Initial commit
Browse files- adapter_config.json +19 -0
- adapter_model.bin +3 -0
- checkpoint-6800/optimizer.pt +3 -0
- checkpoint-6800/pytorch_model.bin +3 -0
- checkpoint-6800/rng_state_0.pth +3 -0
- checkpoint-6800/rng_state_1.pth +3 -0
- checkpoint-6800/scaler.pt +3 -0
- checkpoint-6800/scheduler.pt +3 -0
- checkpoint-6800/trainer_state.json +4368 -0
- checkpoint-6800/training_args.bin +3 -0
adapter_config.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_name_or_path": "models/HF/13B",
|
| 3 |
+
"bias": "none",
|
| 4 |
+
"enable_lora": null,
|
| 5 |
+
"fan_in_fan_out": false,
|
| 6 |
+
"inference_mode": true,
|
| 7 |
+
"init_lora_weights": true,
|
| 8 |
+
"lora_alpha": 16,
|
| 9 |
+
"lora_dropout": 0.05,
|
| 10 |
+
"merge_weights": false,
|
| 11 |
+
"modules_to_save": null,
|
| 12 |
+
"peft_type": "LORA",
|
| 13 |
+
"r": 8,
|
| 14 |
+
"target_modules": [
|
| 15 |
+
"q_proj",
|
| 16 |
+
"v_proj"
|
| 17 |
+
],
|
| 18 |
+
"task_type": "CAUSAL_LM"
|
| 19 |
+
}
|
adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c38f7bcd70b9a3c17444f1986792c6c4b08fab8f742fe038b3193254f0571a98
|
| 3 |
+
size 26271757
|
checkpoint-6800/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ebf2b174d5eca164b2d77471eb09eb6fa5b3ebd2b6ea5ce8fa9ae8e1d7c4272
|
| 3 |
+
size 52564997
|
checkpoint-6800/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:90a9119b27e400a0c131e06aa9288c4862ed7f867267b89b6026010a94082c36
|
| 3 |
+
size 26271757
|
checkpoint-6800/rng_state_0.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd14cfdfcf52ce7a2011c09b24954f3a3fed44a81c2d316260bee0058db44d81
|
| 3 |
+
size 14583
|
checkpoint-6800/rng_state_1.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c22e58698a8f013ada7b3ae20f600228bf2e93adf556bd1fe828642e46fd1434
|
| 3 |
+
size 14583
|
checkpoint-6800/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4c728ca09663d7c80b05c36b62dd781693daeb288dc44e59ee59dd4a735e9859
|
| 3 |
+
size 557
|
checkpoint-6800/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:495d435d9d4f076fa037def2c9abdbb0b90ef189d10bd190b532c3a27512d972
|
| 3 |
+
size 627
|
checkpoint-6800/trainer_state.json
ADDED
|
@@ -0,0 +1,4368 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.7945918440818787,
|
| 3 |
+
"best_model_checkpoint": "train.out/checkpoint-6800",
|
| 4 |
+
"epoch": 1.8709588664190397,
|
| 5 |
+
"global_step": 6800,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.01,
|
| 12 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 13 |
+
"loss": 1.7549,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.01,
|
| 18 |
+
"learning_rate": 4.000000000000001e-06,
|
| 19 |
+
"loss": 1.7891,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.02,
|
| 24 |
+
"learning_rate": 6e-06,
|
| 25 |
+
"loss": 1.7738,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.02,
|
| 30 |
+
"learning_rate": 8.000000000000001e-06,
|
| 31 |
+
"loss": 1.767,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.03,
|
| 36 |
+
"learning_rate": 1e-05,
|
| 37 |
+
"loss": 1.764,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.04,
|
| 42 |
+
"learning_rate": 1.2e-05,
|
| 43 |
+
"loss": 1.7286,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.04,
|
| 48 |
+
"learning_rate": 1.4e-05,
|
| 49 |
+
"loss": 1.68,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.05,
|
| 54 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 55 |
+
"loss": 1.5793,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.06,
|
| 60 |
+
"learning_rate": 1.8e-05,
|
| 61 |
+
"loss": 1.4831,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.06,
|
| 66 |
+
"learning_rate": 2e-05,
|
| 67 |
+
"loss": 1.3956,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.07,
|
| 72 |
+
"learning_rate": 1.998750780762024e-05,
|
| 73 |
+
"loss": 1.2915,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.07,
|
| 78 |
+
"learning_rate": 1.9975015615240476e-05,
|
| 79 |
+
"loss": 1.2065,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.08,
|
| 84 |
+
"learning_rate": 1.9962523422860715e-05,
|
| 85 |
+
"loss": 1.1471,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.09,
|
| 90 |
+
"learning_rate": 1.995003123048095e-05,
|
| 91 |
+
"loss": 1.0787,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.09,
|
| 96 |
+
"learning_rate": 1.993753903810119e-05,
|
| 97 |
+
"loss": 0.9869,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.1,
|
| 102 |
+
"learning_rate": 1.9925046845721424e-05,
|
| 103 |
+
"loss": 0.9633,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.11,
|
| 108 |
+
"learning_rate": 1.9912554653341663e-05,
|
| 109 |
+
"loss": 0.9448,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.11,
|
| 114 |
+
"learning_rate": 1.9900062460961898e-05,
|
| 115 |
+
"loss": 0.9208,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.12,
|
| 120 |
+
"learning_rate": 1.9887570268582137e-05,
|
| 121 |
+
"loss": 0.9104,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.12,
|
| 126 |
+
"learning_rate": 1.9875078076202376e-05,
|
| 127 |
+
"loss": 0.9109,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.12,
|
| 132 |
+
"eval_loss": 0.8898332118988037,
|
| 133 |
+
"eval_runtime": 90.2434,
|
| 134 |
+
"eval_samples_per_second": 11.081,
|
| 135 |
+
"eval_steps_per_second": 0.698,
|
| 136 |
+
"step": 200
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.13,
|
| 140 |
+
"learning_rate": 1.986258588382261e-05,
|
| 141 |
+
"loss": 0.9134,
|
| 142 |
+
"step": 210
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.14,
|
| 146 |
+
"learning_rate": 1.985009369144285e-05,
|
| 147 |
+
"loss": 0.9016,
|
| 148 |
+
"step": 220
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 0.14,
|
| 152 |
+
"learning_rate": 1.983760149906309e-05,
|
| 153 |
+
"loss": 0.8977,
|
| 154 |
+
"step": 230
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.15,
|
| 158 |
+
"learning_rate": 1.9825109306683324e-05,
|
| 159 |
+
"loss": 0.8835,
|
| 160 |
+
"step": 240
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.16,
|
| 164 |
+
"learning_rate": 1.9812617114303563e-05,
|
| 165 |
+
"loss": 0.8859,
|
| 166 |
+
"step": 250
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.16,
|
| 170 |
+
"learning_rate": 1.9800124921923802e-05,
|
| 171 |
+
"loss": 0.8884,
|
| 172 |
+
"step": 260
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.17,
|
| 176 |
+
"learning_rate": 1.9787632729544037e-05,
|
| 177 |
+
"loss": 0.8829,
|
| 178 |
+
"step": 270
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.17,
|
| 182 |
+
"learning_rate": 1.9775140537164276e-05,
|
| 183 |
+
"loss": 0.8905,
|
| 184 |
+
"step": 280
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.18,
|
| 188 |
+
"learning_rate": 1.976264834478451e-05,
|
| 189 |
+
"loss": 0.8824,
|
| 190 |
+
"step": 290
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.19,
|
| 194 |
+
"learning_rate": 1.975015615240475e-05,
|
| 195 |
+
"loss": 0.8844,
|
| 196 |
+
"step": 300
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 0.19,
|
| 200 |
+
"learning_rate": 1.9737663960024985e-05,
|
| 201 |
+
"loss": 0.8817,
|
| 202 |
+
"step": 310
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.2,
|
| 206 |
+
"learning_rate": 1.9725171767645224e-05,
|
| 207 |
+
"loss": 0.8917,
|
| 208 |
+
"step": 320
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.2,
|
| 212 |
+
"learning_rate": 1.971267957526546e-05,
|
| 213 |
+
"loss": 0.8823,
|
| 214 |
+
"step": 330
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.21,
|
| 218 |
+
"learning_rate": 1.97001873828857e-05,
|
| 219 |
+
"loss": 0.8832,
|
| 220 |
+
"step": 340
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.22,
|
| 224 |
+
"learning_rate": 1.9687695190505937e-05,
|
| 225 |
+
"loss": 0.8743,
|
| 226 |
+
"step": 350
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.22,
|
| 230 |
+
"learning_rate": 1.9675202998126173e-05,
|
| 231 |
+
"loss": 0.8676,
|
| 232 |
+
"step": 360
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 0.23,
|
| 236 |
+
"learning_rate": 1.966271080574641e-05,
|
| 237 |
+
"loss": 0.877,
|
| 238 |
+
"step": 370
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 0.24,
|
| 242 |
+
"learning_rate": 1.9650218613366647e-05,
|
| 243 |
+
"loss": 0.8779,
|
| 244 |
+
"step": 380
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.24,
|
| 248 |
+
"learning_rate": 1.9637726420986885e-05,
|
| 249 |
+
"loss": 0.8703,
|
| 250 |
+
"step": 390
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.25,
|
| 254 |
+
"learning_rate": 1.962523422860712e-05,
|
| 255 |
+
"loss": 0.8693,
|
| 256 |
+
"step": 400
