Upload 9 files
Browse files- adapter_config.json +21 -0
- adapter_model.bin +3 -0
- all_results.json +7 -0
- finetuning_args.json +16 -0
- train_results.json +7 -0
- trainer_log.jsonl +155 -0
- trainer_state.json +952 -0
- training_args.bin +3 -0
- training_loss.png +0 -0
adapter_config.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"auto_mapping": null,
|
| 3 |
+
"base_model_name_or_path": "prognosis/cardio-llama-2-7b-miniguanaco-v13",
|
| 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 sha256:7222022803fbec92db08e8d860c0df909da59b2a65110cf4763a32e464fee274
|
| 3 |
+
size 16821197
|
all_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.6278958536241713,
|
| 4 |
+
"train_runtime": 4523.4188,
|
| 5 |
+
"train_samples_per_second": 5.483,
|
| 6 |
+
"train_steps_per_second": 0.342
|
| 7 |
+
}
|
finetuning_args.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"dpo_beta": 0.1,
|
| 3 |
+
"finetuning_type": "lora",
|
| 4 |
+
"lora_alpha": 32.0,
|
| 5 |
+
"lora_dropout": 0.1,
|
| 6 |
+
"lora_rank": 8,
|
| 7 |
+
"lora_target": [
|
| 8 |
+
"q_proj",
|
| 9 |
+
"v_proj"
|
| 10 |
+
],
|
| 11 |
+
"name_module_trainable": "mlp",
|
| 12 |
+
"num_hidden_layers": 32,
|
| 13 |
+
"num_layer_trainable": 3,
|
| 14 |
+
"ppo_score_norm": false,
|
| 15 |
+
"resume_lora_training": true
|
| 16 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.6278958536241713,
|
| 4 |
+
"train_runtime": 4523.4188,
|
| 5 |
+
"train_samples_per_second": 5.483,
|
| 6 |
+
"train_steps_per_second": 0.342
|
| 7 |
+
}
|
trainer_log.jsonl
ADDED
|
@@ -0,0 +1,155 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"current_steps": 10, "total_steps": 1548, "loss": 1.6087, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.999747735219333e-05, "epoch": 0.02, "percentage": 0.65, "elapsed_time": "0:00:31", "remaining_time": "1:21:56"}
|
| 2 |
+
{"current_steps": 20, "total_steps": 1548, "loss": 1.4212, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.998512275175808e-05, "epoch": 0.04, "percentage": 1.29, "elapsed_time": "0:01:01", "remaining_time": "1:17:44"}
|
| 3 |
+
{"current_steps": 30, "total_steps": 1548, "loss": 1.0602, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9962477937145644e-05, "epoch": 0.06, "percentage": 1.94, "elapsed_time": "0:01:30", "remaining_time": "1:16:04"}
|
| 4 |
+
{"current_steps": 40, "total_steps": 1548, "loss": 1.0232, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.992955223470575e-05, "epoch": 0.08, "percentage": 2.58, "elapsed_time": "0:01:59", "remaining_time": "1:15:01"}
|
| 5 |
+
{"current_steps": 50, "total_steps": 1548, "loss": 1.0354, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9886359205009334e-05, "epoch": 0.1, "percentage": 3.23, "elapsed_time": "0:02:28", "remaining_time": "1:14:11"}
|
| 6 |
+
{"current_steps": 60, "total_steps": 1548, "loss": 0.8403, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9832916637263665e-05, "epoch": 0.12, "percentage": 3.88, "elapsed_time": "0:02:57", "remaining_time": "1:13:29"}
|
| 7 |
+
{"current_steps": 70, "total_steps": 1548, "loss": 0.7628, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.976924654198569e-05, "epoch": 0.14, "percentage": 4.52, "elapsed_time": "0:03:27", "remaining_time": "1:12:50"}
|
| 8 |
+
{"current_steps": 80, "total_steps": 1548, "loss": 0.7421, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9695375141937e-05, "epoch": 0.15, "percentage": 5.17, "elapsed_time": "0:03:56", "remaining_time": "1:12:14"}
|
| 9 |
+
{"current_steps": 90, "total_steps": 1548, "loss": 0.7773, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9611332861323875e-05, "epoch": 0.17, "percentage": 5.81, "elapsed_time": "0:04:25", "remaining_time": "1:11:39"}
|
| 10 |
+
{"current_steps": 100, "total_steps": 1548, "loss": 0.7814, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.95171543132669e-05, "epoch": 0.19, "percentage": 6.46, "elapsed_time": "0:04:54", "remaining_time": "1:11:05"}
|
| 11 |
+
{"current_steps": 110, "total_steps": 1548, "loss": 0.7798, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.941287828554553e-05, "epoch": 0.21, "percentage": 7.11, "elapsed_time": "0:05:23", "remaining_time": "1:10:33"}
|
| 12 |
+
{"current_steps": 120, "total_steps": 1548, "loss": 0.7717, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.929854772462312e-05, "epoch": 0.23, "percentage": 7.75, "elapsed_time": "0:05:53", "remaining_time": "1:10:00"}
|
| 13 |
+
{"current_steps": 130, "total_steps": 1548, "loss": 0.8173, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9174209717959294e-05, "epoch": 0.25, "percentage": 8.4, "elapsed_time": "0:06:22", "remaining_time": "1:09:29"}
|
| 14 |
+
{"current_steps": 140, "total_steps": 1548, "loss": 0.8009, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9039915474616805e-05, "epoch": 0.27, "percentage": 9.04, "elapsed_time": "0:06:51", "remaining_time": "1:08:57"}
|
| 15 |
+
{"current_steps": 150, "total_steps": 1548, "loss": 0.7499, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.889572030417091e-05, "epoch": 0.29, "percentage": 9.69, "elapsed_time": "0:07:20", "remaining_time": "1:08:26"}
|
| 16 |
+
{"current_steps": 160, "total_steps": 1548, "loss": 0.6257, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.874168359392987e-05, "epoch": 0.31, "percentage": 10.34, "elapsed_time": "0:07:49", "remaining_time": "1:07:56"}
|
| 17 |
+
{"current_steps": 170, "total_steps": 1548, "loss": 0.7029, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.857786878447612e-05, "epoch": 0.33, "percentage": 10.98, "elapsed_time": "0:08:19", "remaining_time": "1:07:25"}
|
| 18 |
+
{"current_steps": 180, "total_steps": 1548, "loss": 0.6505, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.8404343343538014e-05, "epoch": 0.35, "percentage": 11.63, "elapsed_time": "0:08:48", "remaining_time": "1:06:54"}
|
| 19 |
+
{"current_steps": 190, "total_steps": 1548, "loss": 0.6647, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.822117873820301e-05, "epoch": 0.37, "percentage": 12.27, "elapsed_time": "0:09:17", "remaining_time": "1:06:24"}
|
| 20 |
+
{"current_steps": 200, "total_steps": 1548, "loss": 0.6787, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.802845040548363e-05, "epoch": 0.39, "percentage": 12.92, "elapsed_time": "0:09:46", "remaining_time": "1:05:54"}
|
| 21 |
+
{"current_steps": 210, "total_steps": 1548, "loss": 0.5937, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.782623772124855e-05, "epoch": 0.41, "percentage": 13.57, "elapsed_time": "0:10:15", "remaining_time": "1:05:24"}
|
| 22 |
+
{"current_steps": 220, "total_steps": 1548, "loss": 0.649, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.7614623967531244e-05, "epoch": 0.43, "percentage": 14.21, "elapsed_time": "0:10:45", "remaining_time": "1:04:53"}
|
| 23 |
+
{"current_steps": 230, "total_steps": 1548, "loss": 0.729, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.7393696298230084e-05, "epoch": 0.45, "percentage": 14.86, "elapsed_time": "0:11:14", "remaining_time": "1:04:24"}
|
| 24 |
+
{"current_steps": 240, "total_steps": 1548, "loss": 0.6169, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.716354570321361e-05, "epoch": 0.46, "percentage": 15.5, "elapsed_time": "0:11:43", "remaining_time": "1:03:54"}
|
| 25 |
+
{"current_steps": 250, "total_steps": 1548, "loss": 0.6484, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.692426697084605e-05, "epoch": 0.48, "percentage": 16.15, "elapsed_time": "0:12:12", "remaining_time": "1:03:24"}
|
| 26 |
+
{"current_steps": 260, "total_steps": 1548, "loss": 0.6714, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.6675958648948394e-05, "epoch": 0.5, "percentage": 16.8, "elapsed_time": "0:12:41", "remaining_time": "1:02:54"}
|
| 27 |
+
{"current_steps": 270, "total_steps": 1548, "loss": 0.6608, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.6418723004211075e-05, "epoch": 0.52, "percentage": 17.44, "elapsed_time": "0:13:11", "remaining_time": "1:02:24"}
|
| 28 |
+
