Training in progress, step 16300
Browse files- adapter_config.json +5 -3
- adapter_model.safetensors +1 -1
- trainer_log.jsonl +2 -74
- training_args.bin +1 -1
adapter_config.json
CHANGED
|
@@ -3,6 +3,7 @@
|
|
| 3 |
"auto_mapping": null,
|
| 4 |
"base_model_name_or_path": "Qwen/Qwen3-0.6B",
|
| 5 |
"bias": "none",
|
|
|
|
| 6 |
"eva_config": null,
|
| 7 |
"exclude_modules": null,
|
| 8 |
"fan_in_fan_out": false,
|
|
@@ -23,15 +24,16 @@
|
|
| 23 |
"rank_pattern": {},
|
| 24 |
"revision": null,
|
| 25 |
"target_modules": [
|
|
|
|
| 26 |
"o_proj",
|
|
|
|
| 27 |
"down_proj",
|
| 28 |
-
"q_proj",
|
| 29 |
-
"k_proj",
|
| 30 |
"v_proj",
|
| 31 |
-
"
|
| 32 |
"up_proj"
|
| 33 |
],
|
| 34 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 35 |
"use_dora": false,
|
| 36 |
"use_rslora": false
|
| 37 |
}
|
|
|
|
| 3 |
"auto_mapping": null,
|
| 4 |
"base_model_name_or_path": "Qwen/Qwen3-0.6B",
|
| 5 |
"bias": "none",
|
| 6 |
+
"corda_config": null,
|
| 7 |
"eva_config": null,
|
| 8 |
"exclude_modules": null,
|
| 9 |
"fan_in_fan_out": false,
|
|
|
|
| 24 |
"rank_pattern": {},
|
| 25 |
"revision": null,
|
| 26 |
"target_modules": [
|
| 27 |
+
"q_proj",
|
| 28 |
"o_proj",
|
| 29 |
+
"gate_proj",
|
| 30 |
"down_proj",
|
|
|
|
|
|
|
| 31 |
"v_proj",
|
| 32 |
+
"k_proj",
|
| 33 |
"up_proj"
|
| 34 |
],
|
| 35 |
"task_type": "CAUSAL_LM",
|
| 36 |
+
"trainable_token_indices": null,
|
| 37 |
"use_dora": false,
|
| 38 |
"use_rslora": false
|
| 39 |
}
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 161533160
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:80380594ba85d0fe421fb6120a0f8116f295a982c0790b8effab2d188c06963d
|
| 3 |
size 161533160
|
trainer_log.jsonl
CHANGED
|
@@ -1,74 +1,2 @@
|
|
| 1 |
-
{"current_steps":
|
| 2 |
-
{"current_steps":
|
| 3 |
-
{"current_steps": 12650, "total_steps": 18120, "loss": 0.2893, "lr": 2.529252234174041e-05, "epoch": 2.0943708609271523, "percentage": 69.81, "elapsed_time": "0:09:37", "remaining_time": "0:04:09"}
|
| 4 |
-
{"current_steps": 12700, "total_steps": 18120, "loss": 0.2837, "lr": 2.4874980419378647e-05, "epoch": 2.1026490066225167, "percentage": 70.09, "elapsed_time": "0:12:52", "remaining_time": "0:05:29"}
|
| 5 |
-
{"current_steps": 12750, "total_steps": 18120, "loss": 0.2859, "lr": 2.445976949232676e-05, "epoch": 2.1109271523178808, "percentage": 70.36, "elapsed_time": "0:16:06", "remaining_time": "0:06:46"}
|
| 6 |
-
{"current_steps": 12800, "total_steps": 18120, "loss": 0.2841, "lr": 2.4046928082135733e-05, "epoch": 2.119205298013245, "percentage": 70.64, "elapsed_time": "0:19:17", "remaining_time": "0:08:01"}
|
| 7 |
-
{"current_steps": 12850, "total_steps": 18120, "loss": 0.2772, "lr": 2.3636494490522624e-05, "epoch": 2.127483443708609, "percentage": 70.92, "elapsed_time": "0:22:26", "remaining_time": "0:09:12"}
|
| 8 |
-