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 0.25,
|
| 260 |
+
"eval_loss": 0.8513386249542236,
|
| 261 |
+
"eval_runtime": 90.0667,
|
| 262 |
+
"eval_samples_per_second": 11.103,
|
| 263 |
+
"eval_steps_per_second": 0.699,
|
| 264 |
+
"step": 400
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.11,
|
| 268 |
+
"learning_rate": 1.942603221625625e-05,
|
| 269 |
+
"loss": 0.8775,
|
| 270 |
+
"step": 410
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 0.12,
|
| 274 |
+
"learning_rate": 1.9407517126458065e-05,
|
| 275 |
+
"loss": 0.8583,
|
| 276 |
+
"step": 420
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.12,
|
| 280 |
+
"learning_rate": 1.938900203665988e-05,
|
| 281 |
+
"loss": 0.8687,
|
| 282 |
+
"step": 430
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.12,
|
| 286 |
+
"learning_rate": 1.9370486946861694e-05,
|
| 287 |
+
"loss": 0.8633,
|
| 288 |
+
"step": 440
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.12,
|
| 292 |
+
"learning_rate": 1.9351971857063506e-05,
|
| 293 |
+
"loss": 0.8642,
|
| 294 |
+
"step": 450
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.13,
|
| 298 |
+
"learning_rate": 1.933345676726532e-05,
|
| 299 |
+
"loss": 0.8546,
|
| 300 |
+
"step": 460
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 0.13,
|
| 304 |
+
"learning_rate": 1.931494167746714e-05,
|
| 305 |
+
"loss": 0.87,
|
| 306 |
+
"step": 470
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.13,
|
| 310 |
+
"learning_rate": 1.9296426587668953e-05,
|
| 311 |
+
"loss": 0.8372,
|
| 312 |
+
"step": 480
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.13,
|
| 316 |
+
"learning_rate": 1.9277911497870768e-05,
|
| 317 |
+
"loss": 0.8701,
|
| 318 |
+
"step": 490
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.14,
|
| 322 |
+
"learning_rate": 1.9259396408072583e-05,
|
| 323 |
+
"loss": 0.8682,
|
| 324 |
+
"step": 500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.14,
|
| 328 |
+
"learning_rate": 1.9240881318274397e-05,
|
| 329 |
+
"loss": 0.867,
|
| 330 |
+
"step": 510
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.14,
|
| 334 |
+
"learning_rate": 1.9222366228476212e-05,
|
| 335 |
+
"loss": 0.8455,
|
| 336 |
+
"step": 520
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 0.15,
|
| 340 |
+
"learning_rate": 1.9203851138678023e-05,
|
| 341 |
+
"loss": 0.8479,
|
| 342 |
+
"step": 530
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.15,
|
| 346 |
+
"learning_rate": 1.9185336048879838e-05,
|
| 347 |
+
"loss": 0.8699,
|
| 348 |
+
"step": 540
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.15,
|
| 352 |
+
"learning_rate": 1.9166820959081653e-05,
|
| 353 |
+
"loss": 0.8575,
|
| 354 |
+
"step": 550
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.15,
|
| 358 |
+
"learning_rate": 1.9148305869283467e-05,
|
| 359 |
+
"loss": 0.8577,
|
| 360 |
+
"step": 560
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.16,
|
| 364 |
+
"learning_rate": 1.9129790779485282e-05,
|
| 365 |
+
"loss": 0.8661,
|
| 366 |
+
"step": 570
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.16,
|
| 370 |
+
"learning_rate": 1.9111275689687097e-05,
|
| 371 |
+
"loss": 0.8581,
|
| 372 |
+
"step": 580
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 0.16,
|
| 376 |
+
"learning_rate": 1.909276059988891e-05,
|
| 377 |
+
"loss": 0.8611,
|
| 378 |
+
"step": 590
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.17,
|
| 382 |
+
"learning_rate": 1.9074245510090726e-05,
|
| 383 |
+
"loss": 0.8427,
|
| 384 |
+
"step": 600
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.17,
|
| 388 |
+
"eval_loss": 0.8519569635391235,
|
| 389 |
+
"eval_runtime": 98.6826,
|
| 390 |
+
"eval_samples_per_second": 10.133,
|
| 391 |
+
"eval_steps_per_second": 0.638,
|
| 392 |
+
"step": 600
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.17,
|
| 396 |
+
"learning_rate": 1.905573042029254e-05,
|
| 397 |
+
"loss": 0.8583,
|
| 398 |
+
"step": 610
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.17,
|
| 402 |
+
"learning_rate": 1.9037215330494355e-05,
|
| 403 |
+
"loss": 0.8474,
|
| 404 |
+
"step": 620
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.17,
|
| 408 |
+
"learning_rate": 1.901870024069617e-05,
|
| 409 |
+
"loss": 0.8553,
|
| 410 |
+
"step": 630
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.18,
|
| 414 |
+
"learning_rate": 1.900018515089798e-05,
|
| 415 |
+
"loss": 0.8523,
|
| 416 |
+
"step": 640
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.18,
|
| 420 |
+
"learning_rate": 1.8981670061099796e-05,
|
| 421 |
+
"loss": 0.8431,
|
| 422 |
+
"step": 650
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.18,
|
| 426 |
+
"learning_rate": 1.896315497130161e-05,
|
| 427 |
+
"loss": 0.8567,
|
| 428 |
+
"step": 660
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.18,
|
| 432 |
+
"learning_rate": 1.8944639881503426e-05,
|
| 433 |
+
"loss": 0.846,
|
| 434 |
+
"step": 670
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.19,
|
| 438 |
+
"learning_rate": 1.892612479170524e-05,
|
| 439 |
+
"loss": 0.8591,
|
| 440 |
+
"step": 680
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.19,
|
| 444 |
+
"learning_rate": 1.8907609701907055e-05,
|
| 445 |
+
"loss": 0.8469,
|
| 446 |
+
"step": 690
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.19,
|
| 450 |
+
"learning_rate": 1.888909461210887e-05,
|
| 451 |
+
"loss": 0.8615,
|
| 452 |
+
"step": 700
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.2,
|
| 456 |
+
"learning_rate": 1.8870579522310684e-05,
|
| 457 |
+
"loss": 0.8413,
|
| 458 |
+
"step": 710
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.2,
|
| 462 |
+
"learning_rate": 1.88520644325125e-05,
|
| 463 |
+
"loss": 0.8422,
|
| 464 |
+
"step": 720
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.2,
|
| 468 |
+
"learning_rate": 1.8833549342714314e-05,
|
| 469 |
+
"loss": 0.8535,
|
| 470 |
+
"step": 730
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.2,
|
| 474 |
+
"learning_rate": 1.881503425291613e-05,
|
| 475 |
+
"loss": 0.8567,
|
| 476 |
+
"step": 740
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.21,
|
| 480 |
+
"learning_rate": 1.8796519163117943e-05,
|
| 481 |
+
"loss": 0.8578,
|
| 482 |
+
"step": 750
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.21,
|
| 486 |
+
"learning_rate": 1.8778004073319758e-05,
|
| 487 |
+
"loss": 0.8468,
|
| 488 |
+
"step": 760
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.21,
|
| 492 |
+
"learning_rate": 1.8759488983521572e-05,
|
| 493 |
+
"loss": 0.8469,
|
| 494 |
+
"step": 770
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.21,
|
| 498 |
+
"learning_rate": 1.8740973893723387e-05,
|
| 499 |
+
"loss": 0.8312,
|
| 500 |
+
"step": 780
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.22,
|
| 504 |
+
"learning_rate": 1.8722458803925202e-05,
|
| 505 |
+
"loss": 0.8466,
|
| 506 |
+
"step": 790
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.22,
|
| 510 |
+
"learning_rate": 1.8703943714127017e-05,
|
| 511 |
+
"loss": 0.8483,
|
| 512 |
+
"step": 800
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.22,
|
| 516 |
+
"eval_loss": 0.8436271548271179,
|
| 517 |
+
"eval_runtime": 98.5282,
|
| 518 |
+
"eval_samples_per_second": 10.149,
|
| 519 |
+
"eval_steps_per_second": 0.639,
|
| 520 |
+
"step": 800
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.22,
|
| 524 |
+
"learning_rate": 1.868542862432883e-05,
|
| 525 |
+
"loss": 0.8513,
|
| 526 |
+
"step": 810
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 0.23,
|
| 530 |
+
"learning_rate": 1.8666913534530643e-05,
|
| 531 |
+
"loss": 0.8467,
|
| 532 |
+
"step": 820
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 0.23,
|
| 536 |
+
"learning_rate": 1.8648398444732457e-05,
|
| 537 |
+
"loss": 0.8416,
|
| 538 |
+
"step": 830
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.23,
|
| 542 |
+
"learning_rate": 1.8629883354934272e-05,
|
| 543 |
+
"loss": 0.8398,
|
| 544 |
+
"step": 840
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 0.23,
|
| 548 |
+
"learning_rate": 1.8611368265136087e-05,
|
| 549 |
+
"loss": 0.8361,
|
| 550 |
+
"step": 850
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.24,
|
| 554 |
+
"learning_rate": 1.85928531753379e-05,
|
| 555 |
+
"loss": 0.8474,
|
| 556 |
+
"step": 860
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.24,
|
| 560 |
+
"learning_rate": 1.8574338085539716e-05,
|
| 561 |
+
"loss": 0.8403,
|
| 562 |
+
"step": 870
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.24,
|
| 566 |
+