{"current_steps": 280, "total_steps": 1548, "loss": 0.6982, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.615266598007512e-05, "epoch": 0.54, "percentage": 18.09, "elapsed_time": "0:13:40", "remaining_time": "1:01:55"}
|
| 29 |
+
{"current_steps": 290, "total_steps": 1548, "loss": 0.6304, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.587789715309888e-05, "epoch": 0.56, "percentage": 18.73, "elapsed_time": "0:14:09", "remaining_time": "1:01:25"}
|
| 30 |
+
{"current_steps": 300, "total_steps": 1548, "loss": 0.6261, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.559452968782861e-05, "epoch": 0.58, "percentage": 19.38, "elapsed_time": "0:14:38", "remaining_time": "1:00:55"}
|
| 31 |
+
{"current_steps": 310, "total_steps": 1548, "loss": 0.6385, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.530268029019117e-05, "epoch": 0.6, "percentage": 20.03, "elapsed_time": "0:15:08", "remaining_time": "1:00:26"}
|
| 32 |
+
{"current_steps": 320, "total_steps": 1548, "loss": 0.6703, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.500246915942827e-05, "epoch": 0.62, "percentage": 20.67, "elapsed_time": "0:15:37", "remaining_time": "0:59:56"}
|
| 33 |
+
{"current_steps": 330, "total_steps": 1548, "loss": 0.62, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.469401993859201e-05, "epoch": 0.64, "percentage": 21.32, "elapsed_time": "0:16:06", "remaining_time": "0:59:26"}
|
| 34 |
+
{"current_steps": 340, "total_steps": 1548, "loss": 0.7172, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.437745966362201e-05, "epoch": 0.66, "percentage": 21.96, "elapsed_time": "0:16:35", "remaining_time": "0:58:57"}
|
| 35 |
+
{"current_steps": 350, "total_steps": 1548, "loss": 0.5989, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.4052918711025194e-05, "epoch": 0.68, "percentage": 22.61, "elapsed_time": "0:17:04", "remaining_time": "0:58:27"}
|
| 36 |
+
{"current_steps": 360, "total_steps": 1548, "loss": 0.6586, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.372053074417975e-05, "epoch": 0.7, "percentage": 23.26, "elapsed_time": "0:17:34", "remaining_time": "0:57:58"}
|
| 37 |
+
{"current_steps": 370, "total_steps": 1548, "loss": 0.6459, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.3380432658285367e-05, "epoch": 0.72, "percentage": 23.9, "elapsed_time": "0:18:03", "remaining_time": "0:57:28"}
|
| 38 |
+
{"current_steps": 380, "total_steps": 1548, "loss": 0.6918, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.3032764523982496e-05, "epoch": 0.74, "percentage": 24.55, "elapsed_time": "0:18:32", "remaining_time": "0:56:59"}
|
| 39 |
+
{"current_steps": 390, "total_steps": 1548, "loss": 0.6366, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.267766952966369e-05, "epoch": 0.75, "percentage": 25.19, "elapsed_time": "0:19:01", "remaining_time": "0:56:29"}
|
| 40 |
+
{"current_steps": 400, "total_steps": 1548, "loss": 0.6692, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.231529392250095e-05, "epoch": 0.77, "percentage": 25.84, "elapsed_time": "0:19:30", "remaining_time": "0:56:00"}
|
| 41 |
+
{"current_steps": 410, "total_steps": 1548, "loss": 0.715, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.194578694821332e-05, "epoch": 0.79, "percentage": 26.49, "elapsed_time": "0:20:00", "remaining_time": "0:55:31"}
|
| 42 |
+
{"current_steps": 420, "total_steps": 1548, "loss": 0.6242, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.156930078959946e-05, "epoch": 0.81, "percentage": 27.13, "elapsed_time": "0:20:29", "remaining_time": "0:55:01"}
|
| 43 |
+
{"current_steps": 430, "total_steps": 1548, "loss": 0.7372, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.1185990503860626e-05, "epoch": 0.83, "percentage": 27.78, "elapsed_time": "0:20:58", "remaining_time": "0:54:32"}
|
| 44 |
+
{"current_steps": 440, "total_steps": 1548, "loss": 0.6203, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.079601395873979e-05, "epoch": 0.85, "percentage": 28.42, "elapsed_time": "0:21:27", "remaining_time": "0:54:02"}
|
| 45 |
+
{"current_steps": 450, "total_steps": 1548, "loss": 0.7101, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.0399531767503204e-05, "epoch": 0.87, "percentage": 29.07, "elapsed_time": "0:21:56", "remaining_time": "0:53:33"}
|
| 46 |
+
{"current_steps": 460, "total_steps": 1548, "loss": 0.6866, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.999670722279131e-05, "epoch": 0.89, "percentage": 29.72, "elapsed_time": "0:22:26", "remaining_time": "0:53:04"}
|
| 47 |
+
{"current_steps": 470, "total_steps": 1548, "loss": 0.5917, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.958770622936599e-05, "epoch": 0.91, "percentage": 30.36, "elapsed_time": "0:22:55", "remaining_time": "0:52:34"}
|
| 48 |
+
{"current_steps": 480, "total_steps": 1548, "loss": 0.664, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.917269723578212e-05, "epoch": 0.93, "percentage": 31.01, "elapsed_time": "0:23:24", "remaining_time": "0:52:05"}
|
| 49 |
+
{"current_steps": 490, "total_steps": 1548, "loss": 0.6763, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.8751851165011413e-05, "epoch": 0.95, "percentage": 31.65, "elapsed_time": "0:23:53", "remaining_time": "0:51:35"}
|
| 50 |
+
{"current_steps": 500, "total_steps": 1548, "loss": 0.615, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.8325341344047174e-05, "epoch": 0.97, "percentage": 32.3, "elapsed_time": "0:24:23", "remaining_time": "0:51:06"}
|
| 51 |
+
{"current_steps": 510, "total_steps": 1548, "loss": 0.5707, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.7893343432518946e-05, "epoch": 0.99, "percentage": 32.95, "elapsed_time": "0:24:52", "remaining_time": "0:50:37"}
|
| 52 |
+
{"current_steps": 520, "total_steps": 1548, "loss": 0.6939, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.745603535034641e-05, "epoch": 1.01, "percentage": 33.59, "elapsed_time": "0:25:21", "remaining_time": "0:50:07"}
|
| 53 |
+
{"current_steps": 530, "total_steps": 1548, "loss": 0.6863, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.701359720446249e-05, "epoch": 1.03, "percentage": 34.24, "elapsed_time": "0:25:50", "remaining_time": "0:49:38"}
|
| 54 |
+
{"current_steps": 540, "total_steps": 1548, "loss": 0.6671, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.656621121463557e-05, "epoch": 1.04, "percentage": 34.88, "elapsed_time": "0:26:19", "remaining_time": "0:49:09"}
|
| 55 |
+
{"current_steps": 550, "total_steps": 1548, "loss": 0.518, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.611406163842168e-05, "epoch": 1.06, "percentage": 35.53, "elapsed_time": "0:26:49", "remaining_time": "0:48:39"}
|
| 56 |
+
{"current_steps": 560, "total_steps": 1548, "loss": 0.6747, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.565733469527731e-05, "epoch": 1.08, "percentage": 36.18, "elapsed_time": "0:27:18", "remaining_time": "0:48:10"}
|
| 57 |
+
{"current_steps": 570, "total_steps": 1548, "loss": 0.5754, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.519621848986428e-05, "epoch": 1.1, "percentage": 36.82, "elapsed_time": "0:27:47", "remaining_time": "0:47:41"}
|
| 58 |
+
{"current_steps": 580, "total_steps": 1548, "loss": 0.6061, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.473090293457811e-05, "epoch": 1.12, "percentage": 37.47, "elapsed_time": "0:28:16", "remaining_time": "0:47:11"}
|
| 59 |
+
{"current_steps": 590, "total_steps": 1548, "loss": 0.5859, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.426157967133192e-05, "epoch": 1.14, "percentage": 38.11, "elapsed_time": "0:28:45", "remaining_time": "0:46:42"}
|
| 60 |
+
{"current_steps": 600, "total_steps": 1548, "loss": 0.546, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.3788441992628026e-05, "epoch": 1.16, "percentage": 38.76, "elapsed_time": "0:29:15", "remaining_time": "0:46:13"}
|
| 61 |
+
{"current_steps": 610, "total_steps": 1548, "loss": 0.6333, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.331168476194969e-05, "epoch": 1.18, "percentage": 39.41, "elapsed_time": "0:29:44", "remaining_time": "0:45:43"}
|
| 62 |
+