{"current_steps": 12900, "total_steps": 18120, "loss": 0.2829, "lr": 2.3228506795817072e-05, "epoch": 2.1357615894039736, "percentage": 71.19, "elapsed_time": "0:25:41", "remaining_time": "0:10:23"}
|
| 9 |
-
{"current_steps": 12950, "total_steps": 18120, "loss": 0.293, "lr": 2.282300284942846e-05, "epoch": 2.1440397350993377, "percentage": 71.47, "elapsed_time": "0:29:11", "remaining_time": "0:11:39"}
|
| 10 |
-
{"current_steps": 13000, "total_steps": 18120, "loss": 0.2946, "lr": 2.2420020272334337e-05, "epoch": 2.152317880794702, "percentage": 71.74, "elapsed_time": "0:32:26", "remaining_time": "0:12:46"}
|
| 11 |
-
{"current_steps": 13050, "total_steps": 18120, "loss": 0.2888, "lr": 2.2019596451590047e-05, "epoch": 2.160596026490066, "percentage": 72.02, "elapsed_time": "0:35:25", "remaining_time": "0:13:45"}
|
| 12 |
-
{"current_steps": 13100, "total_steps": 18120, "loss": 0.2921, "lr": 2.162176853686006e-05, "epoch": 2.1688741721854305, "percentage": 72.3, "elapsed_time": "0:38:33", "remaining_time": "0:14:46"}
|
| 13 |
-
{"current_steps": 13150, "total_steps": 18120, "loss": 0.3014, "lr": 2.1226573436971487e-05, "epoch": 2.1771523178807946, "percentage": 72.57, "elapsed_time": "0:41:45", "remaining_time": "0:15:47"}
|
| 14 |
-
{"current_steps": 13200, "total_steps": 18120, "loss": 0.2828, "lr": 2.0834047816489772e-05, "epoch": 2.185430463576159, "percentage": 72.85, "elapsed_time": "0:44:55", "remaining_time": "0:16:44"}
|
| 15 |
-
{"current_steps": 13250, "total_steps": 18120, "loss": 0.2946, "lr": 2.0444228092317057e-05, "epoch": 2.193708609271523, "percentage": 73.12, "elapsed_time": "0:48:04", "remaining_time": "0:17:40"}
|
| 16 |
-
{"current_steps": 13300, "total_steps": 18120, "loss": 0.2905, "lr": 2.005715043031369e-05, "epoch": 2.2019867549668874, "percentage": 73.4, "elapsed_time": "0:51:20", "remaining_time": "0:18:36"}
|
| 17 |
-
{"current_steps": 13350, "total_steps": 18120, "loss": 0.2847, "lr": 1.967285074194283e-05, "epoch": 2.2102649006622515, "percentage": 73.68, "elapsed_time": "0:54:32", "remaining_time": "0:19:29"}
|
| 18 |
-
{"current_steps": 13400, "total_steps": 18120, "loss": 0.2709, "lr": 1.9291364680938688e-05, "epoch": 2.218543046357616, "percentage": 73.95, "elapsed_time": "0:57:53", "remaining_time": "0:20:23"}
|
| 19 |
-
{"current_steps": 13450, "total_steps": 18120, "loss": 0.2754, "lr": 1.891272763999884e-05, "epoch": 2.22682119205298, "percentage": 74.23, "elapsed_time": "1:01:06", "remaining_time": "0:21:12"}
|
| 20 |
-
{"current_steps": 13500, "total_steps": 18120, "loss": 0.278, "lr": 1.8536974747500556e-05, "epoch": 2.2350993377483444, "percentage": 74.5, "elapsed_time": "1:04:22", "remaining_time": "0:22:01"}
|
| 21 |
-
{"current_steps": 13550, "total_steps": 18120, "loss": 0.2863, "lr": 1.8164140864241723e-05, "epoch": 2.243377483443709, "percentage": 74.78, "elapsed_time": "1:07:36", "remaining_time": "0:22:47"}
|
| 22 |
-
