Training in progress, epoch 2

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:41455b4a477a2745e997aceaa5b5cfe7231c0fc4084831de9cd1fd7d6b536cc7
+oid sha256:e7b39f750f80884d8c0560597a9994715815199cdca596a7c2de6bb1132a1c12
 size 4943162336

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c7e8d5d455978adae36dd97dd4d19d32f197a10bd02c50b8dfbbd69917945676
+oid sha256:ea9064e2a3883c2f7d2810b05257c94043e0d41973bbafc26cf0edbc7a19d49f
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:417e72b8c3b122e844654094702c371e5e0b23595b4fbdef80d6122c4ae00f9d
+oid sha256:f5be67176a93b9e0ddf50eeb55211983838c11749e6ef4153298500995475984
 size 4540516344

trainer_log.jsonl

@@ -16,3 +16,20 @@
 {"current_steps": 160, "total_steps": 504, "loss": 0.4983, "lr": 4.53966038240406e-06, "epoch": 0.9523809523809523, "percentage": 31.75, "elapsed_time": "2:34:06", "remaining_time": "5:31:20"}
 {"current_steps": 168, "total_steps": 504, "eval_loss": 0.061965156346559525, "epoch": 1.0, "percentage": 33.33, "elapsed_time": "2:44:44", "remaining_time": "5:29:28"}
 {"current_steps": 170, "total_steps": 504, "loss": 0.4899, "lr": 4.428153298183676e-06, "epoch": 1.0119047619047619, "percentage": 33.73, "elapsed_time": "2:47:24", "remaining_time": "5:28:53"}
+{"current_steps": 180, "total_steps": 504, "loss": 0.4566, "lr": 4.306262355332006e-06, "epoch": 1.0714285714285714, "percentage": 35.71, "elapsed_time": "2:57:02", "remaining_time": "5:18:40"}
+{"current_steps": 190, "total_steps": 504, "loss": 0.4541, "lr": 4.17464398418589e-06, "epoch": 1.130952380952381, "percentage": 37.7, "elapsed_time": "3:06:40", "remaining_time": "5:08:30"}
+{"current_steps": 200, "total_steps": 504, "loss": 0.4525, "lr": 4.034007001082985e-06, "epoch": 1.1904761904761905, "percentage": 39.68, "elapsed_time": "3:16:18", "remaining_time": "4:58:23"}
+{"current_steps": 210, "total_steps": 504, "loss": 0.4533, "lr": 3.8851087911089315e-06, "epoch": 1.25, "percentage": 41.67, "elapsed_time": "3:25:56", "remaining_time": "4:48:18"}
+{"current_steps": 220, "total_steps": 504, "loss": 0.4491, "lr": 3.7287512292828364e-06, "epoch": 1.3095238095238095, "percentage": 43.65, "elapsed_time": "3:35:34", "remaining_time": "4:38:17"}
+{"current_steps": 230, "total_steps": 504, "loss": 0.4493, "lr": 3.5657763621470905e-06, "epoch": 1.369047619047619, "percentage": 45.63, "elapsed_time": "3:45:12", "remaining_time": "4:28:17"}
+{"current_steps": 240, "total_steps": 504, "loss": 0.4476, "lr": 3.39706187301784e-06, "epoch": 1.4285714285714286, "percentage": 47.62, "elapsed_time": "3:54:50", "remaining_time": "4:18:19"}
+{"current_steps": 250, "total_steps": 504, "loss": 0.4437, "lr": 3.2235163553175174e-06, "epoch": 1.4880952380952381, "percentage": 49.6, "elapsed_time": "4:04:28", "remaining_time": "4:08:23"}
+{"current_steps": 260, "total_steps": 504, "loss": 0.4423, "lr": 3.0460744194443658e-06, "epoch": 1.5476190476190477, "percentage": 51.59, "elapsed_time": "4:14:07", "remaining_time": "3:58:28"}
+{"current_steps": 270, "total_steps": 504, "loss": 0.44, "lr": 2.8656916595304026e-06, "epoch": 1.6071428571428572, "percentage": 53.57, "elapsed_time": "4:23:45", "remaining_time": "3:48:35"}
+{"current_steps": 280, "total_steps": 504, "loss": 0.4404, "lr": 2.68333950719376e-06, "epoch": 1.6666666666666665, "percentage": 55.56, "elapsed_time": "4:33:23", "remaining_time": "3:38:42"}
+{"current_steps": 290, "total_steps": 504, "loss": 0.4397, "lr": 2.5e-06, "epoch": 1.7261904761904763, "percentage": 57.54, "elapsed_time": "4:43:02", "remaining_time": "3:28:51"}
+{"current_steps": 300, "total_steps": 504, "loss": 0.4351, "lr": 2.3166604928062407e-06, "epoch": 1.7857142857142856, "percentage": 59.52, "elapsed_time": "4:52:40", "remaining_time": "3:19:00"}
+{"current_steps": 310, "total_steps": 504, "loss": 0.4341, "lr": 2.1343083404695987e-06, "epoch": 1.8452380952380953, "percentage": 61.51, "elapsed_time": "5:02:18", "remaining_time": "3:09:11"}
+{"current_steps": 320, "total_steps": 504, "loss": 0.4338, "lr": 1.9539255805556346e-06, "epoch": 1.9047619047619047, "percentage": 63.49, "elapsed_time": "5:11:56", "remaining_time": "2:59:22"}
+{"current_steps": 330, "total_steps": 504, "loss": 0.4333, "lr": 1.7764836446824835e-06, "epoch": 1.9642857142857144, "percentage": 65.48, "elapsed_time": "5:21:34", "remaining_time": "2:49:33"}
+{"current_steps": 336, "total_steps": 504, "eval_loss": 0.05795557424426079, "epoch": 2.0, "percentage": 66.67, "elapsed_time": "5:30:16", "remaining_time": "2:45:08"}