"learning_rate": 1.855582299574153e-05,
|
| 567 |
+
"loss": 0.8489,
|
| 568 |
+
"step": 880
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.24,
|
| 572 |
+
"learning_rate": 1.8537307905943345e-05,
|
| 573 |
+
"loss": 0.8499,
|
| 574 |
+
"step": 890
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 0.25,
|
| 578 |
+
"learning_rate": 1.851879281614516e-05,
|
| 579 |
+
"loss": 0.8438,
|
| 580 |
+
"step": 900
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 0.25,
|
| 584 |
+
"learning_rate": 1.8500277726346975e-05,
|
| 585 |
+
"loss": 0.839,
|
| 586 |
+
"step": 910
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.25,
|
| 590 |
+
"learning_rate": 1.848176263654879e-05,
|
| 591 |
+
"loss": 0.8429,
|
| 592 |
+
"step": 920
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 0.26,
|
| 596 |
+
"learning_rate": 1.84632475467506e-05,
|
| 597 |
+
"loss": 0.8384,
|
| 598 |
+
"step": 930
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.26,
|
| 602 |
+
"learning_rate": 1.8444732456952415e-05,
|
| 603 |
+
"loss": 0.8446,
|
| 604 |
+
"step": 940
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.26,
|
| 608 |
+
"learning_rate": 1.8426217367154234e-05,
|
| 609 |
+
"loss": 0.8485,
|
| 610 |
+
"step": 950
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 0.26,
|
| 614 |
+
"learning_rate": 1.8407702277356048e-05,
|
| 615 |
+
"loss": 0.8364,
|
| 616 |
+
"step": 960
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 0.27,
|
| 620 |
+
"learning_rate": 1.8389187187557863e-05,
|
| 621 |
+
"loss": 0.8357,
|
| 622 |
+
"step": 970
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.27,
|
| 626 |
+
"learning_rate": 1.8370672097759678e-05,
|
| 627 |
+
"loss": 0.8379,
|
| 628 |
+
"step": 980
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 0.27,
|
| 632 |
+
"learning_rate": 1.8352157007961492e-05,
|
| 633 |
+
"loss": 0.8275,
|
| 634 |
+
"step": 990
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.28,
|
| 638 |
+
"learning_rate": 1.8333641918163304e-05,
|
| 639 |
+
"loss": 0.8491,
|
| 640 |
+
"step": 1000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.28,
|
| 644 |
+
"eval_loss": 0.8377317190170288,
|
| 645 |
+
"eval_runtime": 98.8443,
|
| 646 |
+
"eval_samples_per_second": 10.117,
|
| 647 |
+
"eval_steps_per_second": 0.637,
|
| 648 |
+
"step": 1000
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 0.28,
|
| 652 |
+
"learning_rate": 1.831512682836512e-05,
|
| 653 |
+
"loss": 0.8352,
|
| 654 |
+
"step": 1010
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.28,
|
| 658 |
+
"learning_rate": 1.8296611738566933e-05,
|
| 659 |
+
"loss": 0.8423,
|
| 660 |
+
"step": 1020
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.28,
|
| 664 |
+
"learning_rate": 1.8278096648768748e-05,
|
| 665 |
+
"loss": 0.8337,
|
| 666 |
+
"step": 1030
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 0.29,
|
| 670 |
+
"learning_rate": 1.8259581558970562e-05,
|
| 671 |
+
"loss": 0.8276,
|
| 672 |
+
"step": 1040
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.29,
|
| 676 |
+
"learning_rate": 1.8241066469172377e-05,
|
| 677 |
+
"loss": 0.8313,
|
| 678 |
+
"step": 1050
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.29,
|
| 682 |
+
"learning_rate": 1.8222551379374192e-05,
|
| 683 |
+
"loss": 0.8266,
|
| 684 |
+
"step": 1060
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.29,
|
| 688 |
+
"learning_rate": 1.8204036289576006e-05,
|
| 689 |
+
"loss": 0.838,
|
| 690 |
+
"step": 1070
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.3,
|
| 694 |
+
"learning_rate": 1.818552119977782e-05,
|
| 695 |
+
"loss": 0.8456,
|
| 696 |
+
"step": 1080
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.3,
|
| 700 |
+
"learning_rate": 1.8167006109979636e-05,
|
| 701 |
+
"loss": 0.8281,
|
| 702 |
+
"step": 1090
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.3,
|
| 706 |
+
"learning_rate": 1.814849102018145e-05,
|
| 707 |
+
"loss": 0.835,
|
| 708 |
+
"step": 1100
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 0.31,
|
| 712 |
+
"learning_rate": 1.8129975930383262e-05,
|
| 713 |
+
"loss": 0.84,
|
| 714 |
+
"step": 1110
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.31,
|
| 718 |
+
"learning_rate": 1.8111460840585077e-05,
|
| 719 |
+
"loss": 0.8323,
|
| 720 |
+
"step": 1120
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 0.31,
|
| 724 |
+
"learning_rate": 1.809294575078689e-05,
|
| 725 |
+
"loss": 0.8331,
|
| 726 |
+
"step": 1130
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 0.31,
|
| 730 |
+
"learning_rate": 1.8074430660988706e-05,
|
| 731 |
+
"loss": 0.8304,
|
| 732 |
+
"step": 1140
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 0.32,
|
| 736 |
+
"learning_rate": 1.805591557119052e-05,
|
| 737 |
+
"loss": 0.8329,
|
| 738 |
+
"step": 1150
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.32,
|
| 742 |
+
"learning_rate": 1.8037400481392335e-05,
|
| 743 |
+
"loss": 0.8291,
|
| 744 |
+
"step": 1160
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.32,
|
| 748 |
+
"learning_rate": 1.801888539159415e-05,
|
| 749 |
+
"loss": 0.8437,
|
| 750 |
+
"step": 1170
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 0.32,
|
| 754 |
+
"learning_rate": 1.8000370301795965e-05,
|
| 755 |
+
"loss": 0.825,
|
| 756 |
+
"step": 1180
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 0.33,
|
| 760 |
+
"learning_rate": 1.798185521199778e-05,
|
| 761 |
+
"loss": 0.828,
|
| 762 |
+
"step": 1190
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.33,
|
| 766 |
+
"learning_rate": 1.7963340122199594e-05,
|
| 767 |
+
"loss": 0.8342,
|
| 768 |
+
"step": 1200
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 0.33,
|
| 772 |
+
"eval_loss": 0.8326738476753235,
|
| 773 |
+
"eval_runtime": 98.7519,
|
| 774 |
+
"eval_samples_per_second": 10.126,
|
| 775 |
+
"eval_steps_per_second": 0.638,
|
| 776 |
+
"step": 1200
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.33,
|
| 780 |
+
"learning_rate": 1.794482503240141e-05,
|
| 781 |
+
"loss": 0.8252,
|
| 782 |
+
"step": 1210
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.34,
|
| 786 |
+
"learning_rate": 1.7926309942603223e-05,
|
| 787 |
+
"loss": 0.8227,
|
| 788 |
+
"step": 1220
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 0.34,
|
| 792 |
+
"learning_rate": 1.7907794852805038e-05,
|
| 793 |
+
"loss": 0.8296,
|
| 794 |
+
"step": 1230
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.34,
|
| 798 |
+
"learning_rate": 1.7889279763006853e-05,
|
| 799 |
+
"loss": 0.8429,
|
| 800 |
+
"step": 1240
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.34,
|
| 804 |
+
"learning_rate": 1.7870764673208668e-05,
|
| 805 |
+
"loss": 0.8257,
|
| 806 |
+
"step": 1250
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 0.35,
|
| 810 |
+
"learning_rate": 1.7852249583410482e-05,
|
| 811 |
+
"loss": 0.8267,
|
| 812 |
+
"step": 1260
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.35,
|
| 816 |
+
"learning_rate": 1.7833734493612297e-05,
|
| 817 |
+
"loss": 0.8424,
|
| 818 |
+
"step": 1270
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.35,
|
| 822 |
+
"learning_rate": 1.781521940381411e-05,
|
| 823 |
+
"loss": 0.8262,
|
| 824 |
+
"step": 1280
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.35,
|
| 828 |
+
"learning_rate": 1.7796704314015923e-05,
|
| 829 |
+
"loss": 0.8314,
|
| 830 |
+
"step": 1290
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.36,
|
| 834 |
+
"learning_rate": 1.7778189224217738e-05,
|
| 835 |
+
"loss": 0.8194,
|
| 836 |
+
"step": 1300
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.36,
|
| 840 |
+
"learning_rate": 1.7759674134419552e-05,
|
| 841 |
+
"loss": 0.8316,
|
| 842 |
+
"step": 1310
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.36,
|
| 846 |
+
"learning_rate": 1.7741159044621367e-05,
|
| 847 |
+
"loss": 0.8399,
|
| 848 |
+
"step": 1320
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 0.37,
|
| 852 |
+
"learning_rate": 1.7722643954823182e-05,
|
| 853 |
+
"loss": 0.8273,
|
| 854 |
+
"step": 1330
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.37,
|
| 858 |
+
"learning_rate": 1.7704128865024996e-05,
|
| 859 |
+
"loss": 0.8207,
|
| 860 |
+
"step": 1340
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.37,
|
| 864 |
+
"learning_rate": 1.768561377522681e-05,
|
| 865 |
+
"loss": 0.8346,
|
| 866 |
+
"step": 1350
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.37,
|
| 870 |
+
"learning_rate": 1.7667098685428626e-05,
|
| 871 |
+
"loss": 0.8356,
|
| 872 |
+
"step": 1360
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.38,