{"current_steps": 13600, "total_steps": 18120, "loss": 0.282, "lr": 1.7794260580206673e-05, "epoch": 2.251655629139073, "percentage": 75.06, "elapsed_time": "1:10:41", "remaining_time": "0:23:29"}
|
| 23 |
-
{"current_steps": 13650, "total_steps": 18120, "loss": 0.2756, "lr": 1.742736821135702e-05, "epoch": 2.2599337748344372, "percentage": 75.33, "elapsed_time": "1:13:52", "remaining_time": "0:24:11"}
|
| 24 |
-
{"current_steps": 13700, "total_steps": 18120, "loss": 0.2711, "lr": 1.7063497796447935e-05, "epoch": 2.2682119205298013, "percentage": 75.61, "elapsed_time": "1:17:00", "remaining_time": "0:24:50"}
|
| 25 |
-
{"current_steps": 13750, "total_steps": 18120, "loss": 0.2723, "lr": 1.670268309387029e-05, "epoch": 2.2764900662251657, "percentage": 75.88, "elapsed_time": "1:20:27", "remaining_time": "0:25:34"}
|
| 26 |
-
{"current_steps": 13800, "total_steps": 18120, "loss": 0.2821, "lr": 1.634495757851855e-05, "epoch": 2.2847682119205297, "percentage": 76.16, "elapsed_time": "1:23:36", "remaining_time": "0:26:10"}
|
| 27 |
-
{"current_steps": 13850, "total_steps": 18120, "loss": 0.2701, "lr": 1.599035443868518e-05, "epoch": 2.293046357615894, "percentage": 76.43, "elapsed_time": "1:26:48", "remaining_time": "0:26:45"}
|
| 28 |
-
{"current_steps": 13900, "total_steps": 18120, "loss": 0.2809, "lr": 1.5638906572981604e-05, "epoch": 2.301324503311258, "percentage": 76.71, "elapsed_time": "1:30:00", "remaining_time": "0:27:19"}
|
| 29 |
-
{"current_steps": 13950, "total_steps": 18120, "loss": 0.2847, "lr": 1.529064658728598e-05, "epoch": 2.3096026490066226, "percentage": 76.99, "elapsed_time": "1:33:16", "remaining_time": "0:27:52"}
|
| 30 |
-
{"current_steps": 14000, "total_steps": 18120, "loss": 0.2772, "lr": 1.4945606791718092e-05, "epoch": 2.3178807947019866, "percentage": 77.26, "elapsed_time": "1:36:40", "remaining_time": "0:28:26"}
|
| 31 |
-
{"current_steps": 14050, "total_steps": 18120, "loss": 0.2798, "lr": 1.4603819197641883e-05, "epoch": 2.326158940397351, "percentage": 77.54, "elapsed_time": "1:39:53", "remaining_time": "0:28:56"}
|
| 32 |
-
{"current_steps": 14100, "total_steps": 18120, "loss": 0.2712, "lr": 1.4265315514695488e-05, "epoch": 2.334437086092715, "percentage": 77.81, "elapsed_time": "1:43:08", "remaining_time": "0:29:24"}
|
| 33 |
-
{"current_steps": 14150, "total_steps": 18120, "loss": 0.2805, "lr": 1.3930127147849314e-05, "epoch": 2.3427152317880795, "percentage": 78.09, "elapsed_time": "1:46:32", "remaining_time": "0:29:53"}
|
| 34 |
-
{"current_steps": 14200, "total_steps": 18120, "loss": 0.2882, "lr": 1.3598285194492521e-05, "epoch": 2.3509933774834435, "percentage": 78.37, "elapsed_time": "1:49:38", "remaining_time": "0:30:16"}
|
| 35 |
-
{"current_steps": 14250, "total_steps": 18120, "loss": 0.2837, "lr": 1.326982044154787e-05, "epoch": 2.359271523178808, "percentage": 78.64, "elapsed_time": "1:53:14", "remaining_time": "0:30:45"}
|
| 36 |
-