{"current_steps": 620, "total_steps": 1548, "loss": 0.6332, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.283150433350589e-05, "epoch": 1.2, "percentage": 40.05, "elapsed_time": "0:30:13", "remaining_time": "0:45:14"}
|
| 63 |
+
{"current_steps": 630, "total_steps": 1548, "loss": 0.5709, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.234809847136213e-05, "epoch": 1.22, "percentage": 40.7, "elapsed_time": "0:30:42", "remaining_time": "0:44:45"}
|
| 64 |
+
{"current_steps": 640, "total_steps": 1548, "loss": 0.578, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.1861666267990566e-05, "epoch": 1.24, "percentage": 41.34, "elapsed_time": "0:31:11", "remaining_time": "0:44:15"}
|
| 65 |
+
{"current_steps": 650, "total_steps": 1548, "loss": 0.537, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.137240806227306e-05, "epoch": 1.26, "percentage": 41.99, "elapsed_time": "0:31:41", "remaining_time": "0:43:46"}
|
| 66 |
+
{"current_steps": 660, "total_steps": 1548, "loss": 0.5905, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.08805253569909e-05, "epoch": 1.28, "percentage": 42.64, "elapsed_time": "0:32:10", "remaining_time": "0:43:17"}
|
| 67 |
+
{"current_steps": 670, "total_steps": 1548, "loss": 0.6015, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.038622073583507e-05, "epoch": 1.3, "percentage": 43.28, "elapsed_time": "0:32:39", "remaining_time": "0:42:47"}
|
| 68 |
+
{"current_steps": 680, "total_steps": 1548, "loss": 0.5735, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.9889697779971504e-05, "epoch": 1.32, "percentage": 43.93, "elapsed_time": "0:33:08", "remaining_time": "0:42:18"}
|
| 69 |
+
{"current_steps": 690, "total_steps": 1548, "loss": 0.5791, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.9391160984195382e-05, "epoch": 1.34, "percentage": 44.57, "elapsed_time": "0:33:38", "remaining_time": "0:41:49"}
|
| 70 |
+
{"current_steps": 700, "total_steps": 1548, "loss": 0.5683, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.8890815672709225e-05, "epoch": 1.35, "percentage": 45.22, "elapsed_time": "0:34:07", "remaining_time": "0:41:20"}
|
| 71 |
+
{"current_steps": 710, "total_steps": 1548, "loss": 0.6583, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.8388867914559347e-05, "epoch": 1.37, "percentage": 45.87, "elapsed_time": "0:34:36", "remaining_time": "0:40:50"}
|
| 72 |
+
{"current_steps": 720, "total_steps": 1548, "loss": 0.6091, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.7885524438765603e-05, "epoch": 1.39, "percentage": 46.51, "elapsed_time": "0:35:05", "remaining_time": "0:40:21"}
|
| 73 |
+
{"current_steps": 730, "total_steps": 1548, "loss": 0.5614, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.7380992549179235e-05, "epoch": 1.41, "percentage": 47.16, "elapsed_time": "0:35:34", "remaining_time": "0:39:52"}
|
| 74 |
+
{"current_steps": 740, "total_steps": 1548, "loss": 0.6208, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.6875480039104085e-05, "epoch": 1.43, "percentage": 47.8, "elapsed_time": "0:36:04", "remaining_time": "0:39:22"}
|
| 75 |
+
{"current_steps": 750, "total_steps": 1548, "loss": 0.5686, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.636919510571609e-05, "epoch": 1.45, "percentage": 48.45, "elapsed_time": "0:36:33", "remaining_time": "0:38:53"}
|
| 76 |
+
{"current_steps": 760, "total_steps": 1548, "loss": 0.5565, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.5862346264316605e-05, "epoch": 1.47, "percentage": 49.1, "elapsed_time": "0:37:02", "remaining_time": "0:38:24"}
|
| 77 |
+
{"current_steps": 770, "total_steps": 1548, "loss": 0.5936, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.5355142262454506e-05, "epoch": 1.49, "percentage": 49.74, "elapsed_time": "0:37:31", "remaining_time": "0:37:55"}
|
| 78 |
+
{"current_steps": 780, "total_steps": 1548, "loss": 0.6469, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.484779199395285e-05, "epoch": 1.51, "percentage": 50.39, "elapsed_time": "0:38:00", "remaining_time": "0:37:25"}
|
| 79 |
+
{"current_steps": 790, "total_steps": 1548, "loss": 0.5551, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.4340504412875113e-05, "epoch": 1.53, "percentage": 51.03, "elapsed_time": "0:38:30", "remaining_time": "0:36:56"}
|
| 80 |
+
{"current_steps": 800, "total_steps": 1548, "loss": 0.6454, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.3833488447466746e-05, "epoch": 1.55, "percentage": 51.68, "elapsed_time": "0:38:59", "remaining_time": "0:36:27"}
|
| 81 |
+
{"current_steps": 810, "total_steps": 1548, "loss": 0.61, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.3326952914107268e-05, "epoch": 1.57, "percentage": 52.33, "elapsed_time": "0:39:28", "remaining_time": "0:35:57"}
|
| 82 |
+
{"current_steps": 820, "total_steps": 1548, "loss": 0.6171, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.2821106431308544e-05, "epoch": 1.59, "percentage": 52.97, "elapsed_time": "0:39:57", "remaining_time": "0:35:28"}
|
| 83 |
+
{"current_steps": 830, "total_steps": 1548, "loss": 0.5636, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.2316157333794414e-05, "epoch": 1.61, "percentage": 53.62, "elapsed_time": "0:40:26", "remaining_time": "0:34:59"}
|
| 84 |
+
{"current_steps": 840, "total_steps": 1548, "loss": 0.5094, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.1812313586697307e-05, "epoch": 1.63, "percentage": 54.26, "elapsed_time": "0:40:56", "remaining_time": "0:34:30"}
|
| 85 |
+
{"current_steps": 850, "total_steps": 1548, "loss": 0.6278, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.1309782699907042e-05, "epoch": 1.64, "percentage": 54.91, "elapsed_time": "0:41:25", "remaining_time": "0:34:00"}
|
| 86 |
+
{"current_steps": 860, "total_steps": 1548, "loss": 0.5556, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.0808771642607146e-05, "epoch": 1.66, "percentage": 55.56, "elapsed_time": "0:41:54", "remaining_time": "0:33:31"}
|
| 87 |
+
{"current_steps": 870, "total_steps": 1548, "loss": 0.5103, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.0309486758033773e-05, "epoch": 1.68, "percentage": 56.2, "elapsed_time": "0:42:23", "remaining_time": "0:33:02"}
|
| 88 |
+
{"current_steps": 880, "total_steps": 1548, "loss": 0.5002, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.9812133678492554e-05, "epoch": 1.7, "percentage": 56.85, "elapsed_time": "0:42:52", "remaining_time": "0:32:32"}
|
| 89 |
+
{"current_steps": 890, "total_steps": 1548, "loss": 0.626, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.9316917240668133e-05, "epoch": 1.72, "percentage": 57.49, "elapsed_time": "0:43:21", "remaining_time": "0:32:03"}
|
| 90 |
+
{"current_steps": 900, "total_steps": 1548, "loss": 0.5432, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.8824041401261462e-05, "epoch": 1.74, "percentage": 58.14, "elapsed_time": "0:43:51", "remaining_time": "0:31:34"}
|
| 91 |
+
{"current_steps": 910, "total_steps": 1548, "loss": 0.5423, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.833370915298948e-05, "epoch": 1.76, "percentage": 58.79, "elapsed_time": "0:44:20", "remaining_time": "0:31:05"}
|
| 92 |
+
{"current_steps": 920, "total_steps": 1548, "loss": 0.5179, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.784612244098181e-05, "epoch": 1.78, "percentage": 59.43, "elapsed_time": "0:44:49", "remaining_time": "0:30:35"}
|
| 93 |
+
{"current_steps": 930, "total_steps": 1548, "loss": 0.5541, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.7361482079608914e-05, "epoch": 1.8, "percentage": 60.08, "elapsed_time": "0:45:18", "remaining_time": "0:30:06"}
|
| 94 |
+
{"current_steps": 940, "total_steps": 1548, "loss": 0.6554, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.687998766977597e-05, "epoch": 1.82, "percentage": 60.72, "elapsed_time": "0:45:47", "remaining_time": "0:29:37"}
|
| 95 |
+
{"current_steps": 950, "total_steps": 1548, "loss": 0.601, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.6401837516716546e-05, "epoch": 1.84, "percentage": 61.37, "elapsed_time": "0:46:17", "remaining_time": "0:29:08"}
|
| 96 |
+
