Upload 10 files
Browse filesThis is test model
- README.md +11 -0
- adapter_config.json +21 -0
- adapter_model.bin +3 -0
- all_results.json +7 -0
- finetuning_args.json +13 -0
- train_results.json +7 -0
- trainer_log.jsonl +101 -0
- trainer_state.json +625 -0
- training_args.bin +3 -0
- training_loss.png +0 -0
README.md
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
---
|
| 4 |
+
## Training procedure
|
| 5 |
+
|
| 6 |
+
### Framework versions
|
| 7 |
+
|
| 8 |
+
- PEFT 0.4.0
|
| 9 |
+
- PEFT 0.4.0
|
| 10 |
+
|
| 11 |
+
- PEFT 0.4.0
|
adapter_config.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"auto_mapping": null,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"fan_in_fan_out": false,
|
| 6 |
+
"inference_mode": true,
|
| 7 |
+
"init_lora_weights": true,
|
| 8 |
+
"layers_pattern": null,
|
| 9 |
+
"layers_to_transform": null,
|
| 10 |
+
"lora_alpha": 32.0,
|
| 11 |
+
"lora_dropout": 0.1,
|
| 12 |
+
"modules_to_save": null,
|
| 13 |
+
"peft_type": "LORA",
|
| 14 |
+
"r": 8,
|
| 15 |
+
"revision": null,
|
| 16 |
+
"target_modules": [
|
| 17 |
+
"q_proj",
|
| 18 |
+
"v_proj"
|
| 19 |
+
],
|
| 20 |
+
"task_type": "CAUSAL_LM"
|
| 21 |
+
}
|
adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3011ecb17b213294d67f30210407c359927233cddd803ac107acf4d754907f7f
|
| 3 |
+
size 16821197
|
all_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 666.67,
|
| 3 |
+
"train_loss": 0.06853693281114101,
|
| 4 |
+
"train_runtime": 2944.4055,
|
| 5 |
+
"train_samples_per_second": 3.396,
|
| 6 |
+
"train_steps_per_second": 0.34
|
| 7 |
+
}
|
finetuning_args.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"finetuning_type": "lora",
|
| 3 |
+
"lora_alpha": 32.0,
|
| 4 |
+
"lora_dropout": 0.1,
|
| 5 |
+
"lora_rank": 8,
|
| 6 |
+
"lora_target": [
|
| 7 |
+
"q_proj",
|
| 8 |
+
"v_proj"
|
| 9 |
+
],
|
| 10 |
+
"name_module_trainable": "mlp",
|
| 11 |
+
"num_hidden_layers": 32,
|
| 12 |
+
"num_layer_trainable": 3
|
| 13 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 666.67,
|
| 3 |
+
"train_loss": 0.06853693281114101,
|
| 4 |
+
"train_runtime": 2944.4055,
|
| 5 |
+
"train_samples_per_second": 3.396,
|
| 6 |
+
"train_steps_per_second": 0.34
|
| 7 |
+
}
|
trainer_log.jsonl
ADDED
|
@@ -0,0 +1,101 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"current_steps": 10, "total_steps": 1000, "loss": 1.0516, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.998766400914329e-05, "epoch": 6.67, "percentage": 1.0, "elapsed_time": "0:00:30", "remaining_time": "0:49:40"}
|
| 2 |
+
{"current_steps": 20, "total_steps": 1000, "loss": 1.0167, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.995066821070679e-05, "epoch": 13.33, "percentage": 2.0, "elapsed_time": "0:00:59", "remaining_time": "0:48:17"}
|
| 3 |
+
{"current_steps": 30, "total_steps": 1000, "loss": 0.9676, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9889049115077005e-05, "epoch": 20.0, "percentage": 3.0, "elapsed_time": "0:01:28", "remaining_time": "0:47:33"}
|
| 4 |
+
{"current_steps": 40, "total_steps": 1000, "loss": 0.888, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.980286753286195e-05, "epoch": 26.67, "percentage": 4.0, "elapsed_time": "0:01:58", "remaining_time": "0:47:16"}
|
| 5 |
+
{"current_steps": 50, "total_steps": 1000, "loss": 0.798, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9692208514878444e-05, "epoch": 33.33, "percentage": 5.0, "elapsed_time": "0:02:27", "remaining_time": "0:46:41"}
|
| 6 |
+
{"current_steps": 60, "total_steps": 1000, "loss": 0.6272, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9557181268217227e-05, "epoch": 40.0, "percentage": 6.0, "elapsed_time": "0:02:56", "remaining_time": "0:46:07"}
|
| 7 |
+
{"current_steps": 70, "total_steps": 1000, "loss": 0.5097, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.939791904846869e-05, "epoch": 46.67, "percentage": 7.0, "elapsed_time": "0:03:26", "remaining_time": "0:45:45"}
|
| 8 |
+
{"current_steps": 80, "total_steps": 1000, "loss": 0.3408, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9214579028215776e-05, "epoch": 53.33, "percentage": 8.0, "elapsed_time": "0:03:55", "remaining_time": "0:45:12"}
|
| 9 |
+
{"current_steps": 90, "total_steps": 1000, "loss": 0.1979, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.900734214192358e-05, "epoch": 60.0, "percentage": 9.0, "elapsed_time": "0:04:25", "remaining_time": "0:44:40"}
|
| 10 |
+
{"current_steps": 100, "total_steps": 1000, "loss": 0.109, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.877641290737884e-05, "epoch": 66.67, "percentage": 10.0, "elapsed_time": "0:04:55", "remaining_time": "0:44:16"}
|
| 11 |
+
{"current_steps": 110, "total_steps": 1000, "loss": 0.0503, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.852201922385564e-05, "epoch": 73.33, "percentage": 11.0, "elapsed_time": "0:05:24", "remaining_time": "0:43:44"}
|
| 12 |
+
{"current_steps": 120, "total_steps": 1000, "loss": 0.0281, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.8244412147206284e-05, "epoch": 80.0, "percentage": 12.0, "elapsed_time": "0:05:53", "remaining_time": "0:43:13"}
|
| 13 |
+