{"current_steps": 14300, "total_steps": 18120, "loss": 0.2742, "lr": 1.2944763362615413e-05, "epoch": 2.3675496688741724, "percentage": 78.92, "elapsed_time": "1:56:20", "remaining_time": "0:31:04"}
|
| 37 |
-
{"current_steps": 14350, "total_steps": 18120, "loss": 0.2698, "lr": 1.2623144115145342e-05, "epoch": 2.3758278145695364, "percentage": 79.19, "elapsed_time": "1:59:27", "remaining_time": "0:31:23"}
|
| 38 |
-
{"current_steps": 14400, "total_steps": 18120, "loss": 0.2764, "lr": 1.2304992537640092e-05, "epoch": 2.384105960264901, "percentage": 79.47, "elapsed_time": "2:02:45", "remaining_time": "0:31:42"}
|
| 39 |
-
{"current_steps": 14450, "total_steps": 18120, "loss": 0.2729, "lr": 1.1990338146885977e-05, "epoch": 2.392384105960265, "percentage": 79.75, "elapsed_time": "2:05:53", "remaining_time": "0:31:58"}
|
| 40 |
-
{"current_steps": 14500, "total_steps": 18120, "loss": 0.2764, "lr": 1.1679210135214858e-05, "epoch": 2.4006622516556293, "percentage": 80.02, "elapsed_time": "2:09:03", "remaining_time": "0:32:13"}
|
| 41 |
-
{"current_steps": 14550, "total_steps": 18120, "loss": 0.2718, "lr": 1.1371637367795735e-05, "epoch": 2.4089403973509933, "percentage": 80.3, "elapsed_time": "2:12:09", "remaining_time": "0:32:25"}
|
| 42 |
-
{"current_steps": 14600, "total_steps": 18120, "loss": 0.288, "lr": 1.1067648379956714e-05, "epoch": 2.4172185430463577, "percentage": 80.57, "elapsed_time": "2:15:18", "remaining_time": "0:32:37"}
|
| 43 |
-
{"current_steps": 14650, "total_steps": 18120, "loss": 0.27, "lr": 1.0767271374537724e-05, "epoch": 2.4254966887417218, "percentage": 80.85, "elapsed_time": "2:18:50", "remaining_time": "0:32:53"}
|
| 44 |
-
{"current_steps": 14700, "total_steps": 18120, "loss": 0.2764, "lr": 1.0470534219273903e-05, "epoch": 2.433774834437086, "percentage": 81.13, "elapsed_time": "2:22:05", "remaining_time": "0:33:03"}
|
| 45 |
-
{"current_steps": 14750, "total_steps": 18120, "loss": 0.2685, "lr": 1.0177464444210133e-05, "epoch": 2.44205298013245, "percentage": 81.4, "elapsed_time": "2:25:14", "remaining_time": "0:33:10"}
|
| 46 |
-
{"current_steps": 14800, "total_steps": 18120, "loss": 0.2882, "lr": 9.888089239146963e-06, "epoch": 2.4503311258278146, "percentage": 81.68, "elapsed_time": "2:28:24", "remaining_time": "0:33:17"}
|
| 47 |
-
{"current_steps": 14850, "total_steps": 18120, "loss": 0.288, "lr": 9.602435451118047e-06, "epoch": 2.4586092715231787, "percentage": 81.95, "elapsed_time": "2:31:29", "remaining_time": "0:33:21"}
|
| 48 |
-
{"current_steps": 14900, "total_steps": 18120, "loss": 0.2797, "lr": 9.320529581899335e-06, "epoch": 2.466887417218543, "percentage": 82.23, "elapsed_time": "2:34:32", "remaining_time": "0:33:23"}
|
| 49 |
-
{"current_steps": 14950, "total_steps": 18120, "loss": 0.2812, "lr": 9.042397785550405e-06, "epoch": 2.475165562913907, "percentage": 82.51, "elapsed_time": "2:38:02", "remaining_time": "0:33:30"}
|
| 50 |
-