|
| 876 |
+
"learning_rate": 1.764858359563044e-05,
|
| 877 |
+
"loss": 0.8352,
|
| 878 |
+
"step": 1370
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.38,
|
| 882 |
+
"learning_rate": 1.7630068505832255e-05,
|
| 883 |
+
"loss": 0.8362,
|
| 884 |
+
"step": 1380
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.38,
|
| 888 |
+
"learning_rate": 1.761155341603407e-05,
|
| 889 |
+
"loss": 0.8355,
|
| 890 |
+
"step": 1390
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.39,
|
| 894 |
+
"learning_rate": 1.759303832623588e-05,
|
| 895 |
+
"loss": 0.8203,
|
| 896 |
+
"step": 1400
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 0.39,
|
| 900 |
+
"eval_loss": 0.8290709853172302,
|
| 901 |
+
"eval_runtime": 98.8118,
|
| 902 |
+
"eval_samples_per_second": 10.12,
|
| 903 |
+
"eval_steps_per_second": 0.638,
|
| 904 |
+
"step": 1400
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 0.39,
|
| 908 |
+
"learning_rate": 1.7574523236437696e-05,
|
| 909 |
+
"loss": 0.8267,
|
| 910 |
+
"step": 1410
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.39,
|
| 914 |
+
"learning_rate": 1.755600814663951e-05,
|
| 915 |
+
"loss": 0.814,
|
| 916 |
+
"step": 1420
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 0.39,
|
| 920 |
+
"learning_rate": 1.753749305684133e-05,
|
| 921 |
+
"loss": 0.8227,
|
| 922 |
+
"step": 1430
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 0.4,
|
| 926 |
+
"learning_rate": 1.7518977967043143e-05,
|
| 927 |
+
"loss": 0.8221,
|
| 928 |
+
"step": 1440
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 0.4,
|
| 932 |
+
"learning_rate": 1.7500462877244958e-05,
|
| 933 |
+
"loss": 0.8162,
|
| 934 |
+
"step": 1450
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.4,
|
| 938 |
+
"learning_rate": 1.7481947787446773e-05,
|
| 939 |
+
"loss": 0.8305,
|
| 940 |
+
"step": 1460
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.4,
|
| 944 |
+
"learning_rate": 1.7463432697648587e-05,
|
| 945 |
+
"loss": 0.8188,
|
| 946 |
+
"step": 1470
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 0.41,
|
| 950 |
+
"learning_rate": 1.74449176078504e-05,
|
| 951 |
+
"loss": 0.8265,
|
| 952 |
+
"step": 1480
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 0.41,
|
| 956 |
+
"learning_rate": 1.7426402518052213e-05,
|
| 957 |
+
"loss": 0.8265,
|
| 958 |
+
"step": 1490
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 0.41,
|
| 962 |
+
"learning_rate": 1.7407887428254028e-05,
|
| 963 |
+
"loss": 0.8215,
|
| 964 |
+
"step": 1500
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.42,
|
| 968 |
+
"learning_rate": 1.7389372338455843e-05,
|
| 969 |
+
"loss": 0.8311,
|
| 970 |
+
"step": 1510
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 0.42,
|
| 974 |
+
"learning_rate": 1.7370857248657657e-05,
|
| 975 |
+
"loss": 0.8315,
|
| 976 |
+
"step": 1520
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.42,
|
| 980 |
+
"learning_rate": 1.7352342158859472e-05,
|
| 981 |
+
"loss": 0.8306,
|
| 982 |
+
"step": 1530
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.42,
|
| 986 |
+
"learning_rate": 1.7333827069061287e-05,
|
| 987 |
+
"loss": 0.8212,
|
| 988 |
+
"step": 1540
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 0.43,
|
| 992 |
+
"learning_rate": 1.73153119792631e-05,
|
| 993 |
+
"loss": 0.8263,
|
| 994 |
+
"step": 1550
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.43,
|
| 998 |
+
"learning_rate": 1.7296796889464916e-05,
|
| 999 |
+
"loss": 0.8179,
|
| 1000 |
+
"step": 1560
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 0.43,
|
| 1004 |
+
"learning_rate": 1.727828179966673e-05,
|
| 1005 |
+
"loss": 0.8292,
|
| 1006 |
+
"step": 1570
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 0.43,
|
| 1010 |
+
"learning_rate": 1.7259766709868546e-05,
|
| 1011 |
+
"loss": 0.8115,
|
| 1012 |
+
"step": 1580
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 0.44,
|
| 1016 |
+
"learning_rate": 1.7241251620070357e-05,
|
| 1017 |
+
"loss": 0.8203,
|
| 1018 |
+
"step": 1590
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.44,
|
| 1022 |
+
"learning_rate": 1.722273653027217e-05,
|
| 1023 |
+
"loss": 0.8342,
|
| 1024 |
+
"step": 1600
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.44,
|
| 1028 |
+
"eval_loss": 0.825358510017395,
|
| 1029 |
+
"eval_runtime": 98.8719,
|
| 1030 |
+
"eval_samples_per_second": 10.114,
|
| 1031 |
+
"eval_steps_per_second": 0.637,
|
| 1032 |
+
"step": 1600
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.44,
|
| 1036 |
+
"learning_rate": 1.7204221440473986e-05,
|
| 1037 |
+
"loss": 0.8396,
|
| 1038 |
+
"step": 1610
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.45,
|
| 1042 |
+
"learning_rate": 1.71857063506758e-05,
|
| 1043 |
+
"loss": 0.8271,
|
| 1044 |
+
"step": 1620
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.45,
|
| 1048 |
+
"learning_rate": 1.7167191260877616e-05,
|
| 1049 |
+
"loss": 0.8199,
|
| 1050 |
+
"step": 1630
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.45,
|
| 1054 |
+
"learning_rate": 1.714867617107943e-05,
|
| 1055 |
+
"loss": 0.8315,
|
| 1056 |
+
"step": 1640
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 0.45,
|
| 1060 |
+
"learning_rate": 1.7130161081281245e-05,
|
| 1061 |
+
"loss": 0.8279,
|
| 1062 |
+
"step": 1650
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 0.46,
|
| 1066 |
+
"learning_rate": 1.711164599148306e-05,
|
| 1067 |
+
"loss": 0.8146,
|
| 1068 |
+
"step": 1660
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 0.46,
|
| 1072 |
+
"learning_rate": 1.7093130901684874e-05,
|
| 1073 |
+
"loss": 0.8198,
|
| 1074 |
+
"step": 1670
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.46,
|
| 1078 |
+
"learning_rate": 1.707461581188669e-05,
|
| 1079 |
+
"loss": 0.8146,
|
| 1080 |
+
"step": 1680
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.46,
|
| 1084 |
+
"learning_rate": 1.7056100722088504e-05,
|
| 1085 |
+
"loss": 0.8288,
|
| 1086 |
+
"step": 1690
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 0.47,
|
| 1090 |
+
"learning_rate": 1.703758563229032e-05,
|
| 1091 |
+
"loss": 0.8208,
|
| 1092 |
+
"step": 1700
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 0.47,
|
| 1096 |
+
"learning_rate": 1.7019070542492133e-05,
|
| 1097 |
+
"loss": 0.8262,
|
| 1098 |
+
"step": 1710
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 0.47,
|
| 1102 |
+
"learning_rate": 1.7000555452693948e-05,
|
| 1103 |
+
"loss": 0.8196,
|
| 1104 |
+
"step": 1720
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.48,
|
| 1108 |
+
"learning_rate": 1.6982040362895763e-05,
|
| 1109 |
+
"loss": 0.8191,
|
| 1110 |
+
"step": 1730
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.48,
|
| 1114 |
+
"learning_rate": 1.6963525273097577e-05,
|
| 1115 |
+
"loss": 0.8258,
|
| 1116 |
+
"step": 1740
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.48,
|
| 1120 |
+
"learning_rate": 1.6945010183299392e-05,
|
| 1121 |
+
"loss": 0.827,
|
| 1122 |
+
"step": 1750
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.48,
|
| 1126 |
+
"learning_rate": 1.6926495093501207e-05,
|
| 1127 |
+
"loss": 0.8234,
|
| 1128 |
+
"step": 1760
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 0.49,
|
| 1132 |
+
"learning_rate": 1.6907980003703018e-05,
|
| 1133 |
+
"loss": 0.8225,
|
| 1134 |
+
"step": 1770
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 0.49,
|
| 1138 |
+
"learning_rate": 1.6889464913904833e-05,
|
| 1139 |
+
"loss": 0.8242,
|
| 1140 |
+
"step": 1780
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 0.49,
|
| 1144 |
+
"learning_rate": 1.6870949824106647e-05,
|
| 1145 |
+
"loss": 0.818,
|
| 1146 |
+
"step": 1790
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.5,
|
| 1150 |
+
"learning_rate": 1.6852434734308462e-05,
|
| 1151 |
+
"loss": 0.8235,
|
| 1152 |
+
"step": 1800
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 0.5,
|
| 1156 |
+
"eval_loss": 0.8223171234130859,
|
| 1157 |
+
"eval_runtime": 99.1806,
|
| 1158 |
+
"eval_samples_per_second": 10.083,
|
| 1159 |
+
"eval_steps_per_second": 0.635,
|
| 1160 |
+
"step": 1800
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 0.5,
|
| 1164 |
+
"learning_rate": 1.6833919644510277e-05,
|
| 1165 |
+
"loss": 0.8208,
|
| 1166 |
+
"step": 1810
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 0.5,
|
| 1170 |
+
"learning_rate": 1.681540455471209e-05,
|
| 1171 |
+
"loss": 0.8241,
|
| 1172 |
+
"step": 1820
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.5,
|
| 1176 |
+
"learning_rate": 1.6796889464913906e-05,
|
| 1177 |
+
"loss": 0.8141,