{"current_steps": 130, "total_steps": 1000, "loss": 0.0152, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.794386564209953e-05, "epoch": 86.67, "percentage": 13.0, "elapsed_time": "0:06:23", "remaining_time": "0:42:47"}
|
| 14 |
+
{"current_steps": 140, "total_steps": 1000, "loss": 0.0102, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.762067631165049e-05, "epoch": 93.33, "percentage": 14.0, "elapsed_time": "0:06:52", "remaining_time": "0:42:16"}
|
| 15 |
+
{"current_steps": 150, "total_steps": 1000, "loss": 0.0075, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.72751631047092e-05, "epoch": 100.0, "percentage": 15.0, "elapsed_time": "0:07:22", "remaining_time": "0:41:45"}
|
| 16 |
+
{"current_steps": 160, "total_steps": 1000, "loss": 0.0063, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.690766700109659e-05, "epoch": 106.67, "percentage": 16.0, "elapsed_time": "0:07:52", "remaining_time": "0:41:18"}
|
| 17 |
+
{"current_steps": 170, "total_steps": 1000, "loss": 0.0053, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.65185506750986e-05, "epoch": 113.33, "percentage": 17.0, "elapsed_time": "0:08:21", "remaining_time": "0:40:47"}
|
| 18 |
+
{"current_steps": 180, "total_steps": 1000, "loss": 0.0049, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.610819813755038e-05, "epoch": 120.0, "percentage": 18.0, "elapsed_time": "0:08:50", "remaining_time": "0:40:16"}
|
| 19 |
+
{"current_steps": 190, "total_steps": 1000, "loss": 0.0045, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.567701435686404e-05, "epoch": 126.67, "percentage": 19.0, "elapsed_time": "0:09:20", "remaining_time": "0:39:49"}
|
| 20 |
+
{"current_steps": 200, "total_steps": 1000, "loss": 0.0041, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.522542485937369e-05, "epoch": 133.33, "percentage": 20.0, "elapsed_time": "0:09:49", "remaining_time": "0:39:18"}
|
| 21 |
+
{"current_steps": 210, "total_steps": 1000, "loss": 0.0038, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.4753875309392266e-05, "epoch": 140.0, "percentage": 21.0, "elapsed_time": "0:10:18", "remaining_time": "0:38:48"}
|
| 22 |
+
{"current_steps": 220, "total_steps": 1000, "loss": 0.0036, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.426283106939474e-05, "epoch": 146.67, "percentage": 22.0, "elapsed_time": "0:10:48", "remaining_time": "0:38:20"}
|
| 23 |
+
{"current_steps": 230, "total_steps": 1000, "loss": 0.0035, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.375277674076149e-05, "epoch": 153.33, "percentage": 23.0, "elapsed_time": "0:11:18", "remaining_time": "0:37:50"}
|
| 24 |
+
{"current_steps": 240, "total_steps": 1000, "loss": 0.0033, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.3224215685535294e-05, "epoch": 160.0, "percentage": 24.0, "elapsed_time": "0:11:47", "remaining_time": "0:37:20"}
|
| 25 |
+
{"current_steps": 250, "total_steps": 1000, "loss": 0.0032, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.267766952966369e-05, "epoch": 166.67, "percentage": 25.0, "elapsed_time": "0:12:17", "remaining_time": "0:36:52"}
|
| 26 |
+
{"current_steps": 260, "total_steps": 1000, "loss": 0.0032, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.211367764821722e-05, "epoch": 173.33, "percentage": 26.0, "elapsed_time": "0:12:46", "remaining_time": "0:36:21"}
|
| 27 |
+
{"current_steps": 270, "total_steps": 1000, "loss": 0.0031, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.1532796633091296e-05, "epoch": 180.0, "percentage": 27.0, "elapsed_time": "0:13:15", "remaining_time": "0:35:51"}
|
| 28 |
+
{"current_steps": 280, "total_steps": 1000, "loss": 0.003, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.093559974371725e-05, "epoch": 186.67, "percentage": 28.0, "elapsed_time": "0:13:45", "remaining_time": "0:35:23"}
|
| 29 |
+
{"current_steps": 290, "total_steps": 1000, "loss": 0.0029, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.0322676341324415e-05, "epoch": 193.33, "percentage": 29.0, "elapsed_time": "0:14:15", "remaining_time": "0:34:53"}
|
| 30 |
+
{"current_steps": 300, "total_steps": 1000, "loss": 0.0029, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.969463130731183e-05, "epoch": 200.0, "percentage": 30.0, "elapsed_time": "0:14:44", "remaining_time": "0:34:23"}
|
| 31 |
+
{"current_steps": 310, "total_steps": 1000, "loss": 0.0029, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.905208444630327e-05, "epoch": 206.67, "percentage": 31.0, "elapsed_time": "0:15:14", "remaining_time": "0:33:54"}
|
| 32 |
+
{"current_steps": 320, "total_steps": 1000, "loss": 0.0028, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.8395669874474915e-05, "epoch": 213.33, "percentage": 32.0, "elapsed_time": "0:15:43", "remaining_time": "0:33:24"}
|
| 33 |
+
{"current_steps": 330, "total_steps": 1000, "loss": 0.0028, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.7726035393759285e-05, "epoch": 220.0, "percentage": 33.0, "elapsed_time": "0:16:12", "remaining_time": "0:32:54"}
|
| 34 |
+
{"current_steps": 340, "total_steps": 1000, "loss": 0.0027, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.704384185254288e-05, "epoch": 226.67, "percentage": 34.0, "elapsed_time": "0:16:42", "remaining_time": "0:32:26"}
