Training in progress, step 2120

adapter_config.json

@@ -23,13 +23,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "down_proj",
-    "o_proj",
-    "gate_proj",
     "k_proj",
     "up_proj",
     "v_proj",
-    "q_proj"
+    "o_proj",
+    "down_proj",
+    "q_proj",
+    "gate_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dd625a3b2ffc781553fcda9db9e26e8ce40d5958fc820db9b1754e02fbd1824c
+oid sha256:788640632106b240e696686df5a835b0ed2407a7d55b40141baf52d658a52e52
 size 147770496

trainer_log.jsonl

@@ -1,20 +1,4 @@
-{"current_steps": 1890, "total_steps": 2553, "loss": 0.4759, "lr": 4.783509611417409e-05, "epoch": 2.2208516886930982, "percentage": 74.03, "elapsed_time": "0:33:03", "remaining_time": "0:11:35"}
-{"current_steps": 1900, "total_steps": 2553, "loss": 0.4725, "lr": 4.648611626713082e-05, "epoch": 2.232599118942731, "percentage": 74.42, "elapsed_time": "1:06:18", "remaining_time": "0:22:47"}
-{"current_steps": 1910, "total_steps": 2553, "loss": 0.4872, "lr": 4.515293434716279e-05, "epoch": 2.244346549192364, "percentage": 74.81, "elapsed_time": "1:39:33", "remaining_time": "0:33:30"}
-{"current_steps": 1920, "total_steps": 2553, "loss": 0.4839, "lr": 4.38357538198343e-05, "epoch": 2.256093979441997, "percentage": 75.21, "elapsed_time": "2:12:47", "remaining_time": "0:43:46"}
-{"current_steps": 1930, "total_steps": 2553, "loss": 0.4768, "lr": 4.253477570863275e-05, "epoch": 2.2678414096916297, "percentage": 75.6, "elapsed_time": "2:46:00", "remaining_time": "0:53:35"}
-{"current_steps": 1940, "total_steps": 2553, "loss": 0.4915, "lr": 4.1250198564289644e-05, "epoch": 2.2795888399412627, "percentage": 75.99, "elapsed_time": "3:19:18", "remaining_time": "1:02:58"}
-{"current_steps": 1950, "total_steps": 2553, "loss": 0.4773, "lr": 3.998221843447808e-05, "epoch": 2.2913362701908957, "percentage": 76.38, "elapsed_time": "3:52:20", "remaining_time": "1:11:50"}
-{"current_steps": 1960, "total_steps": 2553, "loss": 0.4728, "lr": 3.8731028833892955e-05, "epoch": 2.3030837004405287, "percentage": 76.77, "elapsed_time": "4:25:08", "remaining_time": "1:20:13"}
-{"current_steps": 1970, "total_steps": 2553, "loss": 0.4978, "lr": 3.749682071471727e-05, "epoch": 2.3148311306901617, "percentage": 77.16, "elapsed_time": "4:58:14", "remaining_time": "1:28:15"}
-{"current_steps": 1980, "total_steps": 2553, "loss": 0.491, "lr": 3.627978243747965e-05, "epoch": 2.3265785609397946, "percentage": 77.56, "elapsed_time": "5:31:32", "remaining_time": "1:35:56"}
-{"current_steps": 1990, "total_steps": 2553, "loss": 0.4672, "lr": 3.5080099742307495e-05, "epoch": 2.338325991189427, "percentage": 77.95, "elapsed_time": "6:04:49", "remaining_time": "1:43:12"}
-{"current_steps": 2000, "total_steps": 2553, "loss": 0.4856, "lr": 3.3897955720579985e-05, "epoch": 2.35007342143906, "percentage": 78.34, "elapsed_time": "6:38:07", "remaining_time": "1:50:05"}
-{"current_steps": 2010, "total_steps": 2553, "loss": 0.4893, "lr": 3.2733530786985124e-05, "epoch": 2.361820851688693, "percentage": 78.73, "elapsed_time": "7:11:25", "remaining_time": "1:56:32"}
-{"current_steps": 2020, "total_steps": 2553, "loss": 0.4737, "lr": 3.1587002651985776e-05, "epoch": 2.373568281938326, "percentage": 79.12, "elapsed_time": "7:44:36", "remaining_time": "2:02:35"}
-{"current_steps": 2030, "total_steps": 2553, "loss": 0.4883, "lr": 3.0458546294697954e-05, "epoch": 2.3853157121879587, "percentage": 79.51, "elapsed_time": "8:17:53", "remaining_time": "2:08:16"}
-{"current_steps": 2040, "total_steps": 2553, "loss": 0.4761, "lr": 2.9348333936186003e-05, "epoch": 2.3970631424375917, "percentage": 79.91, "elapsed_time": "8:51:08", "remaining_time": "2:13:33"}
-{"current_steps": 2050, "total_steps": 2553, "loss": 0.4867, "lr": 2.82565350131791e-05, "epoch": 2.4088105726872246, "percentage": 80.3, "elapsed_time": "9:24:22", "remaining_time": "2:18:28"}
-{"current_steps": 2060, "total_steps": 2553, "loss": 0.4759, "lr": 2.718331615221218e-05, "epoch": 2.4205580029368576, "percentage": 80.69, "elapsed_time": "9:57:12", "remaining_time": "2:22:55"}
-{"current_steps": 2070, "total_steps": 2553, "loss": 0.4881, "lr": 2.61288411441961e-05, "epoch": 2.4323054331864906, "percentage": 81.08, "elapsed_time": "10:30:13", "remaining_time": "2:27:03"}
-{"current_steps": 2080, "total_steps": 2553, "loss": 0.4973, "lr": 2.5093270919420383e-05, "epoch": 2.444052863436123, "percentage": 81.47, "elapsed_time": "11:03:02", "remaining_time": "2:30:46"}
+{"current_steps": 2090, "total_steps": 2553, "loss": 0.4887, "lr": 2.4076763522992665e-05, "epoch": 2.455800293685756, "percentage": 81.86, "elapsed_time": "0:28:01", "remaining_time": "0:06:12"}
+{"current_steps": 2100, "total_steps": 2553, "loss": 0.4791, "lr": 2.307947409071825e-05, "epoch": 2.467547723935389, "percentage": 82.26, "elapsed_time": "0:56:03", "remaining_time": "0:12:05"}
+{"current_steps": 2110, "total_steps": 2553, "loss": 0.4822, "lr": 2.210155482542402e-05, "epoch": 2.479295154185022, "percentage": 82.65, "elapsed_time": "1:24:06", "remaining_time": "0:17:39"}
+{"current_steps": 2120, "total_steps": 2553, "loss": 0.4776, "lr": 2.1143154973729735e-05, "epoch": 2.491042584434655, "percentage": 83.04, "elapsed_time": "1:52:11", "remaining_time": "0:22:54"}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c5002525d09503fe411fa35dafc4fa4f6d8fa9e642e00e997880eefc4a0c6823
+oid sha256:5f37c4f9def600a9e6150e724b6c67a6e551e6b342060fe9dc2e069df8930045
 size 5752