Training in progress, step 2553

adapter_config.json

@@ -23,13 +23,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
+    "up_proj",
+    "v_proj",
     "k_proj",
-    "q_proj",
     "o_proj",
-    "gate_proj",
-    "v_proj",
+    "q_proj",
     "down_proj",
-    "up_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 sha256:5dc260198908eb99d3647d4fdbea5c9b5179e6bf7c27bff0996219bdf691627f
+oid sha256:c2863fab9f406d2423827a29fb4911c4a5eb67f96ba7c0890fb3308a4360f9b8
 size 147770496

trainer_log.jsonl

@@ -1,20 +1,4 @@
-{"current_steps": 2330, "total_steps": 2553, "loss": 0.4834, "lr": 5.7575176455622764e-06, "epoch": 2.7377386196769455, "percentage": 91.27, "elapsed_time": "0:31:33", "remaining_time": "0:03:01"}
-{"current_steps": 2340, "total_steps": 2553, "loss": 0.4812, "lr": 5.260057513552573e-06, "epoch": 2.7494860499265785, "percentage": 91.66, "elapsed_time": "1:03:23", "remaining_time": "0:05:46"}
-{"current_steps": 2350, "total_steps": 2553, "loss": 0.4782, "lr": 4.78468708558864e-06, "epoch": 2.7612334801762115, "percentage": 92.05, "elapsed_time": "1:35:10", "remaining_time": "0:08:13"}
-{"current_steps": 2360, "total_steps": 2553, "loss": 0.4849, "lr": 4.3314789110433675e-06, "epoch": 2.7729809104258445, "percentage": 92.44, "elapsed_time": "2:06:59", "remaining_time": "0:10:23"}
-{"current_steps": 2370, "total_steps": 2553, "loss": 0.477, "lr": 3.90050215696408e-06, "epoch": 2.7847283406754775, "percentage": 92.83, "elapsed_time": "2:38:47", "remaining_time": "0:12:15"}
-{"current_steps": 2380, "total_steps": 2553, "loss": 0.4788, "lr": 3.491822597516375e-06, "epoch": 2.79647577092511, "percentage": 93.22, "elapsed_time": "3:10:42", "remaining_time": "0:13:51"}
-{"current_steps": 2390, "total_steps": 2553, "loss": 0.482, "lr": 3.1055026039459863e-06, "epoch": 2.808223201174743, "percentage": 93.62, "elapsed_time": "3:42:33", "remaining_time": "0:15:10"}
-{"current_steps": 2400, "total_steps": 2553, "loss": 0.4799, "lr": 2.741601135059851e-06, "epoch": 2.819970631424376, "percentage": 94.01, "elapsed_time": "4:14:24", "remaining_time": "0:16:13"}
-{"current_steps": 2410, "total_steps": 2553, "loss": 0.4966, "lr": 2.4001737282280055e-06, "epoch": 2.831718061674009, "percentage": 94.4, "elapsed_time": "4:46:17", "remaining_time": "0:16:59"}
-{"current_steps": 2420, "total_steps": 2553, "loss": 0.4785, "lr": 2.081272490907765e-06, "epoch": 2.8434654919236415, "percentage": 94.79, "elapsed_time": "5:18:07", "remaining_time": "0:17:29"}
-{"current_steps": 2430, "total_steps": 2553, "loss": 0.4849, "lr": 1.784946092691153e-06, "epoch": 2.8552129221732745, "percentage": 95.18, "elapsed_time": "5:49:59", "remaining_time": "0:17:42"}
-{"current_steps": 2440, "total_steps": 2553, "loss": 0.4804, "lr": 1.5112397578771585e-06, "epoch": 2.8669603524229075, "percentage": 95.57, "elapsed_time": "6:21:48", "remaining_time": "0:17:40"}
-{"current_steps": 2450, "total_steps": 2553, "loss": 0.488, "lr": 1.2601952585698405e-06, "epoch": 2.8787077826725405, "percentage": 95.97, "elapsed_time": "6:53:40", "remaining_time": "0:17:23"}
-{"current_steps": 2460, "total_steps": 2553, "loss": 0.4713, "lr": 1.0318509083030447e-06, "epoch": 2.8904552129221734, "percentage": 96.36, "elapsed_time": "7:25:30", "remaining_time": "0:16:50"}
-{"current_steps": 2470, "total_steps": 2553, "loss": 0.4845, "lr": 8.2624155619328e-07, "epoch": 2.9022026431718064, "percentage": 96.75, "elapsed_time": "7:57:19", "remaining_time": "0:16:02"}
-{"current_steps": 2480, "total_steps": 2553, "loss": 0.4881, "lr": 6.43398581621124e-07, "epoch": 2.913950073421439, "percentage": 97.14, "elapsed_time": "8:29:09", "remaining_time": "0:14:59"}
-{"current_steps": 2490, "total_steps": 2553, "loss": 0.4851, "lr": 4.833498894421528e-07, "epoch": 2.925697503671072, "percentage": 97.53, "elapsed_time": "9:01:00", "remaining_time": "0:13:41"}
-{"current_steps": 2500, "total_steps": 2553, "loss": 0.4789, "lr": 3.4611990572829815e-07, "epoch": 2.937444933920705, "percentage": 97.92, "elapsed_time": "9:32:47", "remaining_time": "0:12:08"}
-{"current_steps": 2510, "total_steps": 2553, "loss": 0.4806, "lr": 2.317295740399294e-07, "epoch": 2.9491923641703375, "percentage": 98.32, "elapsed_time": "10:04:41", "remaining_time": "0:10:21"}
-{"current_steps": 2520, "total_steps": 2553, "loss": 0.4829, "lr": 1.4019635222961012e-07, "epoch": 2.9609397944199705, "percentage": 98.71, "elapsed_time": "10:36:29", "remaining_time": "0:08:20"}
+{"current_steps": 2530, "total_steps": 2553, "loss": 0.4732, "lr": 7.153420977766278e-08, "epoch": 2.9726872246696034, "percentage": 99.1, "elapsed_time": "0:31:46", "remaining_time": "0:00:17"}
+{"current_steps": 2540, "total_steps": 2553, "loss": 0.4826, "lr": 2.5753625660224208e-08, "epoch": 2.9844346549192364, "percentage": 99.49, "elapsed_time": "1:03:22", "remaining_time": "0:00:19"}
+{"current_steps": 2550, "total_steps": 2553, "loss": 0.4913, "lr": 2.8615867500414447e-09, "epoch": 2.9961820851688694, "percentage": 99.88, "elapsed_time": "1:34:54", "remaining_time": "0:00:06"}
+{"current_steps": 2553, "total_steps": 2553, "epoch": 2.999706314243759, "percentage": 100.0, "elapsed_time": "1:44:23", "remaining_time": "0:00:00"}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6adccd965aa8bdd0bbc146a114bcb82723fe68d3c66abfbb8609243c758a40f2
+oid sha256:dcdc753965d8ff82bf636220b660b6606d164e4abf268fe254d867087721e8a7
 size 5752