|
| 35 |
+
{"current_steps": 350, "total_steps": 1000, "loss": 0.0027, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.634976249348867e-05, "epoch": 233.33, "percentage": 35.0, "elapsed_time": "0:17:11", "remaining_time": "0:31:56"}
|
| 36 |
+
{"current_steps": 360, "total_steps": 1000, "loss": 0.0027, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.564448228912682e-05, "epoch": 240.0, "percentage": 36.0, "elapsed_time": "0:17:40", "remaining_time": "0:31:26"}
|
| 37 |
+
{"current_steps": 370, "total_steps": 1000, "loss": 0.0027, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.4928697265869515e-05, "epoch": 246.67, "percentage": 37.0, "elapsed_time": "0:18:10", "remaining_time": "0:30:57"}
|
| 38 |
+
{"current_steps": 380, "total_steps": 1000, "loss": 0.0028, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.4203113817116957e-05, "epoch": 253.33, "percentage": 38.0, "elapsed_time": "0:18:40", "remaining_time": "0:30:27"}
|
| 39 |
+
{"current_steps": 390, "total_steps": 1000, "loss": 0.0027, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.346844800613229e-05, "epoch": 260.0, "percentage": 39.0, "elapsed_time": "0:19:09", "remaining_time": "0:29:57"}
|
| 40 |
+
{"current_steps": 400, "total_steps": 1000, "loss": 0.0027, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.272542485937369e-05, "epoch": 266.67, "percentage": 40.0, "elapsed_time": "0:19:39", "remaining_time": "0:29:29"}
|
| 41 |
+
{"current_steps": 410, "total_steps": 1000, "loss": 0.0027, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.1974777650980735e-05, "epoch": 273.33, "percentage": 41.0, "elapsed_time": "0:20:08", "remaining_time": "0:28:59"}
|
| 42 |
+
{"current_steps": 420, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.121724717912138e-05, "epoch": 280.0, "percentage": 42.0, "elapsed_time": "0:20:37", "remaining_time": "0:28:29"}
|
| 43 |
+
{"current_steps": 430, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.045358103491357e-05, "epoch": 286.67, "percentage": 43.0, "elapsed_time": "0:21:07", "remaining_time": "0:28:00"}
|
| 44 |
+
{"current_steps": 440, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.9684532864643122e-05, "epoch": 293.33, "percentage": 44.0, "elapsed_time": "0:21:36", "remaining_time": "0:27:30"}
|
| 45 |
+
{"current_steps": 450, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.8910861626005776e-05, "epoch": 300.0, "percentage": 45.0, "elapsed_time": "0:22:06", "remaining_time": "0:27:00"}
|
| 46 |
+
{"current_steps": 460, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.8133330839107608e-05, "epoch": 306.67, "percentage": 46.0, "elapsed_time": "0:22:36", "remaining_time": "0:26:31"}
|
| 47 |
+
{"current_steps": 470, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.7352707832962865e-05, "epoch": 313.33, "percentage": 47.0, "elapsed_time": "0:23:05", "remaining_time": "0:26:02"}
|
| 48 |
+
{"current_steps": 480, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.656976298823284e-05, "epoch": 320.0, "percentage": 48.0, "elapsed_time": "0:23:34", "remaining_time": "0:25:32"}
|
| 49 |
+
{"current_steps": 490, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.578526897695321e-05, "epoch": 326.67, "percentage": 49.0, "elapsed_time": "0:24:04", "remaining_time": "0:25:03"}
|
| 50 |
+
{"current_steps": 500, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.5e-05, "epoch": 333.33, "percentage": 50.0, "elapsed_time": "0:24:33", "remaining_time": "0:24:33"}
|
| 51 |
+
{"current_steps": 510, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.4214731023046793e-05, "epoch": 340.0, "percentage": 51.0, "elapsed_time": "0:25:02", "remaining_time": "0:24:03"}
|
| 52 |
+
{"current_steps": 520, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.3430237011767167e-05, "epoch": 346.67, "percentage": 52.0, "elapsed_time": "0:25:32", "remaining_time": "0:23:34"}
|
| 53 |
+
{"current_steps": 530, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.2647292167037144e-05, "epoch": 353.33, "percentage": 53.0, "elapsed_time": "0:26:01", "remaining_time": "0:23:04"}
|
| 54 |
+
{"current_steps": 540, "total_steps": 1000, "loss": 0.0026, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.186666916089239e-05, "epoch": 360.0, "percentage": 54.0, "elapsed_time": "0:26:30", "remaining_time": "0:22:35"}
|
| 55 |
+
{"current_steps": 550, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.1089138373994223e-05, "epoch": 366.67, "percentage": 55.0, "elapsed_time": "0:27:00", "remaining_time": "0:22:06"}
|
| 56 |
+
{"current_steps": 560, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.031546713535688e-05, "epoch": 373.33, "percentage": 56.0, "elapsed_time": "0:27:30", "remaining_time": "0:21:36"}
|
| 57 |
+
{"current_steps": 570, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.9546418965086442e-05, "epoch": 380.0, "percentage": 57.0, "elapsed_time": "0:27:59", "remaining_time": "0:21:06"}
|
| 58 |
+
{"current_steps": 580, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.8782752820878634e-05, "epoch": 386.67, "percentage": 58.0, "elapsed_time": "0:28:29", "remaining_time": "0:20:37"}