{"current_steps": 960, "total_steps": 1548, "loss": 0.5267, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.5927228548319767e-05, "epoch": 1.86, "percentage": 62.02, "elapsed_time": "0:46:46", "remaining_time": "0:28:38"}
|
| 97 |
+
{"current_steps": 970, "total_steps": 1548, "loss": 0.5281, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.545635623402496e-05, "epoch": 1.88, "percentage": 62.66, "elapsed_time": "0:47:15", "remaining_time": "0:28:09"}
|
| 98 |
+
{"current_steps": 980, "total_steps": 1548, "loss": 0.5823, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.4989414504316748e-05, "epoch": 1.9, "percentage": 63.31, "elapsed_time": "0:47:44", "remaining_time": "0:27:40"}
|
| 99 |
+
{"current_steps": 990, "total_steps": 1548, "loss": 0.5789, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.4526595670854159e-05, "epoch": 1.92, "percentage": 63.95, "elapsed_time": "0:48:13", "remaining_time": "0:27:11"}
|
| 100 |
+
{"current_steps": 1000, "total_steps": 1548, "loss": 0.6653, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.4068090347266311e-05, "epoch": 1.94, "percentage": 64.6, "elapsed_time": "0:48:43", "remaining_time": "0:26:41"}
|
| 101 |
+
{"current_steps": 1010, "total_steps": 1548, "loss": 0.5199, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.3614087370647479e-05, "epoch": 1.95, "percentage": 65.25, "elapsed_time": "0:49:12", "remaining_time": "0:26:12"}
|
| 102 |
+
{"current_steps": 1020, "total_steps": 1548, "loss": 0.6317, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.3164773723783918e-05, "epoch": 1.97, "percentage": 65.89, "elapsed_time": "0:49:41", "remaining_time": "0:25:43"}
|
| 103 |
+
{"current_steps": 1030, "total_steps": 1548, "loss": 0.4948, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.2720334458144235e-05, "epoch": 1.99, "percentage": 66.54, "elapsed_time": "0:50:10", "remaining_time": "0:25:14"}
|
| 104 |
+
{"current_steps": 1040, "total_steps": 1548, "loss": 0.6401, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.2280952617665334e-05, "epoch": 2.01, "percentage": 67.18, "elapsed_time": "0:50:40", "remaining_time": "0:24:44"}
|
| 105 |
+
{"current_steps": 1050, "total_steps": 1548, "loss": 0.5579, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.1846809163365052e-05, "epoch": 2.03, "percentage": 67.83, "elapsed_time": "0:51:09", "remaining_time": "0:24:15"}
|
| 106 |
+
{"current_steps": 1060, "total_steps": 1548, "loss": 0.4654, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.1418082898812721e-05, "epoch": 2.05, "percentage": 68.48, "elapsed_time": "0:51:38", "remaining_time": "0:23:46"}
|
| 107 |
+
{"current_steps": 1070, "total_steps": 1548, "loss": 0.5219, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.0994950396488275e-05, "epoch": 2.07, "percentage": 69.12, "elapsed_time": "0:52:07", "remaining_time": "0:23:17"}
|
| 108 |
+
{"current_steps": 1080, "total_steps": 1548, "loss": 0.6016, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.057758592506022e-05, "epoch": 2.09, "percentage": 69.77, "elapsed_time": "0:52:36", "remaining_time": "0:22:47"}
|
| 109 |
+
{"current_steps": 1090, "total_steps": 1548, "loss": 0.515, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.0166161377612437e-05, "epoch": 2.11, "percentage": 70.41, "elapsed_time": "0:53:06", "remaining_time": "0:22:18"}
|
| 110 |
+
{"current_steps": 1100, "total_steps": 1548, "loss": 0.5687, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 9.760846200849388e-06, "epoch": 2.13, "percentage": 71.06, "elapsed_time": "0:53:35", "remaining_time": "0:21:49"}
|
| 111 |
+
{"current_steps": 1110, "total_steps": 1548, "loss": 0.5587, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 9.361807325308861e-06, "epoch": 2.15, "percentage": 71.71, "elapsed_time": "0:54:04", "remaining_time": "0:21:20"}
|
| 112 |
+
{"current_steps": 1120, "total_steps": 1548, "loss": 0.5365, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 8.969209096611092e-06, "epoch": 2.17, "percentage": 72.35, "elapsed_time": "0:54:33", "remaining_time": "0:20:50"}
|
| 113 |
+
{"current_steps": 1130, "total_steps": 1548, "loss": 0.514, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 8.5832132077723e-06, "epoch": 2.19, "percentage": 73.0, "elapsed_time": "0:55:02", "remaining_time": "0:20:21"}
|
| 114 |
+
{"current_steps": 1140, "total_steps": 1548, "loss": 0.5671, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 8.203978632610915e-06, "epoch": 2.21, "percentage": 73.64, "elapsed_time": "0:55:32", "remaining_time": "0:19:52"}
|
| 115 |
+
{"current_steps": 1150, "total_steps": 1548, "loss": 0.5803, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 7.831661560273585e-06, "epoch": 2.23, "percentage": 74.29, "elapsed_time": "0:56:01", "remaining_time": "0:19:23"}
|
| 116 |
+
{"current_steps": 1160, "total_steps": 1548, "loss": 0.6003, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 7.466415330908147e-06, "epoch": 2.24, "percentage": 74.94, "elapsed_time": "0:56:30", "remaining_time": "0:18:54"}
|
| 117 |
+
{"current_steps": 1170, "total_steps": 1548, "loss": 0.59, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 7.108390372509893e-06, "epoch": 2.26, "percentage": 75.58, "elapsed_time": "0:56:59", "remaining_time": "0:18:24"}
|
| 118 |
+
{"current_steps": 1180, "total_steps": 1548, "loss": 0.6142, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 6.757734138967248e-06, "epoch": 2.28, "percentage": 76.23, "elapsed_time": "0:57:28", "remaining_time": "0:17:55"}
|
| 119 |
+
{"current_steps": 1190, "total_steps": 1548, "loss": 0.5202, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 6.414591049332366e-06, "epoch": 2.3, "percentage": 76.87, "elapsed_time": "0:57:58", "remaining_time": "0:17:26"}
|
| 120 |
+
{"current_steps": 1200, "total_steps": 1548, "loss": 0.5333, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 6.079102428341588e-06, "epoch": 2.32, "percentage": 77.52, "elapsed_time": "0:58:27", "remaining_time": "0:16:57"}
|
| 121 |
+
{"current_steps": 1210, "total_steps": 1548, "loss": 0.5676, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 5.7514064482104e-06, "epoch": 2.34, "percentage": 78.17, "elapsed_time": "0:58:56", "remaining_time": "0:16:27"}
|
| 122 |
+
{"current_steps": 1220, "total_steps": 1548, "loss": 0.5741, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 5.431638071726602e-06, "epoch": 2.36, "percentage": 78.81, "elapsed_time": "0:59:25", "remaining_time": "0:15:58"}
|
| 123 |
+
{"current_steps": 1230, "total_steps": 1548, "loss": 0.5904, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 5.11992899666546e-06, "epoch": 2.38, "percentage": 79.46, "elapsed_time": "0:59:54", "remaining_time": "0:15:29"}
|
| 124 |
+
{"current_steps": 1240, "total_steps": 1548, "loss": 0.5816, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.8164076015494695e-06, "epoch": 2.4, "percentage": 80.1, "elapsed_time": "1:00:23", "remaining_time": "0:15:00"}
|
| 125 |
+
{"current_steps": 1250, "total_steps": 1548, "loss": 0.5774, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.521198892775203e-06, "epoch": 2.42, "percentage": 80.75, "elapsed_time": "1:00:53", "remaining_time": "0:14:30"}
|
| 126 |
+
{"current_steps": 1260, "total_steps": 1548, "loss": 0.5844, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.234424453128974e-06, "epoch": 2.44, "percentage": 81.4, "elapsed_time": "1:01:22", "remaining_time": "0:14:01"}
|
| 127 |
+
{"current_steps": 1270, "total_steps": 1548, "loss": 0.5407, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.9562023917124905e-06, "epoch": 2.46, "percentage": 82.04, "elapsed_time": "1:01:51", "remaining_time": "0:13:32"}
|
| 128 |
+
{"current_steps": 1280, "total_steps": 1548, "loss": 0.5709, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.6866472952992226e-06, "epoch": 2.48, "percentage": 82.69, "elapsed_time": "1:02:20", "remaining_time": "0:13:03"}
|
| 129 |
+
{"current_steps": 1290, "total_steps": 1548, "loss": 0.5139, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.425870181141394e-06, "epoch": 2.5, "percentage": 83.33, "elapsed_time": "1:02:49", "remaining_time": "0:12:33"}