|
| 1178 |
+
"step": 1830
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.51,
|
| 1182 |
+
"learning_rate": 1.677837437511572e-05,
|
| 1183 |
+
"loss": 0.8234,
|
| 1184 |
+
"step": 1840
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 0.51,
|
| 1188 |
+
"learning_rate": 1.6759859285317536e-05,
|
| 1189 |
+
"loss": 0.8276,
|
| 1190 |
+
"step": 1850
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 0.51,
|
| 1194 |
+
"learning_rate": 1.674134419551935e-05,
|
| 1195 |
+
"loss": 0.8241,
|
| 1196 |
+
"step": 1860
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 0.51,
|
| 1200 |
+
"learning_rate": 1.6722829105721165e-05,
|
| 1201 |
+
"loss": 0.8219,
|
| 1202 |
+
"step": 1870
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 0.52,
|
| 1206 |
+
"learning_rate": 1.6704314015922976e-05,
|
| 1207 |
+
"loss": 0.8259,
|
| 1208 |
+
"step": 1880
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 0.52,
|
| 1212 |
+
"learning_rate": 1.668579892612479e-05,
|
| 1213 |
+
"loss": 0.8142,
|
| 1214 |
+
"step": 1890
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.52,
|
| 1218 |
+
"learning_rate": 1.6667283836326606e-05,
|
| 1219 |
+
"loss": 0.8186,
|
| 1220 |
+
"step": 1900
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.53,
|
| 1224 |
+
"learning_rate": 1.6648768746528424e-05,
|
| 1225 |
+
"loss": 0.8128,
|
| 1226 |
+
"step": 1910
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.53,
|
| 1230 |
+
"learning_rate": 1.663025365673024e-05,
|
| 1231 |
+
"loss": 0.8126,
|
| 1232 |
+
"step": 1920
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 0.53,
|
| 1236 |
+
"learning_rate": 1.6611738566932053e-05,
|
| 1237 |
+
"loss": 0.8451,
|
| 1238 |
+
"step": 1930
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 0.53,
|
| 1242 |
+
"learning_rate": 1.6593223477133868e-05,
|
| 1243 |
+
"loss": 0.8175,
|
| 1244 |
+
"step": 1940
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 0.54,
|
| 1248 |
+
"learning_rate": 1.657470838733568e-05,
|
| 1249 |
+
"loss": 0.8194,
|
| 1250 |
+
"step": 1950
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 0.54,
|
| 1254 |
+
"learning_rate": 1.6556193297537494e-05,
|
| 1255 |
+
"loss": 0.8256,
|
| 1256 |
+
"step": 1960
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.54,
|
| 1260 |
+
"learning_rate": 1.653767820773931e-05,
|
| 1261 |
+
"loss": 0.8086,
|
| 1262 |
+
"step": 1970
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.54,
|
| 1266 |
+
"learning_rate": 1.6519163117941123e-05,
|
| 1267 |
+
"loss": 0.8267,
|
| 1268 |
+
"step": 1980
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 0.55,
|
| 1272 |
+
"learning_rate": 1.6500648028142938e-05,
|
| 1273 |
+
"loss": 0.8115,
|
| 1274 |
+
"step": 1990
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.55,
|
| 1278 |
+
"learning_rate": 1.6482132938344753e-05,
|
| 1279 |
+
"loss": 0.8103,
|
| 1280 |
+
"step": 2000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 0.55,
|
| 1284 |
+
"eval_loss": 0.8191922903060913,
|
| 1285 |
+
"eval_runtime": 99.2272,
|
| 1286 |
+
"eval_samples_per_second": 10.078,
|
| 1287 |
+
"eval_steps_per_second": 0.635,
|
| 1288 |
+
"step": 2000
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 0.55,
|
| 1292 |
+
"learning_rate": 1.6463617848546567e-05,
|
| 1293 |
+
"loss": 0.8309,
|
| 1294 |
+
"step": 2010
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 0.56,
|
| 1298 |
+
"learning_rate": 1.6445102758748382e-05,
|
| 1299 |
+
"loss": 0.8347,
|
| 1300 |
+
"step": 2020
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 0.56,
|
| 1304 |
+
"learning_rate": 1.6426587668950197e-05,
|
| 1305 |
+
"loss": 0.8317,
|
| 1306 |
+
"step": 2030
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.56,
|
| 1310 |
+
"learning_rate": 1.640807257915201e-05,
|
| 1311 |
+
"loss": 0.8177,
|
| 1312 |
+
"step": 2040
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.56,
|
| 1316 |
+
"learning_rate": 1.6389557489353826e-05,
|
| 1317 |
+
"loss": 0.825,
|
| 1318 |
+
"step": 2050
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.57,
|
| 1322 |
+
"learning_rate": 1.6371042399555637e-05,
|
| 1323 |
+
"loss": 0.8213,
|
| 1324 |
+
"step": 2060
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 0.57,
|
| 1328 |
+
"learning_rate": 1.6352527309757452e-05,
|
| 1329 |
+
"loss": 0.8268,
|
| 1330 |
+
"step": 2070
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 0.57,
|
| 1334 |
+
"learning_rate": 1.6334012219959267e-05,
|
| 1335 |
+
"loss": 0.8226,
|
| 1336 |
+
"step": 2080
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 0.58,
|
| 1340 |
+
"learning_rate": 1.631549713016108e-05,
|
| 1341 |
+
"loss": 0.8218,
|
| 1342 |
+
"step": 2090
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 0.58,
|
| 1346 |
+
"learning_rate": 1.6296982040362896e-05,
|
| 1347 |
+
"loss": 0.8081,
|
| 1348 |
+
"step": 2100
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"epoch": 0.58,
|
| 1352 |
+
"learning_rate": 1.627846695056471e-05,
|
| 1353 |
+
"loss": 0.8389,
|
| 1354 |
+
"step": 2110
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.58,
|
| 1358 |
+
"learning_rate": 1.6259951860766525e-05,
|
| 1359 |
+
"loss": 0.8278,
|
| 1360 |
+
"step": 2120
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.59,
|
| 1364 |
+
"learning_rate": 1.624143677096834e-05,
|
| 1365 |
+
"loss": 0.8185,
|
| 1366 |
+
"step": 2130
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 0.59,
|
| 1370 |
+
"learning_rate": 1.6222921681170155e-05,
|
| 1371 |
+
"loss": 0.8111,
|
| 1372 |
+
"step": 2140
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 0.59,
|
| 1376 |
+
"learning_rate": 1.620440659137197e-05,
|
| 1377 |
+
"loss": 0.8192,
|
| 1378 |
+
"step": 2150
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 0.59,
|
| 1382 |
+
"learning_rate": 1.6185891501573784e-05,
|
| 1383 |
+
"loss": 0.8197,
|
| 1384 |
+
"step": 2160
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 0.6,
|
| 1388 |
+
"learning_rate": 1.61673764117756e-05,
|
| 1389 |
+
"loss": 0.8208,
|
| 1390 |
+
"step": 2170
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 0.6,
|
| 1394 |
+
"learning_rate": 1.6148861321977414e-05,
|
| 1395 |
+
"loss": 0.8087,
|
| 1396 |
+
"step": 2180
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.6,
|
| 1400 |
+
"learning_rate": 1.613034623217923e-05,
|
| 1401 |
+
"loss": 0.8217,
|
| 1402 |
+
"step": 2190
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.61,
|
| 1406 |
+
"learning_rate": 1.6111831142381043e-05,
|
| 1407 |
+
"loss": 0.8219,
|
| 1408 |
+
"step": 2200
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 0.61,
|
| 1412 |
+
"eval_loss": 0.8205804228782654,
|
| 1413 |
+
"eval_runtime": 99.3077,
|
| 1414 |
+
"eval_samples_per_second": 10.07,
|
| 1415 |
+
"eval_steps_per_second": 0.634,
|
| 1416 |
+
"step": 2200
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.61,
|
| 1420 |
+
"learning_rate": 1.6093316052582858e-05,
|
| 1421 |
+
"loss": 0.8267,
|
| 1422 |
+
"step": 2210
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 0.61,
|
| 1426 |
+
"learning_rate": 1.6074800962784672e-05,
|
| 1427 |
+
"loss": 0.8009,
|
| 1428 |
+
"step": 2220
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 0.61,
|
| 1432 |
+
"learning_rate": 1.6056285872986487e-05,
|
| 1433 |
+
"loss": 0.8234,
|
| 1434 |
+
"step": 2230
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.62,
|
| 1438 |
+
"learning_rate": 1.60377707831883e-05,
|
| 1439 |
+
"loss": 0.8112,
|
| 1440 |
+
"step": 2240
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 0.62,
|
| 1444 |
+
"learning_rate": 1.6019255693390113e-05,
|
| 1445 |
+
"loss": 0.8181,
|
| 1446 |
+
"step": 2250
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 0.62,
|
| 1450 |
+
"learning_rate": 1.6000740603591928e-05,
|
| 1451 |
+
"loss": 0.8234,
|
| 1452 |
+
"step": 2260
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 0.62,
|
| 1456 |
+
"learning_rate": 1.5982225513793742e-05,
|
| 1457 |
+
"loss": 0.8142,
|
| 1458 |
+
"step": 2270
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.63,
|
| 1462 |
+
"learning_rate": 1.5963710423995557e-05,
|
| 1463 |
+
"loss": 0.8083,
|
| 1464 |
+
"step": 2280
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 0.63,
|
| 1468 |
+
"learning_rate": 1.5945195334197372e-05,
|
| 1469 |
+
"loss": 0.8099,
|
| 1470 |
+
"step": 2290
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 0.63,
|
| 1474 |
+
"learning_rate": 1.5926680244399187e-05,
|
| 1475 |
+
"loss": 0.8164,
|
| 1476 |
+
"step": 2300
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 0.64,