|
| 59 |
+
{"current_steps": 590, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.802522234901927e-05, "epoch": 393.33, "percentage": 59.0, "elapsed_time": "0:28:58", "remaining_time": "0:20:08"}
|
| 60 |
+
{"current_steps": 600, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.7274575140626318e-05, "epoch": 400.0, "percentage": 60.0, "elapsed_time": "0:29:27", "remaining_time": "0:19:38"}
|
| 61 |
+
{"current_steps": 610, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.6531551993867717e-05, "epoch": 406.67, "percentage": 61.0, "elapsed_time": "0:29:57", "remaining_time": "0:19:09"}
|
| 62 |
+
{"current_steps": 620, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.5796886182883053e-05, "epoch": 413.33, "percentage": 62.0, "elapsed_time": "0:30:26", "remaining_time": "0:18:39"}
|
| 63 |
+
{"current_steps": 630, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.5071302734130489e-05, "epoch": 420.0, "percentage": 63.0, "elapsed_time": "0:30:55", "remaining_time": "0:18:09"}
|
| 64 |
+
{"current_steps": 640, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.4355517710873184e-05, "epoch": 426.67, "percentage": 64.0, "elapsed_time": "0:31:25", "remaining_time": "0:17:40"}
|
| 65 |
+
{"current_steps": 650, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.3650237506511331e-05, "epoch": 433.33, "percentage": 65.0, "elapsed_time": "0:31:54", "remaining_time": "0:17:11"}
|
| 66 |
+
{"current_steps": 660, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.2956158147457115e-05, "epoch": 440.0, "percentage": 66.0, "elapsed_time": "0:32:24", "remaining_time": "0:16:41"}
|
| 67 |
+
{"current_steps": 670, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.2273964606240718e-05, "epoch": 446.67, "percentage": 67.0, "elapsed_time": "0:32:53", "remaining_time": "0:16:12"}
|
| 68 |
+
{"current_steps": 680, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.1604330125525079e-05, "epoch": 453.33, "percentage": 68.0, "elapsed_time": "0:33:23", "remaining_time": "0:15:42"}
|
| 69 |
+
{"current_steps": 690, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.0947915553696742e-05, "epoch": 460.0, "percentage": 69.0, "elapsed_time": "0:33:52", "remaining_time": "0:15:13"}
|
| 70 |
+
{"current_steps": 700, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.0305368692688174e-05, "epoch": 466.67, "percentage": 70.0, "elapsed_time": "0:34:22", "remaining_time": "0:14:43"}
|
| 71 |
+
{"current_steps": 710, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 9.677323658675594e-06, "epoch": 473.33, "percentage": 71.0, "elapsed_time": "0:34:51", "remaining_time": "0:14:14"}
|
| 72 |
+
{"current_steps": 720, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 9.064400256282757e-06, "epoch": 480.0, "percentage": 72.0, "elapsed_time": "0:35:20", "remaining_time": "0:13:44"}
|
| 73 |
+
{"current_steps": 730, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 8.467203366908707e-06, "epoch": 486.67, "percentage": 73.0, "elapsed_time": "0:35:50", "remaining_time": "0:13:15"}
|
| 74 |
+
{"current_steps": 740, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 7.886322351782783e-06, "epoch": 493.33, "percentage": 74.0, "elapsed_time": "0:36:19", "remaining_time": "0:12:45"}
|
| 75 |
+
{"current_steps": 750, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 7.3223304703363135e-06, "epoch": 500.0, "percentage": 75.0, "elapsed_time": "0:36:48", "remaining_time": "0:12:16"}
|
| 76 |
+
{"current_steps": 760, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 6.775784314464717e-06, "epoch": 506.67, "percentage": 76.0, "elapsed_time": "0:37:18", "remaining_time": "0:11:46"}
|
| 77 |
+
{"current_steps": 770, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 6.247223259238511e-06, "epoch": 513.33, "percentage": 77.0, "elapsed_time": "0:37:47", "remaining_time": "0:11:17"}
|
| 78 |
+
{"current_steps": 780, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 5.737168930605272e-06, "epoch": 520.0, "percentage": 78.0, "elapsed_time": "0:38:16", "remaining_time": "0:10:47"}
|
| 79 |
+
{"current_steps": 790, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 5.24612469060774e-06, "epoch": 526.67, "percentage": 79.0, "elapsed_time": "0:38:46", "remaining_time": "0:10:18"}
|
| 80 |
+
{"current_steps": 800, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.7745751406263165e-06, "epoch": 533.33, "percentage": 80.0, "elapsed_time": "0:39:16", "remaining_time": "0:09:49"}
|
| 81 |
+
{"current_steps": 810, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.322985643135952e-06, "epoch": 540.0, "percentage": 81.0, "elapsed_time": "0:39:45", "remaining_time": "0:09:19"}
|
| 82 |
+
{"current_steps": 820, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.891801862449629e-06, "epoch": 546.67, "percentage": 82.0, "elapsed_time": "0:40:15", "remaining_time": "0:08:50"}
|
| 83 |
+
{"current_steps": 830, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.4814493249014116e-06, "epoch": 553.33, "percentage": 83.0, "elapsed_time": "0:40:44", "remaining_time": "0:08:20"}
|
| 84 |
+
