Training in progress, epoch 2

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:02b9764647013fde9c3a69bebec5ebfc0293b3d85d923f03718f141916380ffc
+oid sha256:4eb121464670505322f39c00b4d938b9a79a9c8fc3951397958e008c806cddf5
 size 4943162336

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9ae8ffa64dca13106d2ac401979adf36da8370e9d326f698f52058d6329d1fdb
+oid sha256:33293e3a8986eb4524edff6a42009d8dc1c09f9f6e33451e6253bdf5e92ff316
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4ada9d9c2d76440ec5e3d41951f087fd1787c589f3456d57a1758f4a448f1c41
+oid sha256:c89cafdd7aa03b2e26ef31d69ecdfc1397fad448a4ad2e94f2db70c6dedc8b5c
 size 4540516344

trainer_log.jsonl

@@ -16,3 +16,20 @@
 {"current_steps": 160, "total_steps": 504, "loss": 0.5294, "lr": 1.861898114721218e-06, "epoch": 0.9523809523809523, "percentage": 31.75, "elapsed_time": "1:27:55", "remaining_time": "3:09:03"}
 {"current_steps": 168, "total_steps": 504, "eval_loss": 0.06584873795509338, "epoch": 1.0, "percentage": 33.33, "elapsed_time": "1:34:14", "remaining_time": "3:08:28"}
 {"current_steps": 170, "total_steps": 504, "loss": 0.5242, "lr": 1.8284459894551025e-06, "epoch": 1.0119047619047619, "percentage": 33.73, "elapsed_time": "1:36:04", "remaining_time": "3:08:46"}
+{"current_steps": 180, "total_steps": 504, "loss": 0.506, "lr": 1.7918787065996015e-06, "epoch": 1.0714285714285714, "percentage": 35.71, "elapsed_time": "1:41:32", "remaining_time": "3:02:47"}
+{"current_steps": 190, "total_steps": 504, "loss": 0.5037, "lr": 1.7523931952557666e-06, "epoch": 1.130952380952381, "percentage": 37.7, "elapsed_time": "1:47:01", "remaining_time": "2:56:52"}
+{"current_steps": 200, "total_steps": 504, "loss": 0.5017, "lr": 1.7102021003248955e-06, "epoch": 1.1904761904761905, "percentage": 39.68, "elapsed_time": "1:52:31", "remaining_time": "2:51:02"}
+{"current_steps": 210, "total_steps": 504, "loss": 0.5023, "lr": 1.6655326373326793e-06, "epoch": 1.25, "percentage": 41.67, "elapsed_time": "1:58:00", "remaining_time": "2:45:12"}
+{"current_steps": 220, "total_steps": 504, "loss": 0.497, "lr": 1.6186253687848507e-06, "epoch": 1.3095238095238095, "percentage": 43.65, "elapsed_time": "2:03:29", "remaining_time": "2:39:24"}
+{"current_steps": 230, "total_steps": 504, "loss": 0.4962, "lr": 1.569732908644127e-06, "epoch": 1.369047619047619, "percentage": 45.63, "elapsed_time": "2:08:56", "remaining_time": "2:33:36"}
+{"current_steps": 240, "total_steps": 504, "loss": 0.4943, "lr": 1.5191185619053519e-06, "epoch": 1.4285714285714286, "percentage": 47.62, "elapsed_time": "2:14:24", "remaining_time": "2:27:50"}
+{"current_steps": 250, "total_steps": 504, "loss": 0.4903, "lr": 1.4670549065952552e-06, "epoch": 1.4880952380952381, "percentage": 49.6, "elapsed_time": "2:19:54", "remaining_time": "2:22:08"}
+{"current_steps": 260, "total_steps": 504, "loss": 0.4885, "lr": 1.4138223258333096e-06, "epoch": 1.5476190476190477, "percentage": 51.59, "elapsed_time": "2:25:23", "remaining_time": "2:16:26"}
+{"current_steps": 270, "total_steps": 504, "loss": 0.4864, "lr": 1.3597074978591206e-06, "epoch": 1.6071428571428572, "percentage": 53.57, "elapsed_time": "2:30:50", "remaining_time": "2:10:44"}
+{"current_steps": 280, "total_steps": 504, "loss": 0.4871, "lr": 1.3050018521581279e-06, "epoch": 1.6666666666666665, "percentage": 55.56, "elapsed_time": "2:36:19", "remaining_time": "2:05:03"}
+{"current_steps": 290, "total_steps": 504, "loss": 0.4864, "lr": 1.2499999999999999e-06, "epoch": 1.7261904761904763, "percentage": 57.54, "elapsed_time": "2:41:50", "remaining_time": "1:59:25"}
+{"current_steps": 300, "total_steps": 504, "loss": 0.4813, "lr": 1.1949981478418721e-06, "epoch": 1.7857142857142856, "percentage": 59.52, "elapsed_time": "2:47:19", "remaining_time": "1:53:46"}
+{"current_steps": 310, "total_steps": 504, "loss": 0.4801, "lr": 1.1402925021408796e-06, "epoch": 1.8452380952380953, "percentage": 61.51, "elapsed_time": "2:52:47", "remaining_time": "1:48:08"}
+{"current_steps": 320, "total_steps": 504, "loss": 0.4795, "lr": 1.0861776741666901e-06, "epoch": 1.9047619047619047, "percentage": 63.49, "elapsed_time": "2:58:16", "remaining_time": "1:42:30"}
+{"current_steps": 330, "total_steps": 504, "loss": 0.4791, "lr": 1.032945093404745e-06, "epoch": 1.9642857142857144, "percentage": 65.48, "elapsed_time": "3:03:46", "remaining_time": "1:36:53"}
+{"current_steps": 336, "total_steps": 504, "eval_loss": 0.061718959361314774, "epoch": 2.0, "percentage": 66.67, "elapsed_time": "3:08:58", "remaining_time": "1:34:29"}