|
| 1480 |
+
"learning_rate": 1.5908165154601e-05,
|
| 1481 |
+
"loss": 0.8095,
|
| 1482 |
+
"step": 2310
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 0.64,
|
| 1486 |
+
"learning_rate": 1.5889650064802816e-05,
|
| 1487 |
+
"loss": 0.809,
|
| 1488 |
+
"step": 2320
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 0.64,
|
| 1492 |
+
"learning_rate": 1.587113497500463e-05,
|
| 1493 |
+
"loss": 0.8144,
|
| 1494 |
+
"step": 2330
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.64,
|
| 1498 |
+
"learning_rate": 1.5852619885206445e-05,
|
| 1499 |
+
"loss": 0.813,
|
| 1500 |
+
"step": 2340
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.65,
|
| 1504 |
+
"learning_rate": 1.5834104795408257e-05,
|
| 1505 |
+
"loss": 0.8114,
|
| 1506 |
+
"step": 2350
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 0.65,
|
| 1510 |
+
"learning_rate": 1.581558970561007e-05,
|
| 1511 |
+
"loss": 0.8217,
|
| 1512 |
+
"step": 2360
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 0.65,
|
| 1516 |
+
"learning_rate": 1.5797074615811886e-05,
|
| 1517 |
+
"loss": 0.7975,
|
| 1518 |
+
"step": 2370
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 0.65,
|
| 1522 |
+
"learning_rate": 1.57785595260137e-05,
|
| 1523 |
+
"loss": 0.8268,
|
| 1524 |
+
"step": 2380
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 0.66,
|
| 1528 |
+
"learning_rate": 1.5760044436215515e-05,
|
| 1529 |
+
"loss": 0.8188,
|
| 1530 |
+
"step": 2390
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 0.66,
|
| 1534 |
+
"learning_rate": 1.5741529346417333e-05,
|
| 1535 |
+
"loss": 0.8103,
|
| 1536 |
+
"step": 2400
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.66,
|
| 1540 |
+
"eval_loss": 0.8151404857635498,
|
| 1541 |
+
"eval_runtime": 99.0346,
|
| 1542 |
+
"eval_samples_per_second": 10.097,
|
| 1543 |
+
"eval_steps_per_second": 0.636,
|
| 1544 |
+
"step": 2400
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 0.66,
|
| 1548 |
+
"learning_rate": 1.5723014256619148e-05,
|
| 1549 |
+
"loss": 0.813,
|
| 1550 |
+
"step": 2410
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.67,
|
| 1554 |
+
"learning_rate": 1.5704499166820963e-05,
|
| 1555 |
+
"loss": 0.8175,
|
| 1556 |
+
"step": 2420
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.67,
|
| 1560 |
+
"learning_rate": 1.5685984077022774e-05,
|
| 1561 |
+
"loss": 0.8167,
|
| 1562 |
+
"step": 2430
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 0.67,
|
| 1566 |
+
"learning_rate": 1.566746898722459e-05,
|
| 1567 |
+
"loss": 0.8215,
|
| 1568 |
+
"step": 2440
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 0.67,
|
| 1572 |
+
"learning_rate": 1.5648953897426404e-05,
|
| 1573 |
+
"loss": 0.8129,
|
| 1574 |
+
"step": 2450
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 0.68,
|
| 1578 |
+
"learning_rate": 1.5630438807628218e-05,
|
| 1579 |
+
"loss": 0.8164,
|
| 1580 |
+
"step": 2460
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 0.68,
|
| 1584 |
+
"learning_rate": 1.5611923717830033e-05,
|
| 1585 |
+
"loss": 0.7983,
|
| 1586 |
+
"step": 2470
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.68,
|
| 1590 |
+
"learning_rate": 1.5593408628031848e-05,
|
| 1591 |
+
"loss": 0.8108,
|
| 1592 |
+
"step": 2480
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.69,
|
| 1596 |
+
"learning_rate": 1.5574893538233662e-05,
|
| 1597 |
+
"loss": 0.8064,
|
| 1598 |
+
"step": 2490
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.69,
|
| 1602 |
+
"learning_rate": 1.5556378448435477e-05,
|
| 1603 |
+
"loss": 0.8165,
|
| 1604 |
+
"step": 2500
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 0.69,
|
| 1608 |
+
"learning_rate": 1.5537863358637292e-05,
|
| 1609 |
+
"loss": 0.8008,
|
| 1610 |
+
"step": 2510
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.69,
|
| 1614 |
+
"learning_rate": 1.5519348268839106e-05,
|
| 1615 |
+
"loss": 0.8078,
|
| 1616 |
+
"step": 2520
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 0.7,
|
| 1620 |
+
"learning_rate": 1.550083317904092e-05,
|
| 1621 |
+
"loss": 0.7909,
|
| 1622 |
+
"step": 2530
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 0.7,
|
| 1626 |
+
"learning_rate": 1.5482318089242732e-05,
|
| 1627 |
+
"loss": 0.8132,
|
| 1628 |
+
"step": 2540
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 0.7,
|
| 1632 |
+
"learning_rate": 1.5463802999444547e-05,
|
| 1633 |
+
"loss": 0.8092,
|
| 1634 |
+
"step": 2550
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.7,
|
| 1638 |
+
"learning_rate": 1.5445287909646362e-05,
|
| 1639 |
+
"loss": 0.823,
|
| 1640 |
+
"step": 2560
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.71,
|
| 1644 |
+
"learning_rate": 1.5426772819848176e-05,
|
| 1645 |
+
"loss": 0.8063,
|
| 1646 |
+
"step": 2570
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 0.71,
|
| 1650 |
+
"learning_rate": 1.540825773004999e-05,
|
| 1651 |
+
"loss": 0.8111,
|
| 1652 |
+
"step": 2580
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 0.71,
|
| 1656 |
+
"learning_rate": 1.5389742640251806e-05,
|
| 1657 |
+
"loss": 0.8013,
|
| 1658 |
+
"step": 2590
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.72,
|
| 1662 |
+
"learning_rate": 1.537122755045362e-05,
|
| 1663 |
+
"loss": 0.7979,
|
| 1664 |
+
"step": 2600
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 0.72,
|
| 1668 |
+
"eval_loss": 0.8133894801139832,
|
| 1669 |
+
"eval_runtime": 98.8127,
|
| 1670 |
+
"eval_samples_per_second": 10.12,
|
| 1671 |
+
"eval_steps_per_second": 0.638,
|
| 1672 |
+
"step": 2600
|
| 1673 |
+
},
|
| 1674 |
+
{
|
| 1675 |
+
"epoch": 0.72,
|
| 1676 |
+
"learning_rate": 1.5352712460655435e-05,
|
| 1677 |
+
"loss": 0.8191,
|
| 1678 |
+
"step": 2610
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 0.72,
|
| 1682 |
+
"learning_rate": 1.533419737085725e-05,
|
| 1683 |
+
"loss": 0.815,
|
| 1684 |
+
"step": 2620
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 0.72,
|
| 1688 |
+
"learning_rate": 1.5315682281059065e-05,
|
| 1689 |
+
"loss": 0.8041,
|
| 1690 |
+
"step": 2630
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.73,
|
| 1694 |
+
"learning_rate": 1.529716719126088e-05,
|
| 1695 |
+
"loss": 0.8123,
|
| 1696 |
+
"step": 2640
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 0.73,
|
| 1700 |
+
"learning_rate": 1.5278652101462694e-05,
|
| 1701 |
+
"loss": 0.8097,
|
| 1702 |
+
"step": 2650
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 0.73,
|
| 1706 |
+
"learning_rate": 1.526013701166451e-05,
|
| 1707 |
+
"loss": 0.8256,
|
| 1708 |
+
"step": 2660
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"epoch": 0.73,
|
| 1712 |
+
"learning_rate": 1.5241621921866323e-05,
|
| 1713 |
+
"loss": 0.8071,
|
| 1714 |
+
"step": 2670
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 0.74,
|
| 1718 |
+
"learning_rate": 1.5223106832068138e-05,
|
| 1719 |
+
"loss": 0.7995,
|
| 1720 |
+
"step": 2680
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 0.74,
|
| 1724 |
+
"learning_rate": 1.5204591742269951e-05,
|
| 1725 |
+
"loss": 0.8017,
|
| 1726 |
+
"step": 2690
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 0.74,
|
| 1730 |
+
"learning_rate": 1.5186076652471766e-05,
|
| 1731 |
+
"loss": 0.8111,
|
| 1732 |
+
"step": 2700
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 0.75,
|
| 1736 |
+
"learning_rate": 1.516756156267358e-05,
|
| 1737 |
+
"loss": 0.8154,
|
| 1738 |
+
"step": 2710
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.75,
|
| 1742 |
+
"learning_rate": 1.5149046472875395e-05,
|
| 1743 |
+
"loss": 0.8056,
|
| 1744 |
+
"step": 2720
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 0.75,
|
| 1748 |
+
"learning_rate": 1.513053138307721e-05,
|
| 1749 |
+
"loss": 0.7996,
|
| 1750 |
+
"step": 2730
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 0.75,
|
| 1754 |
+
"learning_rate": 1.5112016293279025e-05,
|
| 1755 |
+
"loss": 0.8149,
|
| 1756 |
+
"step": 2740
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 0.76,
|
| 1760 |
+
"learning_rate": 1.5093501203480838e-05,
|
| 1761 |
+
"loss": 0.8108,
|
| 1762 |
+
"step": 2750
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 0.76,
|
| 1766 |
+
"learning_rate": 1.5074986113682652e-05,
|
| 1767 |
+
"loss": 0.8093,
|
| 1768 |
+
"step": 2760
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 0.76,
|
| 1772 |
+
"learning_rate": 1.5056471023884467e-05,
|
| 1773 |
+
"loss": 0.8063,
|
| 1774 |
+
"step": 2770
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.76,
|
| 1778 |
+
"learning_rate": 1.5037955934086282e-05,
|
| 1779 |
+
"loss": 0.8089,