{"current_steps": 840, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.092332998903416e-06, "epoch": 560.0, "percentage": 84.0, "elapsed_time": "0:41:13", "remaining_time": "0:07:51"}
|
| 85 |
+
{"current_steps": 850, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.7248368952908053e-06, "epoch": 566.67, "percentage": 85.0, "elapsed_time": "0:41:43", "remaining_time": "0:07:21"}
|
| 86 |
+
{"current_steps": 860, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.379323688349516e-06, "epoch": 573.33, "percentage": 86.0, "elapsed_time": "0:42:12", "remaining_time": "0:06:52"}
|
| 87 |
+
{"current_steps": 870, "total_steps": 1000, "loss": 0.0024, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.0561343579004715e-06, "epoch": 580.0, "percentage": 87.0, "elapsed_time": "0:42:41", "remaining_time": "0:06:22"}
|
| 88 |
+
{"current_steps": 880, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.7555878527937164e-06, "epoch": 586.67, "percentage": 88.0, "elapsed_time": "0:43:11", "remaining_time": "0:05:53"}
|
| 89 |
+
{"current_steps": 890, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.4779807761443636e-06, "epoch": 593.33, "percentage": 89.0, "elapsed_time": "0:43:40", "remaining_time": "0:05:23"}
|
| 90 |
+
{"current_steps": 900, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.2235870926211619e-06, "epoch": 600.0, "percentage": 90.0, "elapsed_time": "0:44:09", "remaining_time": "0:04:54"}
|
| 91 |
+
{"current_steps": 910, "total_steps": 1000, "loss": 0.0024, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 9.926578580764234e-07, "epoch": 606.67, "percentage": 91.0, "elapsed_time": "0:44:39", "remaining_time": "0:04:25"}
|
| 92 |
+
{"current_steps": 920, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 7.854209717842231e-07, "epoch": 613.33, "percentage": 92.0, "elapsed_time": "0:45:08", "remaining_time": "0:03:55"}
|
| 93 |
+
{"current_steps": 930, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 6.020809515313142e-07, "epoch": 620.0, "percentage": 93.0, "elapsed_time": "0:45:37", "remaining_time": "0:03:26"}
|
| 94 |
+
{"current_steps": 940, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.4281873178278475e-07, "epoch": 626.67, "percentage": 94.0, "elapsed_time": "0:46:07", "remaining_time": "0:02:56"}
|
| 95 |
+
{"current_steps": 950, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.077914851215585e-07, "epoch": 633.33, "percentage": 95.0, "elapsed_time": "0:46:37", "remaining_time": "0:02:27"}
|
| 96 |
+
{"current_steps": 960, "total_steps": 1000, "loss": 0.0024, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.9713246713805588e-07, "epoch": 640.0, "percentage": 96.0, "elapsed_time": "0:47:06", "remaining_time": "0:01:57"}
|
| 97 |
+
{"current_steps": 970, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.109508849230001e-07, "epoch": 646.67, "percentage": 97.0, "elapsed_time": "0:47:36", "remaining_time": "0:01:28"}
|
| 98 |
+
{"current_steps": 980, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9331789293211026e-08, "epoch": 653.33, "percentage": 98.0, "elapsed_time": "0:48:05", "remaining_time": "0:00:58"}
|
| 99 |
+
{"current_steps": 990, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.233599085671e-08, "epoch": 660.0, "percentage": 99.0, "elapsed_time": "0:48:34", "remaining_time": "0:00:29"}
|
| 100 |
+
{"current_steps": 1000, "total_steps": 1000, "loss": 0.0025, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 0.0, "epoch": 666.67, "percentage": 100.0, "elapsed_time": "0:49:04", "remaining_time": "0:00:00"}
|
| 101 |
+
{"current_steps": 1000, "total_steps": 1000, "loss": null, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": null, "epoch": 666.67, "percentage": 100.0, "elapsed_time": "0:49:04", "remaining_time": "0:00:00"}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,625 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 666.6666666666666,
|
| 5 |
+
"global_step": 1000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 6.67,
|
| 12 |
+
"learning_rate": 4.998766400914329e-05,
|
| 13 |
+
"loss": 1.0516,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 13.33,
|
| 18 |
+
"learning_rate": 4.995066821070679e-05,
|
| 19 |
+
"loss": 1.0167,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 20.0,
|
| 24 |
+
"learning_rate": 4.9889049115077005e-05,
|
| 25 |
+
"loss": 0.9676,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 26.67,
|
| 30 |
+
"learning_rate": 4.980286753286195e-05,
|
| 31 |
+
"loss": 0.888,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 33.33,
|
| 36 |
+
"learning_rate": 4.9692208514878444e-05,
|
| 37 |
+
"loss": 0.798,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 40.0,
|
| 42 |
+
"learning_rate": 4.9557181268217227e-05,
|
| 43 |
+
"loss": 0.6272,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 46.67,
|
| 48 |
+
"learning_rate": 4.939791904846869e-05,
|
| 49 |
+
"loss": 0.5097,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 53.33,
|
| 54 |
+
"learning_rate": 4.9214579028215776e-05,
|
| 55 |
+
"loss": 0.3408,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 60.0,
|
| 60 |
+
"learning_rate": 4.900734214192358e-05,
|
| 61 |
+
"loss": 0.1979,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 66.67,