|
| 130 |
+
{"current_steps": 1300, "total_steps": 1548, "loss": 0.5911, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.173978451247153e-06, "epoch": 2.52, "percentage": 83.98, "elapsed_time": "1:03:19", "remaining_time": "0:12:04"}
|
| 131 |
+
{"current_steps": 1310, "total_steps": 1548, "loss": 0.5552, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.931075848146647e-06, "epoch": 2.54, "percentage": 84.63, "elapsed_time": "1:03:48", "remaining_time": "0:11:35"}
|
| 132 |
+
{"current_steps": 1320, "total_steps": 1548, "loss": 0.5725, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.697262412165261e-06, "epoch": 2.55, "percentage": 85.27, "elapsed_time": "1:04:17", "remaining_time": "0:11:06"}
|
| 133 |
+
{"current_steps": 1330, "total_steps": 1548, "loss": 0.5334, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.4726344402216662e-06, "epoch": 2.57, "percentage": 85.92, "elapsed_time": "1:04:46", "remaining_time": "0:10:37"}
|
| 134 |
+
{"current_steps": 1340, "total_steps": 1548, "loss": 0.4746, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.2572844461675902e-06, "epoch": 2.59, "percentage": 86.56, "elapsed_time": "1:05:15", "remaining_time": "0:10:07"}
|
| 135 |
+
{"current_steps": 1350, "total_steps": 1548, "loss": 0.5849, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.051301122685634e-06, "epoch": 2.61, "percentage": 87.21, "elapsed_time": "1:05:45", "remaining_time": "0:09:38"}
|
| 136 |
+
{"current_steps": 1360, "total_steps": 1548, "loss": 0.5648, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.8547693047608588e-06, "epoch": 2.63, "percentage": 87.86, "elapsed_time": "1:06:14", "remaining_time": "0:09:09"}
|
| 137 |
+
{"current_steps": 1370, "total_steps": 1548, "loss": 0.5596, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.6677699347412035e-06, "epoch": 2.65, "percentage": 88.5, "elapsed_time": "1:06:43", "remaining_time": "0:08:40"}
|
| 138 |
+
{"current_steps": 1380, "total_steps": 1548, "loss": 0.508, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.4903800290010817e-06, "epoch": 2.67, "percentage": 89.15, "elapsed_time": "1:07:12", "remaining_time": "0:08:10"}
|
| 139 |
+
{"current_steps": 1390, "total_steps": 1548, "loss": 0.5818, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.3226726462218897e-06, "epoch": 2.69, "percentage": 89.79, "elapsed_time": "1:07:41", "remaining_time": "0:07:41"}
|
| 140 |
+
{"current_steps": 1400, "total_steps": 1548, "loss": 0.5936, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.1647168573025474e-06, "epoch": 2.71, "percentage": 90.44, "elapsed_time": "1:08:11", "remaining_time": "0:07:12"}
|
| 141 |
+
{"current_steps": 1410, "total_steps": 1548, "loss": 0.5606, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.0165777169123703e-06, "epoch": 2.73, "percentage": 91.09, "elapsed_time": "1:08:40", "remaining_time": "0:06:43"}
|
| 142 |
+
{"current_steps": 1420, "total_steps": 1548, "loss": 0.6648, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 8.783162366980763e-07, "epoch": 2.75, "percentage": 91.73, "elapsed_time": "1:09:09", "remaining_time": "0:06:14"}
|
| 143 |
+
{"current_steps": 1430, "total_steps": 1548, "loss": 0.6249, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 7.499893601559255e-07, "epoch": 2.77, "percentage": 92.38, "elapsed_time": "1:09:38", "remaining_time": "0:05:44"}
|
| 144 |
+
{"current_steps": 1440, "total_steps": 1548, "loss": 0.5482, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 6.316499391793212e-07, "epoch": 2.79, "percentage": 93.02, "elapsed_time": "1:10:08", "remaining_time": "0:05:15"}
|
| 145 |
+
{"current_steps": 1450, "total_steps": 1548, "loss": 0.4884, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 5.233467122915642e-07, "epoch": 2.81, "percentage": 93.67, "elapsed_time": "1:10:37", "remaining_time": "0:04:46"}
|
| 146 |
+
{"current_steps": 1460, "total_steps": 1548, "loss": 0.6196, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.2512428457271435e-07, "epoch": 2.83, "percentage": 94.32, "elapsed_time": "1:11:06", "remaining_time": "0:04:17"}
|
| 147 |
+
{"current_steps": 1470, "total_steps": 1548, "loss": 0.5377, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 3.370231092888365e-07, "epoch": 2.84, "percentage": 94.96, "elapsed_time": "1:11:35", "remaining_time": "0:03:47"}
|
| 148 |
+
{"current_steps": 1480, "total_steps": 1548, "loss": 0.5291, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.590794712311606e-07, "epoch": 2.86, "percentage": 95.61, "elapsed_time": "1:12:04", "remaining_time": "0:03:18"}
|
| 149 |
+
{"current_steps": 1490, "total_steps": 1548, "loss": 0.5918, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.913254717720664e-07, "epoch": 2.88, "percentage": 96.25, "elapsed_time": "1:12:34", "remaining_time": "0:02:49"}
|
| 150 |
+
{"current_steps": 1500, "total_steps": 1548, "loss": 0.5625, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 1.3378901564400636e-07, "epoch": 2.9, "percentage": 96.9, "elapsed_time": "1:13:03", "remaining_time": "0:02:20"}
|
| 151 |
+
{"current_steps": 1510, "total_steps": 1548, "loss": 0.4924, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 8.649379944685732e-08, "epoch": 2.92, "percentage": 97.55, "elapsed_time": "1:13:32", "remaining_time": "0:01:51"}
|
| 152 |
+
{"current_steps": 1520, "total_steps": 1548, "loss": 0.5611, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 4.9459301888366004e-08, "epoch": 2.94, "percentage": 98.19, "elapsed_time": "1:14:01", "remaining_time": "0:01:21"}
|
| 153 |
+
{"current_steps": 1530, "total_steps": 1548, "loss": 0.5466, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 2.2700775761791416e-08, "epoch": 2.96, "percentage": 98.84, "elapsed_time": "1:14:30", "remaining_time": "0:00:52"}
|
| 154 |
+
{"current_steps": 1540, "total_steps": 1548, "loss": 0.5867, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": 6.229241663974206e-09, "epoch": 2.98, "percentage": 99.48, "elapsed_time": "1:15:00", "remaining_time": "0:00:23"}
|
| 155 |
+
{"current_steps": 1548, "total_steps": 1548, "loss": null, "eval_loss": null, "predict_loss": null, "reward": null, "learning_rate": null, "epoch": 3.0, "percentage": 100.0, "elapsed_time": "1:15:23", "remaining_time": "0:00:00"}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,952 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.9956458635703918,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1548,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.02,
|
| 13 |
+
"learning_rate": 4.999747735219333e-05,
|
| 14 |
+
"loss": 1.6087,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.04,
|
| 19 |
+
"learning_rate": 4.998512275175808e-05,
|
| 20 |
+
"loss": 1.4212,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.06,
|
| 25 |
+
"learning_rate": 4.9962477937145644e-05,
|
| 26 |
+
"loss": 1.0602,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.08,
|
| 31 |
+
"learning_rate": 4.992955223470575e-05,
|
| 32 |
+
"loss": 1.0232,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.1,
|
| 37 |
+
"learning_rate": 4.9886359205009334e-05,
|
| 38 |
+
"loss": 1.0354,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.12,
|
| 43 |
+
"learning_rate": 4.9832916637263665e-05,
|
| 44 |
+
"loss": 0.8403,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.14,
|
| 49 |
+
"learning_rate": 4.976924654198569e-05,
|
| 50 |
+
"loss": 0.7628,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.15,
|
| 55 |
+
"learning_rate": 4.9695375141937e-05,
|
| 56 |
+
"loss": 0.7421,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.17,
|
| 61 |
+
"learning_rate": 4.9611332861323875e-05,
|
| 62 |
+
"loss": 0.7773,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.19,
|
| 67 |
+
"learning_rate": 4.95171543132669e-05,
|
| 68 |
+
"loss": 0.7814,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.21,
|
| 73 |
+
"learning_rate": 4.941287828554553e-05,
|
| 74 |
+