|
| 1780 |
+
"step": 2780
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 0.77,
|
| 1784 |
+
"learning_rate": 1.5019440844288096e-05,
|
| 1785 |
+
"loss": 0.8059,
|
| 1786 |
+
"step": 2790
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 0.77,
|
| 1790 |
+
"learning_rate": 1.500092575448991e-05,
|
| 1791 |
+
"loss": 0.8136,
|
| 1792 |
+
"step": 2800
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 0.77,
|
| 1796 |
+
"eval_loss": 0.8110187649726868,
|
| 1797 |
+
"eval_runtime": 98.9424,
|
| 1798 |
+
"eval_samples_per_second": 10.107,
|
| 1799 |
+
"eval_steps_per_second": 0.637,
|
| 1800 |
+
"step": 2800
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 0.77,
|
| 1804 |
+
"learning_rate": 1.4982410664691724e-05,
|
| 1805 |
+
"loss": 0.8123,
|
| 1806 |
+
"step": 2810
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 0.78,
|
| 1810 |
+
"learning_rate": 1.4963895574893539e-05,
|
| 1811 |
+
"loss": 0.8198,
|
| 1812 |
+
"step": 2820
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 0.78,
|
| 1816 |
+
"learning_rate": 1.4945380485095353e-05,
|
| 1817 |
+
"loss": 0.8014,
|
| 1818 |
+
"step": 2830
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 0.78,
|
| 1822 |
+
"learning_rate": 1.4926865395297168e-05,
|
| 1823 |
+
"loss": 0.8017,
|
| 1824 |
+
"step": 2840
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 0.78,
|
| 1828 |
+
"learning_rate": 1.4908350305498983e-05,
|
| 1829 |
+
"loss": 0.8104,
|
| 1830 |
+
"step": 2850
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 0.79,
|
| 1834 |
+
"learning_rate": 1.4889835215700796e-05,
|
| 1835 |
+
"loss": 0.7939,
|
| 1836 |
+
"step": 2860
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 0.79,
|
| 1840 |
+
"learning_rate": 1.487132012590261e-05,
|
| 1841 |
+
"loss": 0.8072,
|
| 1842 |
+
"step": 2870
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 0.79,
|
| 1846 |
+
"learning_rate": 1.4852805036104427e-05,
|
| 1847 |
+
"loss": 0.818,
|
| 1848 |
+
"step": 2880
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 0.8,
|
| 1852 |
+
"learning_rate": 1.4834289946306242e-05,
|
| 1853 |
+
"loss": 0.8024,
|
| 1854 |
+
"step": 2890
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 0.8,
|
| 1858 |
+
"learning_rate": 1.4815774856508056e-05,
|
| 1859 |
+
"loss": 0.8166,
|
| 1860 |
+
"step": 2900
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 0.8,
|
| 1864 |
+
"learning_rate": 1.4797259766709871e-05,
|
| 1865 |
+
"loss": 0.8047,
|
| 1866 |
+
"step": 2910
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 0.8,
|
| 1870 |
+
"learning_rate": 1.4778744676911686e-05,
|
| 1871 |
+
"loss": 0.8186,
|
| 1872 |
+
"step": 2920
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 0.81,
|
| 1876 |
+
"learning_rate": 1.4760229587113499e-05,
|
| 1877 |
+
"loss": 0.8089,
|
| 1878 |
+
"step": 2930
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 0.81,
|
| 1882 |
+
"learning_rate": 1.4741714497315313e-05,
|
| 1883 |
+
"loss": 0.8053,
|
| 1884 |
+
"step": 2940
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 0.81,
|
| 1888 |
+
"learning_rate": 1.4723199407517128e-05,
|
| 1889 |
+
"loss": 0.8016,
|
| 1890 |
+
"step": 2950
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 0.81,
|
| 1894 |
+
"learning_rate": 1.4704684317718943e-05,
|
| 1895 |
+
"loss": 0.8004,
|
| 1896 |
+
"step": 2960
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 0.82,
|
| 1900 |
+
"learning_rate": 1.4686169227920757e-05,
|
| 1901 |
+
"loss": 0.815,
|
| 1902 |
+
"step": 2970
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 0.82,
|
| 1906 |
+
"learning_rate": 1.466765413812257e-05,
|
| 1907 |
+
"loss": 0.8055,
|
| 1908 |
+
"step": 2980
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 0.82,
|
| 1912 |
+
"learning_rate": 1.4649139048324385e-05,
|
| 1913 |
+
"loss": 0.8142,
|
| 1914 |
+
"step": 2990
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.83,
|
| 1918 |
+
"learning_rate": 1.46306239585262e-05,
|
| 1919 |
+
"loss": 0.8149,
|
| 1920 |
+
"step": 3000
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 0.83,
|
| 1924 |
+
"eval_loss": 0.809707760810852,
|
| 1925 |
+
"eval_runtime": 98.7818,
|
| 1926 |
+
"eval_samples_per_second": 10.123,
|
| 1927 |
+
"eval_steps_per_second": 0.638,
|
| 1928 |
+
"step": 3000
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 0.83,
|
| 1932 |
+
"learning_rate": 1.4612108868728014e-05,
|
| 1933 |
+
"loss": 0.7967,
|
| 1934 |
+
"step": 3010
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.83,
|
| 1938 |
+
"learning_rate": 1.4593593778929829e-05,
|
| 1939 |
+
"loss": 0.8142,
|
| 1940 |
+
"step": 3020
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 0.83,
|
| 1944 |
+
"learning_rate": 1.4575078689131644e-05,
|
| 1945 |
+
"loss": 0.8023,
|
| 1946 |
+
"step": 3030
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.84,
|
| 1950 |
+
"learning_rate": 1.4556563599333457e-05,
|
| 1951 |
+
"loss": 0.8143,
|
| 1952 |
+
"step": 3040
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 0.84,
|
| 1956 |
+
"learning_rate": 1.4538048509535272e-05,
|
| 1957 |
+
"loss": 0.8165,
|
| 1958 |
+
"step": 3050
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 0.84,
|
| 1962 |
+
"learning_rate": 1.4519533419737086e-05,
|
| 1963 |
+
"loss": 0.8052,
|
| 1964 |
+
"step": 3060
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 0.84,
|
| 1968 |
+
"learning_rate": 1.4501018329938901e-05,
|
| 1969 |
+
"loss": 0.8136,
|
| 1970 |
+
"step": 3070
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 0.85,
|
| 1974 |
+
"learning_rate": 1.4482503240140716e-05,
|
| 1975 |
+
"loss": 0.8128,
|
| 1976 |
+
"step": 3080
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 0.85,
|
| 1980 |
+
"learning_rate": 1.4463988150342529e-05,
|
| 1981 |
+
"loss": 0.7964,
|
| 1982 |
+
"step": 3090
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 0.85,
|
| 1986 |
+
"learning_rate": 1.4445473060544343e-05,
|
| 1987 |
+
"loss": 0.8157,
|
| 1988 |
+
"step": 3100
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 0.86,
|
| 1992 |
+
"learning_rate": 1.4426957970746158e-05,
|
| 1993 |
+
"loss": 0.8031,
|
| 1994 |
+
"step": 3110
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.86,
|
| 1998 |
+
"learning_rate": 1.4408442880947974e-05,
|
| 1999 |
+
"loss": 0.8041,
|
| 2000 |
+
"step": 3120
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 0.86,
|
| 2004 |
+
"learning_rate": 1.4389927791149789e-05,
|
| 2005 |
+
"loss": 0.8112,
|
| 2006 |
+
"step": 3130
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 0.86,
|
| 2010 |
+
"learning_rate": 1.4371412701351604e-05,
|
| 2011 |
+
"loss": 0.8067,
|
| 2012 |
+
"step": 3140
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 0.87,
|
| 2016 |
+
"learning_rate": 1.4352897611553418e-05,
|
| 2017 |
+
"loss": 0.8163,
|
| 2018 |
+
"step": 3150
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 0.87,
|
| 2022 |
+
"learning_rate": 1.4334382521755233e-05,
|
| 2023 |
+
"loss": 0.8051,
|
| 2024 |
+
"step": 3160
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 0.87,
|
| 2028 |
+
"learning_rate": 1.4315867431957046e-05,
|
| 2029 |
+
"loss": 0.8179,
|
| 2030 |
+
"step": 3170
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 0.87,
|
| 2034 |
+
"learning_rate": 1.429735234215886e-05,
|
| 2035 |
+
"loss": 0.7973,
|
| 2036 |
+
"step": 3180
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 0.88,
|
| 2040 |
+
"learning_rate": 1.4278837252360676e-05,
|
| 2041 |
+
"loss": 0.811,
|
| 2042 |
+
"step": 3190
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 0.88,
|
| 2046 |
+
"learning_rate": 1.426032216256249e-05,
|
| 2047 |
+
"loss": 0.7944,
|
| 2048 |
+
"step": 3200
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 0.88,
|
| 2052 |
+
"eval_loss": 0.8083726763725281,
|
| 2053 |
+
"eval_runtime": 99.0291,
|
| 2054 |
+
"eval_samples_per_second": 10.098,
|
| 2055 |
+
"eval_steps_per_second": 0.636,
|
| 2056 |
+
"step": 3200
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 0.88,
|
| 2060 |
+
"learning_rate": 1.4241807072764305e-05,
|
| 2061 |
+
"loss": 0.8138,
|
| 2062 |
+
"step": 3210
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 0.89,
|
| 2066 |
+
"learning_rate": 1.4223291982966118e-05,
|
| 2067 |
+
"loss": 0.7919,
|
| 2068 |
+
"step": 3220
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 0.89,
|
| 2072 |
+
"learning_rate": 1.4204776893167933e-05,
|
| 2073 |
+
"loss": 0.8061,
|
| 2074 |
+
"step": 3230
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 0.89,
|
| 2078 |
+
"learning_rate": 1.4186261803369747e-05,