|
| 66 |
+
"learning_rate": 4.877641290737884e-05,
|
| 67 |
+
"loss": 0.109,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 73.33,
|
| 72 |
+
"learning_rate": 4.852201922385564e-05,
|
| 73 |
+
"loss": 0.0503,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 80.0,
|
| 78 |
+
"learning_rate": 4.8244412147206284e-05,
|
| 79 |
+
"loss": 0.0281,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 86.67,
|
| 84 |
+
"learning_rate": 4.794386564209953e-05,
|
| 85 |
+
"loss": 0.0152,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 93.33,
|
| 90 |
+
"learning_rate": 4.762067631165049e-05,
|
| 91 |
+
"loss": 0.0102,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 100.0,
|
| 96 |
+
"learning_rate": 4.72751631047092e-05,
|
| 97 |
+
"loss": 0.0075,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 106.67,
|
| 102 |
+
"learning_rate": 4.690766700109659e-05,
|
| 103 |
+
"loss": 0.0063,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 113.33,
|
| 108 |
+
"learning_rate": 4.65185506750986e-05,
|
| 109 |
+
"loss": 0.0053,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 120.0,
|
| 114 |
+
"learning_rate": 4.610819813755038e-05,
|
| 115 |
+
"loss": 0.0049,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 126.67,
|
| 120 |
+
"learning_rate": 4.567701435686404e-05,
|
| 121 |
+
"loss": 0.0045,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 133.33,
|
| 126 |
+
"learning_rate": 4.522542485937369e-05,
|
| 127 |
+
"loss": 0.0041,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 140.0,
|
| 132 |
+
"learning_rate": 4.4753875309392266e-05,
|
| 133 |
+
"loss": 0.0038,
|
| 134 |
+
"step": 210
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 146.67,
|
| 138 |
+
"learning_rate": 4.426283106939474e-05,
|
| 139 |
+
"loss": 0.0036,
|
| 140 |
+
"step": 220
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 153.33,
|
| 144 |
+
"learning_rate": 4.375277674076149e-05,
|
| 145 |
+
"loss": 0.0035,
|
| 146 |
+
"step": 230
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 160.0,
|
| 150 |
+
"learning_rate": 4.3224215685535294e-05,
|
| 151 |
+
"loss": 0.0033,
|
| 152 |
+
"step": 240
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 166.67,
|
| 156 |
+
"learning_rate": 4.267766952966369e-05,
|
| 157 |
+
"loss": 0.0032,
|
| 158 |
+
"step": 250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 173.33,
|
| 162 |
+
"learning_rate": 4.211367764821722e-05,
|
| 163 |
+
"loss": 0.0032,
|
| 164 |
+
"step": 260
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 180.0,
|
| 168 |
+
"learning_rate": 4.1532796633091296e-05,
|
| 169 |
+
"loss": 0.0031,
|
| 170 |
+
"step": 270
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 186.67,
|
| 174 |
+
"learning_rate": 4.093559974371725e-05,
|
| 175 |
+
"loss": 0.003,
|
| 176 |
+
"step": 280
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 193.33,
|
| 180 |
+
"learning_rate": 4.0322676341324415e-05,
|
| 181 |
+
"loss": 0.0029,
|
| 182 |
+
"step": 290
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 200.0,
|
| 186 |
+
"learning_rate": 3.969463130731183e-05,
|
| 187 |
+
"loss": 0.0029,
|
| 188 |
+
"step": 300
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 206.67,
|
| 192 |
+
"learning_rate": 3.905208444630327e-05,
|
| 193 |
+
"loss": 0.0029,
|
| 194 |
+
"step": 310
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 213.33,
|
| 198 |
+
"learning_rate": 3.8395669874474915e-05,
|
| 199 |
+
"loss": 0.0028,
|
| 200 |
+
"step": 320
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 220.0,
|
| 204 |
+
"learning_rate": 3.7726035393759285e-05,
|
| 205 |
+
"loss": 0.0028,
|
| 206 |
+
"step": 330
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 226.67,
|
| 210 |
+
"learning_rate": 3.704384185254288e-05,
|
| 211 |
+
"loss": 0.0027,
|
| 212 |
+
"step": 340
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 233.33,
|
| 216 |
+
"learning_rate": 3.634976249348867e-05,
|
| 217 |
+
"loss": 0.0027,
|
| 218 |
+
"step": 350
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 240.0,
|
| 222 |
+
"learning_rate": 3.564448228912682e-05,
|
| 223 |
+
"loss": 0.0027,
|
| 224 |
+
"step": 360
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 246.67,
|
| 228 |
+
"learning_rate": 3.4928697265869515e-05,
|
| 229 |
+
"loss": 0.0027,
|
| 230 |
+
"step": 370
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 253.33,
|
| 234 |
+
"learning_rate": 3.4203113817116957e-05,
|
| 235 |
+
"loss": 0.0028,
|
| 236 |
+
"step": 380
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 260.0,
|
| 240 |
+
"learning_rate": 3.346844800613229e-05,
|
| 241 |
+
"loss": 0.0027,
|
| 242 |
+
"step": 390
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 266.67,
|
| 246 |
+
"learning_rate": 3.272542485937369e-05,
|
| 247 |
+
"loss": 0.0027,
|
| 248 |
+
"step": 400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 273.33,
|
| 252 |
+
"learning_rate": 3.1974777650980735e-05,
|
| 253 |
+
"loss": 0.0027,
|
| 254 |
+
"step": 410
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 280.0,
|
| 258 |
+
"learning_rate": 3.121724717912138e-05,
|
| 259 |
+
"loss": 0.0026,
|
| 260 |
+
"step": 420
