Training in progress, step 1920

adapter_config.json

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

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3562cd06abdaec56814cb61f23304fdeaca253c81d9dc09b0dd14141ca2d13b3
+oid sha256:417cba96386ae73badac7fde603c7dc4ffdc905626abd5178d5e76503e04a642
 size 147770496

trainer_log.jsonl

@@ -1,20 +1,4 @@
-{"current_steps": 1690, "total_steps": 2553, "loss": 0.5072, "lr": 7.779866385336391e-05, "epoch": 1.9856093979441996, "percentage": 66.2, "elapsed_time": "0:29:09", "remaining_time": "0:14:53"}
-{"current_steps": 1700, "total_steps": 2553, "loss": 0.5024, "lr": 7.617992717972585e-05, "epoch": 1.9973568281938325, "percentage": 66.59, "elapsed_time": "0:58:56", "remaining_time": "0:29:34"}
-{"current_steps": 1710, "total_steps": 2553, "loss": 0.541, "lr": 7.457245666754417e-05, "epoch": 2.0093979441997063, "percentage": 66.98, "elapsed_time": "1:28:55", "remaining_time": "0:43:50"}
-{"current_steps": 1720, "total_steps": 2553, "loss": 0.4831, "lr": 7.297649764334912e-05, "epoch": 2.0211453744493393, "percentage": 67.37, "elapsed_time": "1:58:39", "remaining_time": "0:57:28"}
-{"current_steps": 1730, "total_steps": 2553, "loss": 0.486, "lr": 7.139229367682778e-05, "epoch": 2.0328928046989723, "percentage": 67.76, "elapsed_time": "2:27:57", "remaining_time": "1:10:23"}
-{"current_steps": 1740, "total_steps": 2553, "loss": 0.4842, "lr": 6.982008654365156e-05, "epoch": 2.044640234948605, "percentage": 68.16, "elapsed_time": "2:57:39", "remaining_time": "1:23:00"}
-{"current_steps": 1750, "total_steps": 2553, "loss": 0.4911, "lr": 6.82601161885771e-05, "epoch": 2.056387665198238, "percentage": 68.55, "elapsed_time": "3:26:56", "remaining_time": "1:34:57"}
-{"current_steps": 1760, "total_steps": 2553, "loss": 0.4924, "lr": 6.671262068882665e-05, "epoch": 2.0681350954478708, "percentage": 68.94, "elapsed_time": "3:56:34", "remaining_time": "1:46:35"}
-{"current_steps": 1770, "total_steps": 2553, "loss": 0.4736, "lr": 6.517783621775382e-05, "epoch": 2.0798825256975038, "percentage": 69.33, "elapsed_time": "4:26:04", "remaining_time": "1:57:42"}
-{"current_steps": 1780, "total_steps": 2553, "loss": 0.4824, "lr": 6.36559970087992e-05, "epoch": 2.0916299559471367, "percentage": 69.72, "elapsed_time": "4:55:46", "remaining_time": "2:08:26"}
-{"current_steps": 1790, "total_steps": 2553, "loss": 0.4834, "lr": 6.214733531974292e-05, "epoch": 2.1033773861967693, "percentage": 70.11, "elapsed_time": "5:25:11", "remaining_time": "2:18:37"}
-{"current_steps": 1800, "total_steps": 2553, "loss": 0.4889, "lr": 6.065208139725811e-05, "epoch": 2.1151248164464023, "percentage": 70.51, "elapsed_time": "5:54:50", "remaining_time": "2:28:26"}
-{"current_steps": 1810, "total_steps": 2553, "loss": 0.4893, "lr": 5.917046344177123e-05, "epoch": 2.1268722466960353, "percentage": 70.9, "elapsed_time": "6:24:19", "remaining_time": "2:37:46"}
-{"current_steps": 1820, "total_steps": 2553, "loss": 0.4876, "lr": 5.770270757263536e-05, "epoch": 2.1386196769456682, "percentage": 71.29, "elapsed_time": "6:53:58", "remaining_time": "2:46:43"}
-{"current_steps": 1830, "total_steps": 2553, "loss": 0.4764, "lr": 5.624903779362031e-05, "epoch": 2.150367107195301, "percentage": 71.68, "elapsed_time": "7:23:26", "remaining_time": "2:55:11"}
-{"current_steps": 1840, "total_steps": 2553, "loss": 0.4781, "lr": 5.480967595872602e-05, "epoch": 2.1621145374449338, "percentage": 72.07, "elapsed_time": "7:53:01", "remaining_time": "3:03:17"}
-{"current_steps": 1850, "total_steps": 2553, "loss": 0.4854, "lr": 5.338484173832413e-05, "epoch": 2.1738619676945667, "percentage": 72.46, "elapsed_time": "8:22:33", "remaining_time": "3:10:58"}
-{"current_steps": 1860, "total_steps": 2553, "loss": 0.4815, "lr": 5.197475258563249e-05, "epoch": 2.1856093979441997, "percentage": 72.86, "elapsed_time": "8:52:07", "remaining_time": "3:18:15"}
-{"current_steps": 1870, "total_steps": 2553, "loss": 0.4878, "lr": 5.057962370352815e-05, "epoch": 2.1973568281938327, "percentage": 73.25, "elapsed_time": "9:21:38", "remaining_time": "3:25:08"}
-{"current_steps": 1880, "total_steps": 2553, "loss": 0.4737, "lr": 4.91996680117041e-05, "epoch": 2.2091042584434657, "percentage": 73.64, "elapsed_time": "9:51:11", "remaining_time": "3:31:38"}
+{"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"}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e2db69a706d840c833279feb43cd380caf2ea3109923fe2a30592c8b14b96f0e
+oid sha256:c5002525d09503fe411fa35dafc4fa4f6d8fa9e642e00e997880eefc4a0c6823
 size 5752