"loss": 0.7798,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.23,
|
| 79 |
+
"learning_rate": 4.929854772462312e-05,
|
| 80 |
+
"loss": 0.7717,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.25,
|
| 85 |
+
"learning_rate": 4.9174209717959294e-05,
|
| 86 |
+
"loss": 0.8173,
|
| 87 |
+
"step": 130
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.27,
|
| 91 |
+
"learning_rate": 4.9039915474616805e-05,
|
| 92 |
+
"loss": 0.8009,
|
| 93 |
+
"step": 140
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.29,
|
| 97 |
+
"learning_rate": 4.889572030417091e-05,
|
| 98 |
+
"loss": 0.7499,
|
| 99 |
+
"step": 150
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.31,
|
| 103 |
+
"learning_rate": 4.874168359392987e-05,
|
| 104 |
+
"loss": 0.6257,
|
| 105 |
+
"step": 160
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.33,
|
| 109 |
+
"learning_rate": 4.857786878447612e-05,
|
| 110 |
+
"loss": 0.7029,
|
| 111 |
+
"step": 170
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.35,
|
| 115 |
+
"learning_rate": 4.8404343343538014e-05,
|
| 116 |
+
"loss": 0.6505,
|
| 117 |
+
"step": 180
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.37,
|
| 121 |
+
"learning_rate": 4.822117873820301e-05,
|
| 122 |
+
"loss": 0.6647,
|
| 123 |
+
"step": 190
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.39,
|
| 127 |
+
"learning_rate": 4.802845040548363e-05,
|
| 128 |
+
"loss": 0.6787,
|
| 129 |
+
"step": 200
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.41,
|
| 133 |
+
"learning_rate": 4.782623772124855e-05,
|
| 134 |
+
"loss": 0.5937,
|
| 135 |
+
"step": 210
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.43,
|
| 139 |
+
"learning_rate": 4.7614623967531244e-05,
|
| 140 |
+
"loss": 0.649,
|
| 141 |
+
"step": 220
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.45,
|
| 145 |
+
"learning_rate": 4.7393696298230084e-05,
|
| 146 |
+
"loss": 0.729,
|
| 147 |
+
"step": 230
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.46,
|
| 151 |
+
"learning_rate": 4.716354570321361e-05,
|
| 152 |
+
"loss": 0.6169,
|
| 153 |
+
"step": 240
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.48,
|
| 157 |
+
"learning_rate": 4.692426697084605e-05,
|
| 158 |
+
"loss": 0.6484,
|
| 159 |
+
"step": 250
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.5,
|
| 163 |
+
"learning_rate": 4.6675958648948394e-05,
|
| 164 |
+
"loss": 0.6714,
|
| 165 |
+
"step": 260
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.52,
|
| 169 |
+
"learning_rate": 4.6418723004211075e-05,
|
| 170 |
+
"loss": 0.6608,
|
| 171 |
+
"step": 270
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.54,
|
| 175 |
+
"learning_rate": 4.615266598007512e-05,
|
| 176 |
+
"loss": 0.6982,
|
| 177 |
+
"step": 280
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.56,
|
| 181 |
+
"learning_rate": 4.587789715309888e-05,
|
| 182 |
+
"loss": 0.6304,
|
| 183 |
+
"step": 290
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.58,
|
| 187 |
+
"learning_rate": 4.559452968782861e-05,
|
| 188 |
+
"loss": 0.6261,
|
| 189 |
+
"step": 300
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.6,
|
| 193 |
+
"learning_rate": 4.530268029019117e-05,
|
| 194 |
+
"loss": 0.6385,
|
| 195 |
+
"step": 310
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.62,
|
| 199 |
+
"learning_rate": 4.500246915942827e-05,
|
| 200 |
+
"loss": 0.6703,
|
| 201 |
+
"step": 320
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.64,
|
| 205 |
+
"learning_rate": 4.469401993859201e-05,
|
| 206 |
+
"loss": 0.62,
|
| 207 |
+
"step": 330
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.66,
|
| 211 |
+
"learning_rate": 4.437745966362201e-05,
|
| 212 |
+
"loss": 0.7172,
|
| 213 |
+
"step": 340
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.68,
|
| 217 |
+
"learning_rate": 4.4052918711025194e-05,
|
| 218 |
+
"loss": 0.5989,
|
| 219 |
+
"step": 350
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.7,
|
| 223 |
+
"learning_rate": 4.372053074417975e-05,
|
| 224 |
+
"loss": 0.6586,
|
| 225 |
+
"step": 360
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.72,
|
| 229 |
+
"learning_rate": 4.3380432658285367e-05,
|
| 230 |
+
"loss": 0.6459,
|
| 231 |
+
"step": 370
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.74,
|
| 235 |
+
"learning_rate": 4.3032764523982496e-05,
|
| 236 |
+
"loss": 0.6918,
|
| 237 |
+
"step": 380
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.75,
|
| 241 |
+
"learning_rate": 4.267766952966369e-05,
|
| 242 |
+
"loss": 0.6366,
|
| 243 |
+
"step": 390
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.77,
|
| 247 |
+
"learning_rate": 4.231529392250095e-05,
|
| 248 |
+
"loss": 0.6692,
|
| 249 |
+
"step": 400
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.79,
|
| 253 |
+
"learning_rate": 4.194578694821332e-05,
|
| 254 |
+
"loss": 0.715,
|
| 255 |
+
"step": 410
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.81,
|
| 259 |
+
"learning_rate": 4.156930078959946e-05,
|
| 260 |
+
"loss": 0.6242,
|
| 261 |
+
"step": 420
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.83,
|
| 265 |
+
"learning_rate": 4.1185990503860626e-05,
|
| 266 |
+
"loss": 0.7372,
|
| 267 |
+
"step": 430
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.85,
|
| 271 |
+
"learning_rate": 4.079601395873979e-05,
|
| 272 |
+
"loss": 0.6203,
|
| 273 |
+
"step": 440
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.87,
|
| 277 |
+
"learning_rate": 4.0399531767503204e-05,
|
| 278 |
+
"loss": 0.7101,
|
| 279 |
+
"step": 450
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.89,
|
| 283 |
+
"learning_rate": 3.999670722279131e-05,
|
| 284 |
+
"loss": 0.6866,
|
| 285 |
+
"step": 460
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.91,
|
| 289 |
+
"learning_rate": 3.958770622936599e-05,
|
| 290 |
+
"loss": 0.5917,
|
| 291 |
+
"step": 470
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.93,
|
| 295 |
+
"learning_rate": 3.917269723578212e-05,
|
| 296 |
+
"loss": 0.664,
|
| 297 |
+
"step": 480
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.95,
|
| 301 |
+
"learning_rate": 3.8751851165011413e-05,
|
| 302 |
+
"loss": 0.6763,
|
| 303 |
+
"step": 490
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.97,
|
| 307 |
+
"learning_rate": 3.8325341344047174e-05,
|
| 308 |
+
"loss": 0.615,
|
| 309 |
+
"step": 500
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.99,
|
| 313 |
+
"learning_rate": 3.7893343432518946e-05,
|
| 314 |
+
"loss": 0.5707,
|
| 315 |
+
"step": 510
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 1.01,
|
| 319 |
+
"learning_rate": 3.745603535034641e-05,
|
| 320 |
+
"loss": 0.6939,
|
| 321 |
+
"step": 520
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 1.03,
|
| 325 |
+
"learning_rate": 3.701359720446249e-05,
|
| 326 |
+
"loss": 0.6863,
|
| 327 |
+
"step": 530
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 1.04,
|
| 331 |
+
"learning_rate": 3.656621121463557e-05,
|
| 332 |
+
"loss": 0.6671,
|
| 333 |
+
"step": 540
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 1.06,
|
| 337 |
+
"learning_rate": 3.611406163842168e-05,
|
| 338 |
+
"loss": 0.518,
|
| 339 |
+
"step": 550
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.08,
|
| 343 |
+
"learning_rate": 3.565733469527731e-05,
|
| 344 |
+
"loss": 0.6747,
|
| 345 |
+
"step": 560
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.1,
|
| 349 |
+
"learning_rate": 3.519621848986428e-05,
|
| 350 |
+
"loss": 0.5754,
|
| 351 |
+
"step": 570
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 1.12,
|
| 355 |
+
"learning_rate": 3.473090293457811e-05,
|
| 356 |
+
"loss": 0.6061,
|
| 357 |
+
"step": 580
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 1.14,
|
| 361 |
+
"learning_rate": 3.426157967133192e-05,
|
| 362 |
+
"loss": 0.5859,
|
| 363 |
+
"step": 590
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 1.16,
|
| 367 |
+
"learning_rate": 3.3788441992628026e-05,
|
| 368 |
+
"loss": 0.546,
|
| 369 |
+
"step": 600
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.18,
|
| 373 |
+
"learning_rate": 3.331168476194969e-05,
|
| 374 |
+