{"current_steps": 15000, "total_steps": 18120, "loss": 0.2719, "lr": 8.768065865987995e-06, "epoch": 2.4834437086092715, "percentage": 82.78, "elapsed_time": "2:41:26", "remaining_time": "0:33:34"}
|
| 51 |
-
{"current_steps": 15050, "total_steps": 18120, "loss": 0.2862, "lr": 8.49755927459196e-06, "epoch": 2.491721854304636, "percentage": 83.06, "elapsed_time": "2:44:37", "remaining_time": "0:33:34"}
|
| 52 |
-
{"current_steps": 15100, "total_steps": 18120, "loss": 0.2622, "lr": 8.230903107844078e-06, "epoch": 2.5, "percentage": 83.33, "elapsed_time": "2:47:54", "remaining_time": "0:33:34"}
|
| 53 |
-
{"current_steps": 15150, "total_steps": 18120, "loss": 0.2735, "lr": 7.968122104999676e-06, "epoch": 2.508278145695364, "percentage": 83.61, "elapsed_time": "2:51:01", "remaining_time": "0:33:31"}
|
| 54 |
-
{"current_steps": 15200, "total_steps": 18120, "loss": 0.2857, "lr": 7.70924064579236e-06, "epoch": 2.5165562913907285, "percentage": 83.89, "elapsed_time": "2:54:10", "remaining_time": "0:33:27"}
|
| 55 |
-
{"current_steps": 15250, "total_steps": 18120, "loss": 0.2647, "lr": 7.454282748172281e-06, "epoch": 2.524834437086093, "percentage": 84.16, "elapsed_time": "2:57:22", "remaining_time": "0:33:22"}
|
| 56 |
-
{"current_steps": 15300, "total_steps": 18120, "loss": 0.272, "lr": 7.2032720660777706e-06, "epoch": 2.533112582781457, "percentage": 84.44, "elapsed_time": "3:00:38", "remaining_time": "0:33:17"}
|
| 57 |
-
{"current_steps": 15350, "total_steps": 18120, "loss": 0.2815, "lr": 6.95623188724081e-06, "epoch": 2.5413907284768213, "percentage": 84.71, "elapsed_time": "3:03:40", "remaining_time": "0:33:08"}
|
| 58 |
-
{"current_steps": 15400, "total_steps": 18120, "loss": 0.2808, "lr": 6.713185131026567e-06, "epoch": 2.5496688741721854, "percentage": 84.99, "elapsed_time": "3:06:47", "remaining_time": "0:32:59"}
|
| 59 |
-
{"current_steps": 15450, "total_steps": 18120, "loss": 0.2719, "lr": 6.474154346306999e-06, "epoch": 2.55794701986755, "percentage": 85.26, "elapsed_time": "3:10:06", "remaining_time": "0:32:51"}
|
| 60 |
-
{"current_steps": 15500, "total_steps": 18120, "loss": 0.2743, "lr": 6.239161709368774e-06, "epoch": 2.566225165562914, "percentage": 85.54, "elapsed_time": "3:13:20", "remaining_time": "0:32:40"}
|
| 61 |
-
{"current_steps": 15550, "total_steps": 18120, "loss": 0.2817, "lr": 6.00822902185601e-06, "epoch": 2.5745033112582782, "percentage": 85.82, "elapsed_time": "3:16:31", "remaining_time": "0:32:28"}
|
| 62 |
-
{"current_steps": 15600, "total_steps": 18120, "loss": 0.2665, "lr": 5.781377708747493e-06, "epoch": 2.5827814569536423, "percentage": 86.09, "elapsed_time": "3:19:39", "remaining_time": "0:32:15"}
|
| 63 |
-
{"current_steps": 15650, "total_steps": 18120, "loss": 0.2767, "lr": 5.558628816368972e-06, "epoch": 2.5910596026490067, "percentage": 86.37, "elapsed_time": "3:22:45", "remaining_time": "0:32:00"}
|
| 64 |
-
