17:30", "remaining_time": "4:33:33"}
+{"current_steps": 2310, "total_steps": 2658, "loss": 0.4663, "learning_rate": 5e-06, "epoch": 2.606488011283498, "percentage": 86.91, "elapsed_time": "1 day, 5:25:01", "remaining_time": "4:25:54"}
+{"current_steps": 2320, "total_steps": 2658, "loss": 0.4664, "learning_rate": 5e-06, "epoch": 2.617771509167842, "percentage": 87.28, "elapsed_time": "1 day, 5:32:33", "remaining_time": "4:18:14"}
+{"current_steps": 2330, "total_steps": 2658, "loss": 0.4543, "learning_rate": 5e-06, "epoch": 2.629055007052186, "percentage": 87.66, "elapsed_time": "1 day, 5:40:05", "remaining_time": "4:10:35"}
+{"current_steps": 2340, "total_steps": 2658, "loss": 0.4559, "learning_rate": 5e-06, "epoch": 2.64033850493653, "percentage": 88.04, "elapsed_time": "1 day, 5:47:37", "remaining_time": "4:02:55"}
+{"current_steps": 2350, "total_steps": 2658, "loss": 0.4635, "learning_rate": 5e-06, "epoch": 2.6516220028208743, "percentage": 88.41, "elapsed_time": "1 day, 5:55:08", "remaining_time": "3:55:16"}
+{"current_steps": 2360, "total_steps": 2658, "loss": 0.4686, "learning_rate": 5e-06, "epoch": 2.6629055007052185, "percentage": 88.79, "elapsed_time": "1 day, 6:02:40", "remaining_time": "3:47:37"}
+{"current_steps": 2370, "total_steps": 2658, "loss": 0.46, "learning_rate": 5e-06, "epoch": 2.6741889985895626, "percentage": 89.16, "elapsed_time": "1 day, 6:10:11", "remaining_time": "3:39:58"}
+{"current_steps": 2380, "total_steps": 2658, "loss": 0.4727, "learning_rate": 5e-06, "epoch": 2.685472496473907, "percentage": 89.54, "elapsed_time": "1 day, 6:17:43", "remaining_time": "3:32:19"}
+{"current_steps": 2390, "total_steps": 2658, "loss": 0.4667, "learning_rate": 5e-06, "epoch": 2.6967559943582513, "percentage": 89.92, "elapsed_time": "1 day, 6:25:14", "remaining_time": "3:24:40"}
+{"current_steps": 2400, "total_steps": 2658, "loss": 0.465, "learning_rate": 5e-06, "epoch": 2.7080394922425954, "percentage": 90.29, "elapsed_time": "1 day, 6:32:46", "remaining_time": "3:17:01"}
+{"current_steps": 2410, "total_steps": 2658, "loss": 0.4659, "learning_rate": 5e-06, "epoch": 2.7193229901269396, "percentage": 90.67, "elapsed_time": "1 day, 6:40:18", "remaining_time": "3:09:22"}
+{"current_steps": 2420, "total_steps": 2658, "loss": 0.4679, "learning_rate": 5e-06, "epoch": 2.7306064880112837, "percentage": 91.05, "elapsed_time": "1 day, 6:47:50", "remaining_time": "3:01:43"}
+{"current_steps": 2430, "total_steps": 2658, "loss": 0.4628, "learning_rate": 5e-06, "epoch": 2.741889985895628, "percentage": 91.42, "elapsed_time": "1 day, 6:55:22", "remaining_time": "2:54:05"}
+{"current_steps": 2440, "total_steps": 2658, "loss": 0.4654, "learning_rate": 5e-06, "epoch": 2.753173483779972, "percentage": 91.8, "elapsed_time": "1 day, 7:02:54", "remaining_time": "2:46:26"}
+{"current_steps": 2450, "total_steps": 2658, "loss": 0.4697, "learning_rate": 5e-06, "epoch": 2.764456981664316, "percentage": 92.17, "elapsed_time": "1 day, 7:10:26", "remaining_time": "2:38:47"}
+{"current_steps": 2460, "total_steps": 2658, "loss": 0.4705, "learning_rate": 5e-06, "epoch": 2.77574047954866, "percentage": 92.55, "elapsed_time": "1 day, 7:17:57", "remaining_time": "2:31:09"}
+{"current_steps": 2470, "total_steps": 2658, "loss": 0.4641, "learning_rate": 5e-06, "epoch": 2.7870239774330043, "percentage": 92.93, "elapsed_time": "1 day, 7:25:29", "remaining_time": "2:23:30"}