"loss": 0.6333,
|
| 375 |
+
"step": 610
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 1.2,
|
| 379 |
+
"learning_rate": 3.283150433350589e-05,
|
| 380 |
+
"loss": 0.6332,
|
| 381 |
+
"step": 620
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 1.22,
|
| 385 |
+
"learning_rate": 3.234809847136213e-05,
|
| 386 |
+
"loss": 0.5709,
|
| 387 |
+
"step": 630
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.24,
|
| 391 |
+
"learning_rate": 3.1861666267990566e-05,
|
| 392 |
+
"loss": 0.578,
|
| 393 |
+
"step": 640
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 1.26,
|
| 397 |
+
"learning_rate": 3.137240806227306e-05,
|
| 398 |
+
"loss": 0.537,
|
| 399 |
+
"step": 650
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 1.28,
|
| 403 |
+
"learning_rate": 3.08805253569909e-05,
|
| 404 |
+
"loss": 0.5905,
|
| 405 |
+
"step": 660
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.3,
|
| 409 |
+
"learning_rate": 3.038622073583507e-05,
|
| 410 |
+
"loss": 0.6015,
|
| 411 |
+
"step": 670
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 1.32,
|
| 415 |
+
"learning_rate": 2.9889697779971504e-05,
|
| 416 |
+
"loss": 0.5735,
|
| 417 |
+
"step": 680
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 1.34,
|
| 421 |
+
"learning_rate": 2.9391160984195382e-05,
|
| 422 |
+
"loss": 0.5791,
|
| 423 |
+
"step": 690
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.35,
|
| 427 |
+
"learning_rate": 2.8890815672709225e-05,
|
| 428 |
+
"loss": 0.5683,
|
| 429 |
+
"step": 700
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 1.37,
|
| 433 |
+
"learning_rate": 2.8388867914559347e-05,
|
| 434 |
+
"loss": 0.6583,
|
| 435 |
+
"step": 710
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 1.39,
|
| 439 |
+
"learning_rate": 2.7885524438765603e-05,
|
| 440 |
+
"loss": 0.6091,
|
| 441 |
+
"step": 720
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.41,
|
| 445 |
+
"learning_rate": 2.7380992549179235e-05,
|
| 446 |
+
"loss": 0.5614,
|
| 447 |
+
"step": 730
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 1.43,
|
| 451 |
+
"learning_rate": 2.6875480039104085e-05,
|
| 452 |
+
"loss": 0.6208,
|
| 453 |
+
"step": 740
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 1.45,
|
| 457 |
+
"learning_rate": 2.636919510571609e-05,
|
| 458 |
+
"loss": 0.5686,
|
| 459 |
+
"step": 750
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.47,
|
| 463 |
+
"learning_rate": 2.5862346264316605e-05,
|
| 464 |
+
"loss": 0.5565,
|
| 465 |
+
"step": 760
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.49,
|
| 469 |
+
"learning_rate": 2.5355142262454506e-05,
|
| 470 |
+
"loss": 0.5936,
|
| 471 |
+
"step": 770
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 1.51,
|
| 475 |
+
"learning_rate": 2.484779199395285e-05,
|
| 476 |
+
"loss": 0.6469,
|
| 477 |
+
"step": 780
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.53,
|
| 481 |
+
"learning_rate": 2.4340504412875113e-05,
|
| 482 |
+
"loss": 0.5551,
|
| 483 |
+
"step": 790
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 1.55,
|
| 487 |
+
"learning_rate": 2.3833488447466746e-05,
|
| 488 |
+
"loss": 0.6454,
|
| 489 |
+
"step": 800
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.57,
|
| 493 |
+
"learning_rate": 2.3326952914107268e-05,
|
| 494 |
+
"loss": 0.61,
|
| 495 |
+
"step": 810
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.59,
|
| 499 |
+
"learning_rate": 2.2821106431308544e-05,
|
| 500 |
+
"loss": 0.6171,
|
| 501 |
+
"step": 820
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.61,
|
| 505 |
+
"learning_rate": 2.2316157333794414e-05,
|
| 506 |
+
"loss": 0.5636,
|
| 507 |
+
"step": 830
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.63,
|
| 511 |
+
"learning_rate": 2.1812313586697307e-05,
|
| 512 |
+
"loss": 0.5094,
|
| 513 |
+
"step": 840
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.64,
|
| 517 |
+
"learning_rate": 2.1309782699907042e-05,
|
| 518 |
+
"loss": 0.6278,
|
| 519 |
+
"step": 850
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 1.66,
|
| 523 |
+
"learning_rate": 2.0808771642607146e-05,
|
| 524 |
+
"loss": 0.5556,
|
| 525 |
+
"step": 860
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 1.68,
|
| 529 |
+
"learning_rate": 2.0309486758033773e-05,
|
| 530 |
+
"loss": 0.5103,
|
| 531 |
+
"step": 870
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.7,
|
| 535 |
+
"learning_rate": 1.9812133678492554e-05,
|
| 536 |
+
"loss": 0.5002,
|
| 537 |
+
"step": 880
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 1.72,
|
| 541 |
+
"learning_rate": 1.9316917240668133e-05,
|
| 542 |
+
"loss": 0.626,
|
| 543 |
+
"step": 890
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.74,
|
| 547 |
+
"learning_rate": 1.8824041401261462e-05,
|
| 548 |
+
"loss": 0.5432,
|
| 549 |
+
"step": 900
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.76,
|
| 553 |
+
"learning_rate": 1.833370915298948e-05,
|
| 554 |
+
"loss": 0.5423,
|
| 555 |
+
"step": 910
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 1.78,
|
| 559 |
+
"learning_rate": 1.784612244098181e-05,
|
| 560 |
+
"loss": 0.5179,
|
| 561 |
+
"step": 920
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 1.8,
|
| 565 |
+
"learning_rate": 1.7361482079608914e-05,
|
| 566 |
+
"loss": 0.5541,
|
| 567 |
+
"step": 930
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.82,
|
| 571 |
+
"learning_rate": 1.687998766977597e-05,
|
| 572 |
+
"loss": 0.6554,
|
| 573 |
+
"step": 940
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 1.84,
|
| 577 |
+
"learning_rate": 1.6401837516716546e-05,
|
| 578 |
+
"loss": 0.601,
|
| 579 |
+
"step": 950
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 1.86,
|
| 583 |
+
"learning_rate": 1.5927228548319767e-05,
|
| 584 |
+
"loss": 0.5267,
|
| 585 |
+
"step": 960
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.88,
|
| 589 |
+
"learning_rate": 1.545635623402496e-05,
|
| 590 |
+
"loss": 0.5281,
|
| 591 |
+
"step": 970
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 1.9,
|
| 595 |
+
"learning_rate": 1.4989414504316748e-05,
|
| 596 |
+
"loss": 0.5823,
|
| 597 |
+
"step": 980
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 1.92,
|
| 601 |
+
"learning_rate": 1.4526595670854159e-05,
|
| 602 |
+
"loss": 0.5789,
|
| 603 |
+
"step": 990
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.94,
|
| 607 |
+
"learning_rate": 1.4068090347266311e-05,
|
| 608 |
+
"loss": 0.6653,
|
| 609 |
+
"step": 1000
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 1.95,
|
| 613 |
+
"learning_rate": 1.3614087370647479e-05,
|
| 614 |
+
"loss": 0.5199,
|
| 615 |
+
"step": 1010
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.97,
|
| 619 |
+
"learning_rate": 1.3164773723783918e-05,
|
| 620 |
+
"loss": 0.6317,
|
| 621 |
+
"step": 1020
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.99,
|
| 625 |
+
"learning_rate": 1.2720334458144235e-05,
|
| 626 |
+
"loss": 0.4948,
|
| 627 |
+
"step": 1030
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 2.01,
|
| 631 |
+
"learning_rate": 1.2280952617665334e-05,
|
| 632 |
+
"loss": 0.6401,
|
| 633 |
+
"step": 1040
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 2.03,
|
| 637 |
+
"learning_rate": 1.1846809163365052e-05,
|
| 638 |
+
"loss": 0.5579,
|
| 639 |
+
"step": 1050
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 2.05,
|
| 643 |
+
"learning_rate": 1.1418082898812721e-05,
|
| 644 |
+
"loss": 0.4654,
|
| 645 |
+
"step": 1060
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 2.07,
|
| 649 |
+
"learning_rate": 1.0994950396488275e-05,
|
| 650 |
+
"loss": 0.5219,
|
| 651 |
+
"step": 1070
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 2.09,
|
| 655 |
+
"learning_rate": 1.057758592506022e-05,
|
| 656 |
+
"loss": 0.6016,
|
| 657 |
+
"step": 1080
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 2.11,
|
| 661 |
+
"learning_rate": 1.0166161377612437e-05,
|
| 662 |
+
"loss": 0.515,
|
| 663 |
+
"step": 1090
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 2.13,
|
| 667 |
+
"learning_rate": 9.760846200849388e-06,
|
| 668 |
+
"loss": 0.5687,