{"current_steps": 15700, "total_steps": 18120, "loss": 0.2721, "lr": 5.340003010440603e-06, "epoch": 2.5993377483443707, "percentage": 86.64, "elapsed_time": "3:25:49", "remaining_time": "0:31:43"}
|
| 65 |
-
{"current_steps": 15750, "total_steps": 18120, "loss": 0.2797, "lr": 5.125520574159654e-06, "epoch": 2.607615894039735, "percentage": 86.92, "elapsed_time": "3:29:15", "remaining_time": "0:31:29"}
|
| 66 |
-
{"current_steps": 15800, "total_steps": 18120, "loss": 0.2669, "lr": 4.915201406318676e-06, "epoch": 2.6158940397350996, "percentage": 87.2, "elapsed_time": "3:32:30", "remaining_time": "0:31:12"}
|
| 67 |
-
{"current_steps": 15850, "total_steps": 18120, "loss": 0.2756, "lr": 4.7090650194594465e-06, "epoch": 2.6241721854304636, "percentage": 87.47, "elapsed_time": "3:35:57", "remaining_time": "0:30:55"}
|
| 68 |
-
{"current_steps": 15900, "total_steps": 18120, "loss": 0.2737, "lr": 4.50713053806262e-06, "epoch": 2.6324503311258276, "percentage": 87.75, "elapsed_time": "3:39:06", "remaining_time": "0:30:35"}
|
| 69 |
-
{"current_steps": 15950, "total_steps": 18120, "loss": 0.27, "lr": 4.309416696773455e-06, "epoch": 2.640728476821192, "percentage": 88.02, "elapsed_time": "3:42:23", "remaining_time": "0:30:15"}
|
| 70 |
-
{"current_steps": 16000, "total_steps": 18120, "loss": 0.2659, "lr": 4.1159418386636895e-06, "epoch": 2.6490066225165565, "percentage": 88.3, "elapsed_time": "3:45:45", "remaining_time": "0:29:54"}
|
| 71 |
-
{"current_steps": 16050, "total_steps": 18120, "loss": 0.2739, "lr": 3.926723913529773e-06, "epoch": 2.6572847682119205, "percentage": 88.58, "elapsed_time": "3:49:00", "remaining_time": "0:29:32"}
|
| 72 |
-
{"current_steps": 16100, "total_steps": 18120, "loss": 0.2801, "lr": 3.7417804762274968e-06, "epoch": 2.6655629139072845, "percentage": 88.85, "elapsed_time": "3:52:15", "remaining_time": "0:29:08"}
|
| 73 |
-
{"current_steps": 16150, "total_steps": 18120, "loss": 0.2646, "lr": 3.5611286850433967e-06, "epoch": 2.673841059602649, "percentage": 89.13, "elapsed_time": "3:55:37", "remaining_time": "0:28:44"}
|
| 74 |
-
{"current_steps": 16200, "total_steps": 18120, "loss": 0.2699, "lr": 3.3847853001028495e-06, "epoch": 2.6821192052980134, "percentage": 89.4, "elapsed_time": "3:58:52", "remaining_time": "0:28:18"}
|
|
|
|
| 1 |
+
{"current_steps": 16250, "total_steps": 18120, "loss": 0.2717, "lr": 3.2127666818151046e-06, "epoch": 2.6903973509933774, "percentage": 89.68, "elapsed_time": "0:05:28", "remaining_time": "0:00:37"}
|
| 2 |
+
{"current_steps": 16300, "total_steps": 18120, "loss": 0.277, "lr": 3.045088789355488e-06, "epoch": 2.698675496688742, "percentage": 89.96, "elapsed_time": "0:11:01", "remaining_time": "0:01:13"}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5752
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e7fd7045b4a15373c4d905f915edbb1436664635b44fe1df1bfa95de5188043
|
| 3 |
size 5752
|