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 286.67,
|
| 264 |
+
"learning_rate": 3.045358103491357e-05,
|
| 265 |
+
"loss": 0.0026,
|
| 266 |
+
"step": 430
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 293.33,
|
| 270 |
+
"learning_rate": 2.9684532864643122e-05,
|
| 271 |
+
"loss": 0.0026,
|
| 272 |
+
"step": 440
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 300.0,
|
| 276 |
+
"learning_rate": 2.8910861626005776e-05,
|
| 277 |
+
"loss": 0.0026,
|
| 278 |
+
"step": 450
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 306.67,
|
| 282 |
+
"learning_rate": 2.8133330839107608e-05,
|
| 283 |
+
"loss": 0.0026,
|
| 284 |
+
"step": 460
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 313.33,
|
| 288 |
+
"learning_rate": 2.7352707832962865e-05,
|
| 289 |
+
"loss": 0.0026,
|
| 290 |
+
"step": 470
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 320.0,
|
| 294 |
+
"learning_rate": 2.656976298823284e-05,
|
| 295 |
+
"loss": 0.0026,
|
| 296 |
+
"step": 480
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 326.67,
|
| 300 |
+
"learning_rate": 2.578526897695321e-05,
|
| 301 |
+
"loss": 0.0026,
|
| 302 |
+
"step": 490
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 333.33,
|
| 306 |
+
"learning_rate": 2.5e-05,
|
| 307 |
+
"loss": 0.0026,
|
| 308 |
+
"step": 500
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 340.0,
|
| 312 |
+
"learning_rate": 2.4214731023046793e-05,
|
| 313 |
+
"loss": 0.0025,
|
| 314 |
+
"step": 510
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 346.67,
|
| 318 |
+
"learning_rate": 2.3430237011767167e-05,
|
| 319 |
+
"loss": 0.0026,
|
| 320 |
+
"step": 520
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 353.33,
|
| 324 |
+
"learning_rate": 2.2647292167037144e-05,
|
| 325 |
+
"loss": 0.0025,
|
| 326 |
+
"step": 530
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 360.0,
|
| 330 |
+
"learning_rate": 2.186666916089239e-05,
|
| 331 |
+
"loss": 0.0026,
|
| 332 |
+
"step": 540
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 366.67,
|
| 336 |
+
"learning_rate": 2.1089138373994223e-05,
|
| 337 |
+
"loss": 0.0025,
|
| 338 |
+
"step": 550
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 373.33,
|
| 342 |
+
"learning_rate": 2.031546713535688e-05,
|
| 343 |
+
"loss": 0.0025,
|
| 344 |
+
"step": 560
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 380.0,
|
| 348 |
+
"learning_rate": 1.9546418965086442e-05,
|
| 349 |
+
"loss": 0.0025,
|
| 350 |
+
"step": 570
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 386.67,
|
| 354 |
+
"learning_rate": 1.8782752820878634e-05,
|
| 355 |
+
"loss": 0.0025,
|
| 356 |
+
"step": 580
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 393.33,
|
| 360 |
+
"learning_rate": 1.802522234901927e-05,
|
| 361 |
+
"loss": 0.0025,
|
| 362 |
+
"step": 590
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 400.0,
|
| 366 |
+
"learning_rate": 1.7274575140626318e-05,
|
| 367 |
+
"loss": 0.0025,
|
| 368 |
+
"step": 600
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 406.67,
|
| 372 |
+
"learning_rate": 1.6531551993867717e-05,
|
| 373 |
+
"loss": 0.0025,
|
| 374 |
+
"step": 610
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 413.33,
|
| 378 |
+
"learning_rate": 1.5796886182883053e-05,
|
| 379 |
+
"loss": 0.0025,
|
| 380 |
+
"step": 620
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 420.0,
|
| 384 |
+
"learning_rate": 1.5071302734130489e-05,
|
| 385 |
+
"loss": 0.0025,
|
| 386 |
+
"step": 630
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 426.67,
|
| 390 |
+
"learning_rate": 1.4355517710873184e-05,
|
| 391 |
+
"loss": 0.0025,
|
| 392 |
+
"step": 640
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 433.33,
|
| 396 |
+
"learning_rate": 1.3650237506511331e-05,
|
| 397 |
+
"loss": 0.0025,
|
| 398 |
+
"step": 650
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 440.0,
|
| 402 |
+
"learning_rate": 1.2956158147457115e-05,
|
| 403 |
+
"loss": 0.0025,
|
| 404 |
+
"step": 660
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 446.67,
|
| 408 |
+
"learning_rate": 1.2273964606240718e-05,
|
| 409 |
+
"loss": 0.0025,
|
| 410 |
+
"step": 670
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 453.33,
|
| 414 |
+
"learning_rate": 1.1604330125525079e-05,
|
| 415 |
+
"loss": 0.0025,
|
| 416 |
+
"step": 680
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 460.0,
|
| 420 |
+
"learning_rate": 1.0947915553696742e-05,
|
| 421 |
+
"loss": 0.0025,
|
| 422 |
+
"step": 690
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 466.67,
|
| 426 |
+
"learning_rate": 1.0305368692688174e-05,
|
| 427 |
+
"loss": 0.0025,
|
| 428 |
+
"step": 700
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 473.33,
|
| 432 |
+
"learning_rate": 9.677323658675594e-06,
|
| 433 |
+
"loss": 0.0025,
|
| 434 |
+
"step": 710
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 480.0,
|
| 438 |
+
"learning_rate": 9.064400256282757e-06,
|
| 439 |
+
"loss": 0.0025,
|
| 440 |
+
"step": 720
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 486.67,
|
| 444 |
+
"learning_rate": 8.467203366908707e-06,
|
| 445 |
+
"loss": 0.0025,
|
| 446 |
+
"step": 730