|
| 669 |
+
"step": 1100
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.15,
|
| 673 |
+
"learning_rate": 9.361807325308861e-06,
|
| 674 |
+
"loss": 0.5587,
|
| 675 |
+
"step": 1110
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 2.17,
|
| 679 |
+
"learning_rate": 8.969209096611092e-06,
|
| 680 |
+
"loss": 0.5365,
|
| 681 |
+
"step": 1120
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 2.19,
|
| 685 |
+
"learning_rate": 8.5832132077723e-06,
|
| 686 |
+
"loss": 0.514,
|
| 687 |
+
"step": 1130
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 2.21,
|
| 691 |
+
"learning_rate": 8.203978632610915e-06,
|
| 692 |
+
"loss": 0.5671,
|
| 693 |
+
"step": 1140
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 2.23,
|
| 697 |
+
"learning_rate": 7.831661560273585e-06,
|
| 698 |
+
"loss": 0.5803,
|
| 699 |
+
"step": 1150
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 2.24,
|
| 703 |
+
"learning_rate": 7.466415330908147e-06,
|
| 704 |
+
"loss": 0.6003,
|
| 705 |
+
"step": 1160
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 2.26,
|
| 709 |
+
"learning_rate": 7.108390372509893e-06,
|
| 710 |
+
"loss": 0.59,
|
| 711 |
+
"step": 1170
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 2.28,
|
| 715 |
+
"learning_rate": 6.757734138967248e-06,
|
| 716 |
+
"loss": 0.6142,
|
| 717 |
+
"step": 1180
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 2.3,
|
| 721 |
+
"learning_rate": 6.414591049332366e-06,
|
| 722 |
+
"loss": 0.5202,
|
| 723 |
+
"step": 1190
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 2.32,
|
| 727 |
+
"learning_rate": 6.079102428341588e-06,
|
| 728 |
+
"loss": 0.5333,
|
| 729 |
+
"step": 1200
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 2.34,
|
| 733 |
+
"learning_rate": 5.7514064482104e-06,
|
| 734 |
+
"loss": 0.5676,
|
| 735 |
+
"step": 1210
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.36,
|
| 739 |
+
"learning_rate": 5.431638071726602e-06,
|
| 740 |
+
"loss": 0.5741,
|
| 741 |
+
"step": 1220
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 2.38,
|
| 745 |
+
"learning_rate": 5.11992899666546e-06,
|
| 746 |
+
"loss": 0.5904,
|
| 747 |
+
"step": 1230
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 2.4,
|
| 751 |
+
"learning_rate": 4.8164076015494695e-06,
|
| 752 |
+
"loss": 0.5816,
|
| 753 |
+
"step": 1240
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 2.42,
|
| 757 |
+
"learning_rate": 4.521198892775203e-06,
|
| 758 |
+
"loss": 0.5774,
|
| 759 |
+
"step": 1250
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 2.44,
|
| 763 |
+
"learning_rate": 4.234424453128974e-06,
|
| 764 |
+
"loss": 0.5844,
|
| 765 |
+
"step": 1260
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.46,
|
| 769 |
+
"learning_rate": 3.9562023917124905e-06,
|
| 770 |
+
"loss": 0.5407,
|
| 771 |
+
"step": 1270
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 2.48,
|
| 775 |
+
"learning_rate": 3.6866472952992226e-06,
|
| 776 |
+
"loss": 0.5709,
|
| 777 |
+
"step": 1280
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 2.5,
|
| 781 |
+
"learning_rate": 3.425870181141394e-06,
|
| 782 |
+
"loss": 0.5139,
|
| 783 |
+
"step": 1290
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 2.52,
|
| 787 |
+
"learning_rate": 3.173978451247153e-06,
|
| 788 |
+
"loss": 0.5911,
|
| 789 |
+
"step": 1300
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 2.54,
|
| 793 |
+
"learning_rate": 2.931075848146647e-06,
|
| 794 |
+
"loss": 0.5552,
|
| 795 |
+
"step": 1310
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 2.55,
|
| 799 |
+
"learning_rate": 2.697262412165261e-06,
|
| 800 |
+
"loss": 0.5725,
|
| 801 |
+
"step": 1320
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 2.57,
|
| 805 |
+
"learning_rate": 2.4726344402216662e-06,
|
| 806 |
+
"loss": 0.5334,
|
| 807 |
+
"step": 1330
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 2.59,
|
| 811 |
+
"learning_rate": 2.2572844461675902e-06,
|
| 812 |
+
"loss": 0.4746,
|
| 813 |
+
"step": 1340
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 2.61,
|
| 817 |
+
"learning_rate": 2.051301122685634e-06,
|
| 818 |
+
"loss": 0.5849,
|
| 819 |
+
"step": 1350
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.63,
|
| 823 |
+
"learning_rate": 1.8547693047608588e-06,
|
| 824 |
+
"loss": 0.5648,
|
| 825 |
+
"step": 1360
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 2.65,
|
| 829 |
+
"learning_rate": 1.6677699347412035e-06,
|
| 830 |
+
"loss": 0.5596,
|
| 831 |
+
"step": 1370
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 2.67,
|
| 835 |
+
"learning_rate": 1.4903800290010817e-06,
|
| 836 |
+
"loss": 0.508,
|
| 837 |
+
"step": 1380
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.69,
|
| 841 |
+
"learning_rate": 1.3226726462218897e-06,
|
| 842 |
+
"loss": 0.5818,
|
| 843 |
+
"step": 1390
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 2.71,
|
| 847 |
+
"learning_rate": 1.1647168573025474e-06,
|
| 848 |
+
"loss": 0.5936,
|
| 849 |
+
"step": 1400
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 2.73,
|
| 853 |
+
"learning_rate": 1.0165777169123703e-06,
|
| 854 |
+
"loss": 0.5606,
|
| 855 |
+
"step": 1410
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.75,
|
| 859 |
+
"learning_rate": 8.783162366980763e-07,
|
| 860 |
+
"loss": 0.6648,
|
| 861 |
+
"step": 1420
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 2.77,
|
| 865 |
+
"learning_rate": 7.499893601559255e-07,
|
| 866 |
+
"loss": 0.6249,
|
| 867 |
+
"step": 1430
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 2.79,
|
| 871 |
+
"learning_rate": 6.316499391793212e-07,
|
| 872 |
+
"loss": 0.5482,
|
| 873 |
+
"step": 1440
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.81,
|
| 877 |
+
"learning_rate": 5.233467122915642e-07,
|
| 878 |
+
"loss": 0.4884,
|
| 879 |
+
"step": 1450
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 2.83,
|
| 883 |
+
"learning_rate": 4.2512428457271435e-07,
|
| 884 |
+
"loss": 0.6196,
|
| 885 |
+
"step": 1460
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 2.84,
|
| 889 |
+
"learning_rate": 3.370231092888365e-07,
|
| 890 |
+
"loss": 0.5377,
|
| 891 |
+
"step": 1470
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.86,
|
| 895 |
+
"learning_rate": 2.590794712311606e-07,
|
| 896 |
+
"loss": 0.5291,
|
| 897 |
+
"step": 1480
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 2.88,
|
| 901 |
+
"learning_rate": 1.913254717720664e-07,
|
| 902 |
+
"loss": 0.5918,
|
| 903 |
+
"step": 1490
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 2.9,
|
| 907 |
+
"learning_rate": 1.3378901564400636e-07,
|
| 908 |
+
"loss": 0.5625,
|
| 909 |
+
"step": 1500
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.92,
|
| 913 |
+
"learning_rate": 8.649379944685732e-08,
|
| 914 |
+
"loss": 0.4924,
|
| 915 |
+
"step": 1510
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 2.94,
|
| 919 |
+
"learning_rate": 4.9459301888366004e-08,
|
| 920 |
+
"loss": 0.5611,
|
| 921 |
+
"step": 1520
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 2.96,
|
| 925 |
+
"learning_rate": 2.2700775761791416e-08,
|
| 926 |
+
"loss": 0.5466,
|
| 927 |
+
"step": 1530
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.98,
|
| 931 |
+
"learning_rate": 6.229241663974206e-09,
|
| 932 |
+
"loss": 0.5867,
|
| 933 |
+
"step": 1540
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 3.0,
|
| 937 |
+
"step": 1548,
|
| 938 |
+
"total_flos": 5.030540427342643e+17,
|
| 939 |
+
"train_loss": 0.6278958536241713,
|
| 940 |
+
"train_runtime": 4523.4188,
|
| 941 |
+
"train_samples_per_second": 5.483,
|
| 942 |
+
"train_steps_per_second": 0.342
|
| 943 |
+
}
|
| 944 |
+
],
|
| 945 |
+
"logging_steps": 10,
|
| 946 |
+
"max_steps": 1548,
|
| 947 |
+
"num_train_epochs": 3,
|
| 948 |
+
"save_steps": 1000,
|
| 949 |
+
"total_flos": 5.030540427342643e+17,
|
| 950 |
+
"trial_name": null,
|
| 951 |
+
"trial_params": null
|
| 952 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18ab75398da7f09d2453a0b310cf7ec2cce62d39d751b7660e3ac874ed58af4e
|
| 3 |
+
size 3431
|
training_loss.png
ADDED
|