+{"current_steps": 2480, "total_steps": 2658, "loss": 0.4726, "learning_rate": 5e-06, "epoch": 2.7983074753173485, "percentage": 93.3, "elapsed_time": "1 day, 7:33:01", "remaining_time": "2:15:52"}
+{"current_steps": 2490, "total_steps": 2658, "loss": 0.4734, "learning_rate": 5e-06, "epoch": 2.8095909732016926, "percentage": 93.68, "elapsed_time": "1 day, 7:40:33", "remaining_time": "2:08:13"}
+{"current_steps": 2500, "total_steps": 2658, "loss": 0.4644, "learning_rate": 5e-06, "epoch": 2.8208744710860367, "percentage": 94.06, "elapsed_time": "1 day, 7:48:05", "remaining_time": "2:00:35"}
+{"current_steps": 2510, "total_steps": 2658, "loss": 0.4794, "learning_rate": 5e-06, "epoch": 2.832157968970381, "percentage": 94.43, "elapsed_time": "1 day, 7:55:37", "remaining_time": "1:52:57"}
+{"current_steps": 2520, "total_steps": 2658, "loss": 0.4628, "learning_rate": 5e-06, "epoch": 2.843441466854725, "percentage": 94.81, "elapsed_time": "1 day, 8:03:08", "remaining_time": "1:45:18"}
+{"current_steps": 2530, "total_steps": 2658, "loss": 0.4715, "learning_rate": 5e-06, "epoch": 2.854724964739069, "percentage": 95.18, "elapsed_time": "1 day, 8:10:40", "remaining_time": "1:37:40"}
+{"current_steps": 2540, "total_steps": 2658, "loss": 0.4633, "learning_rate": 5e-06, "epoch": 2.8660084626234132, "percentage": 95.56, "elapsed_time": "1 day, 8:18:12", "remaining_time": "1:30:02"}
+{"current_steps": 2550, "total_steps": 2658, "loss": 0.4689, "learning_rate": 5e-06, "epoch": 2.8772919605077574, "percentage": 95.94, "elapsed_time": "1 day, 8:25:43", "remaining_time": "1:22:24"}
+{"current_steps": 2560, "total_steps": 2658, "loss": 0.4659, "learning_rate": 5e-06, "epoch": 2.8885754583921015, "percentage": 96.31, "elapsed_time": "1 day, 8:33:15", "remaining_time": "1:14:46"}
+{"current_steps": 2570, "total_steps": 2658, "loss": 0.4668, "learning_rate": 5e-06, "epoch": 2.8998589562764456, "percentage": 96.69, "elapsed_time": "1 day, 8:40:47", "remaining_time": "1:07:08"}
+{"current_steps": 2580, "total_steps": 2658, "loss": 0.4723, "learning_rate": 5e-06, "epoch": 2.9111424541607898, "percentage": 97.07, "elapsed_time": "1 day, 8:48:19", "remaining_time": "0:59:30"}
+{"current_steps": 2590, "total_steps": 2658, "loss": 0.4631, "learning_rate": 5e-06, "epoch": 2.922425952045134, "percentage": 97.44, "elapsed_time": "1 day, 8:55:50", "remaining_time": "0:51:52"}
+{"current_steps": 2600, "total_steps": 2658, "loss": 0.4686, "learning_rate": 5e-06, "epoch": 2.933709449929478, "percentage": 97.82, "elapsed_time": "1 day, 9:03:22", "remaining_time": "0:44:14"}
+{"current_steps": 2610, "total_steps": 2658, "loss": 0.4641, "learning_rate": 5e-06, "epoch": 2.944992947813822, "percentage": 98.19, "elapsed_time": "1 day, 9:10:54", "remaining_time": "0:36:36"}
+{"current_steps": 2620, "total_steps": 2658, "loss": 0.4748, "learning_rate": 5e-06, "epoch": 2.9562764456981663, "percentage": 98.57, "elapsed_time": "1 day, 9:18:26", "remaining_time": "0:28:59"}
+{"current_steps": 2630, "total_steps": 2658, "loss": 0.472, "learning_rate": 5e-06, "epoch": 2.9675599435825104, "percentage": 98.95, "elapsed_time": "1 day, 9:25:58", "remaining_time": "0:21:21"}
+{"current_steps": 2640, "total_steps": 2658, "loss": 0.4627, "learning_rate": 5e-06, "epoch": 2.9788434414668545, "percentage": 99.32, "elapsed_time": "1 day, 9:33:30", "remaining_time": "0:13:43"}
+{"current_steps": 2650, "total_steps": 2658, "loss": 0.4672, "learning_rate": 5e-06, "epoch": 2.9901269393511987, "percentage": 99.7, "elapsed_time": "1 day, 9:41:01", "remaining_time": "0:06:06"}