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 493.33,
|
| 450 |
+
"learning_rate": 7.886322351782783e-06,
|
| 451 |
+
"loss": 0.0025,
|
| 452 |
+
"step": 740
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 500.0,
|
| 456 |
+
"learning_rate": 7.3223304703363135e-06,
|
| 457 |
+
"loss": 0.0025,
|
| 458 |
+
"step": 750
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 506.67,
|
| 462 |
+
"learning_rate": 6.775784314464717e-06,
|
| 463 |
+
"loss": 0.0025,
|
| 464 |
+
"step": 760
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 513.33,
|
| 468 |
+
"learning_rate": 6.247223259238511e-06,
|
| 469 |
+
"loss": 0.0025,
|
| 470 |
+
"step": 770
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 520.0,
|
| 474 |
+
"learning_rate": 5.737168930605272e-06,
|
| 475 |
+
"loss": 0.0025,
|
| 476 |
+
"step": 780
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 526.67,
|
| 480 |
+
"learning_rate": 5.24612469060774e-06,
|
| 481 |
+
"loss": 0.0025,
|
| 482 |
+
"step": 790
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 533.33,
|
| 486 |
+
"learning_rate": 4.7745751406263165e-06,
|
| 487 |
+
"loss": 0.0025,
|
| 488 |
+
"step": 800
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 540.0,
|
| 492 |
+
"learning_rate": 4.322985643135952e-06,
|
| 493 |
+
"loss": 0.0025,
|
| 494 |
+
"step": 810
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 546.67,
|
| 498 |
+
"learning_rate": 3.891801862449629e-06,
|
| 499 |
+
"loss": 0.0025,
|
| 500 |
+
"step": 820
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 553.33,
|
| 504 |
+
"learning_rate": 3.4814493249014116e-06,
|
| 505 |
+
"loss": 0.0025,
|
| 506 |
+
"step": 830
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 560.0,
|
| 510 |
+
"learning_rate": 3.092332998903416e-06,
|
| 511 |
+
"loss": 0.0025,
|
| 512 |
+
"step": 840
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 566.67,
|
| 516 |
+
"learning_rate": 2.7248368952908053e-06,
|
| 517 |
+
"loss": 0.0025,
|
| 518 |
+
"step": 850
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 573.33,
|
| 522 |
+
"learning_rate": 2.379323688349516e-06,
|
| 523 |
+
"loss": 0.0025,
|
| 524 |
+
"step": 860
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 580.0,
|
| 528 |
+
"learning_rate": 2.0561343579004715e-06,
|
| 529 |
+
"loss": 0.0024,
|
| 530 |
+
"step": 870
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 586.67,
|
| 534 |
+
"learning_rate": 1.7555878527937164e-06,
|
| 535 |
+
"loss": 0.0025,
|
| 536 |
+
"step": 880
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 593.33,
|
| 540 |
+
"learning_rate": 1.4779807761443636e-06,
|
| 541 |
+
"loss": 0.0025,
|
| 542 |
+
"step": 890
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 600.0,
|
| 546 |
+
"learning_rate": 1.2235870926211619e-06,
|
| 547 |
+
"loss": 0.0025,
|
| 548 |
+
"step": 900
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 606.67,
|
| 552 |
+
"learning_rate": 9.926578580764234e-07,
|
| 553 |
+
"loss": 0.0024,
|
| 554 |
+
"step": 910
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 613.33,
|
| 558 |
+
"learning_rate": 7.854209717842231e-07,
|
| 559 |
+
"loss": 0.0025,
|
| 560 |
+
"step": 920
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 620.0,
|
| 564 |
+
"learning_rate": 6.020809515313142e-07,
|
| 565 |
+
"loss": 0.0025,
|
| 566 |
+
"step": 930
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 626.67,
|
| 570 |
+
"learning_rate": 4.4281873178278475e-07,
|
| 571 |
+
"loss": 0.0025,
|
| 572 |
+
"step": 940
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 633.33,
|
| 576 |
+
"learning_rate": 3.077914851215585e-07,
|
| 577 |
+
"loss": 0.0025,
|
| 578 |
+
"step": 950
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 640.0,
|
| 582 |
+
"learning_rate": 1.9713246713805588e-07,
|
| 583 |
+
"loss": 0.0024,
|
| 584 |
+
"step": 960
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 646.67,
|
| 588 |
+
"learning_rate": 1.109508849230001e-07,
|
| 589 |
+
"loss": 0.0025,
|
| 590 |
+
"step": 970
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 653.33,
|
| 594 |
+
"learning_rate": 4.9331789293211026e-08,
|
| 595 |
+
"loss": 0.0025,
|
| 596 |
+
"step": 980
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 660.0,
|
| 600 |
+
"learning_rate": 1.233599085671e-08,
|
| 601 |
+
"loss": 0.0025,
|
| 602 |
+
"step": 990
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 666.67,
|
| 606 |
+
"learning_rate": 0.0,
|
| 607 |
+
"loss": 0.0025,
|
| 608 |
+
"step": 1000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 666.67,
|
| 612 |
+
"step": 1000,
|
| 613 |
+
"total_flos": 1.3543137746092032e+17,
|
| 614 |
+
"train_loss": 0.06853693281114101,
|
| 615 |
+
"train_runtime": 2944.4055,
|
| 616 |
+
"train_samples_per_second": 3.396,
|
| 617 |
+
"train_steps_per_second": 0.34
|
| 618 |
+
}
|
| 619 |
+
],
|
| 620 |
+
"max_steps": 1000,
|
| 621 |
+
"num_train_epochs": 1000,
|
| 622 |
+
"total_flos": 1.3543137746092032e+17,
|
| 623 |
+
"trial_name": null,
|
| 624 |
+
"trial_params": null
|
| 625 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ca49ced37f60858313781f6d053a0ecdf141b073e86b9cd38e9457eaea69ea72
|
| 3 |
+
size 3262
|
training_loss.png
ADDED
|