|
| 2079 |
+
"loss": 0.8014,
|
| 2080 |
+
"step": 3240
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 0.89,
|
| 2084 |
+
"learning_rate": 1.4167746713571562e-05,
|
| 2085 |
+
"loss": 0.8299,
|
| 2086 |
+
"step": 3250
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 0.9,
|
| 2090 |
+
"learning_rate": 1.4149231623773377e-05,
|
| 2091 |
+
"loss": 0.8038,
|
| 2092 |
+
"step": 3260
|
| 2093 |
+
},
|
| 2094 |
+
{
|
| 2095 |
+
"epoch": 0.9,
|
| 2096 |
+
"learning_rate": 1.4130716533975191e-05,
|
| 2097 |
+
"loss": 0.7951,
|
| 2098 |
+
"step": 3270
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"epoch": 0.9,
|
| 2102 |
+
"learning_rate": 1.4112201444177004e-05,
|
| 2103 |
+
"loss": 0.8043,
|
| 2104 |
+
"step": 3280
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 0.91,
|
| 2108 |
+
"learning_rate": 1.4093686354378819e-05,
|
| 2109 |
+
"loss": 0.8158,
|
| 2110 |
+
"step": 3290
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 0.91,
|
| 2114 |
+
"learning_rate": 1.4075171264580634e-05,
|
| 2115 |
+
"loss": 0.8041,
|
| 2116 |
+
"step": 3300
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 0.91,
|
| 2120 |
+
"learning_rate": 1.4056656174782448e-05,
|
| 2121 |
+
"loss": 0.799,
|
| 2122 |
+
"step": 3310
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 0.91,
|
| 2126 |
+
"learning_rate": 1.4038141084984263e-05,
|
| 2127 |
+
"loss": 0.8033,
|
| 2128 |
+
"step": 3320
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"epoch": 0.92,
|
| 2132 |
+
"learning_rate": 1.4019625995186076e-05,
|
| 2133 |
+
"loss": 0.8097,
|
| 2134 |
+
"step": 3330
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"epoch": 0.92,
|
| 2138 |
+
"learning_rate": 1.400111090538789e-05,
|
| 2139 |
+
"loss": 0.8047,
|
| 2140 |
+
"step": 3340
|
| 2141 |
+
},
|
| 2142 |
+
{
|
| 2143 |
+
"epoch": 0.92,
|
| 2144 |
+
"learning_rate": 1.3982595815589706e-05,
|
| 2145 |
+
"loss": 0.82,
|
| 2146 |
+
"step": 3350
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 0.92,
|
| 2150 |
+
"learning_rate": 1.3964080725791522e-05,
|
| 2151 |
+
"loss": 0.8003,
|
| 2152 |
+
"step": 3360
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 0.93,
|
| 2156 |
+
"learning_rate": 1.3945565635993337e-05,
|
| 2157 |
+
"loss": 0.7975,
|
| 2158 |
+
"step": 3370
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 0.93,
|
| 2162 |
+
"learning_rate": 1.3927050546195151e-05,
|
| 2163 |
+
"loss": 0.8006,
|
| 2164 |
+
"step": 3380
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 0.93,
|
| 2168 |
+
"learning_rate": 1.3908535456396966e-05,
|
| 2169 |
+
"loss": 0.7952,
|
| 2170 |
+
"step": 3390
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 0.94,
|
| 2174 |
+
"learning_rate": 1.3890020366598779e-05,
|
| 2175 |
+
"loss": 0.8084,
|
| 2176 |
+
"step": 3400
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"epoch": 0.94,
|
| 2180 |
+
"eval_loss": 0.807253360748291,
|
| 2181 |
+
"eval_runtime": 98.8373,
|
| 2182 |
+
"eval_samples_per_second": 10.118,
|
| 2183 |
+
"eval_steps_per_second": 0.637,
|
| 2184 |
+
"step": 3400
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 0.94,
|
| 2188 |
+
"learning_rate": 1.3871505276800594e-05,
|
| 2189 |
+
"loss": 0.8168,
|
| 2190 |
+
"step": 3410
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"epoch": 0.94,
|
| 2194 |
+
"learning_rate": 1.3852990187002408e-05,
|
| 2195 |
+
"loss": 0.8134,
|
| 2196 |
+
"step": 3420
|
| 2197 |
+
},
|
| 2198 |
+
{
|
| 2199 |
+
"epoch": 0.94,
|
| 2200 |
+
"learning_rate": 1.3834475097204223e-05,
|
| 2201 |
+
"loss": 0.7968,
|
| 2202 |
+
"step": 3430
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 0.95,
|
| 2206 |
+
"learning_rate": 1.3815960007406038e-05,
|
| 2207 |
+
"loss": 0.8145,
|
| 2208 |
+
"step": 3440
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 0.95,
|
| 2212 |
+
"learning_rate": 1.3797444917607852e-05,
|
| 2213 |
+
"loss": 0.7996,
|
| 2214 |
+
"step": 3450
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 0.95,
|
| 2218 |
+
"learning_rate": 1.3778929827809665e-05,
|
| 2219 |
+
"loss": 0.8016,
|
| 2220 |
+
"step": 3460
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 0.95,
|
| 2224 |
+
"learning_rate": 1.376041473801148e-05,
|
| 2225 |
+
"loss": 0.7863,
|
| 2226 |
+
"step": 3470
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 0.96,
|
| 2230 |
+
"learning_rate": 1.3741899648213295e-05,
|
| 2231 |
+
"loss": 0.8193,
|
| 2232 |
+
"step": 3480
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 0.96,
|
| 2236 |
+
"learning_rate": 1.372338455841511e-05,
|
| 2237 |
+
"loss": 0.8148,
|
| 2238 |
+
"step": 3490
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 0.96,
|
| 2242 |
+
"learning_rate": 1.3704869468616924e-05,
|
| 2243 |
+
"loss": 0.8016,
|
| 2244 |
+
"step": 3500
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 0.97,
|
| 2248 |
+
"learning_rate": 1.3686354378818739e-05,
|
| 2249 |
+
"loss": 0.8077,
|
| 2250 |
+
"step": 3510
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 0.97,
|
| 2254 |
+
"learning_rate": 1.3667839289020552e-05,
|
| 2255 |
+
"loss": 0.806,
|
| 2256 |
+
"step": 3520
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 0.97,
|
| 2260 |
+
"learning_rate": 1.3649324199222367e-05,
|
| 2261 |
+
"loss": 0.8108,
|
| 2262 |
+
"step": 3530
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 0.97,
|
| 2266 |
+
"learning_rate": 1.3630809109424181e-05,
|
| 2267 |
+
"loss": 0.8101,
|
| 2268 |
+
"step": 3540
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 0.98,
|
| 2272 |
+
"learning_rate": 1.3612294019625996e-05,
|
| 2273 |
+
"loss": 0.8051,
|
| 2274 |
+
"step": 3550
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 0.98,
|
| 2278 |
+
"learning_rate": 1.359377892982781e-05,
|
| 2279 |
+
"loss": 0.8079,
|
| 2280 |
+
"step": 3560
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 0.98,
|
| 2284 |
+
"learning_rate": 1.3575263840029624e-05,
|
| 2285 |
+
"loss": 0.8078,
|
| 2286 |
+
"step": 3570
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 0.99,
|
| 2290 |
+
"learning_rate": 1.3556748750231438e-05,
|
| 2291 |
+
"loss": 0.7974,
|
| 2292 |
+
"step": 3580
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 0.99,
|
| 2296 |
+
"learning_rate": 1.3538233660433253e-05,
|
| 2297 |
+
"loss": 0.8152,
|
| 2298 |
+
"step": 3590
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 0.99,
|
| 2302 |
+
"learning_rate": 1.351971857063507e-05,
|
| 2303 |
+
"loss": 0.7966,
|
| 2304 |
+
"step": 3600
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 0.99,
|
| 2308 |
+
"eval_loss": 0.8059168457984924,
|
| 2309 |
+
"eval_runtime": 98.7407,
|
| 2310 |
+
"eval_samples_per_second": 10.128,
|
| 2311 |
+
"eval_steps_per_second": 0.638,
|
| 2312 |
+
"step": 3600
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 0.99,
|
| 2316 |
+
"learning_rate": 1.3501203480836884e-05,
|
| 2317 |
+
"loss": 0.8112,
|
| 2318 |
+
"step": 3610
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 1.0,
|
| 2322 |
+
"learning_rate": 1.3482688391038699e-05,
|
| 2323 |
+
"loss": 0.8148,
|
| 2324 |
+
"step": 3620
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 1.0,
|
| 2328 |
+
"learning_rate": 1.3464173301240514e-05,
|
| 2329 |
+
"loss": 0.8081,
|
| 2330 |
+
"step": 3630
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 1.0,
|
| 2334 |
+
"learning_rate": 1.3445658211442327e-05,
|
| 2335 |
+
"loss": 0.8169,
|
| 2336 |
+
"step": 3640
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 1.0,
|
| 2340 |
+
"learning_rate": 1.3427143121644141e-05,
|
| 2341 |
+
"loss": 0.8125,
|
| 2342 |
+
"step": 3650
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 1.01,
|
| 2346 |
+
"learning_rate": 1.3408628031845956e-05,
|
| 2347 |
+
"loss": 0.81,
|
| 2348 |
+
"step": 3660
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 1.01,
|
| 2352 |
+
"learning_rate": 1.339011294204777e-05,
|
| 2353 |
+
"loss": 0.7986,
|
| 2354 |
+
"step": 3670
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 1.01,
|
| 2358 |
+
"learning_rate": 1.3371597852249585e-05,
|
| 2359 |
+
"loss": 0.8034,
|
| 2360 |
+
"step": 3680
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 1.02,
|
| 2364 |
+
"learning_rate": 1.33530827624514e-05,
|
| 2365 |
+
"loss": 0.8135,
|
| 2366 |
+
"step": 3690
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 1.02,
|
| 2370 |
+
"learning_rate": 1.3334567672653213e-05,
|
| 2371 |
+
"loss": 0.7991,
|
| 2372 |
+
"step": 3700
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 1.02,
|
| 2376 |
+
"learning_rate": 1.3316052582855028e-05,
|
| 2377 |
+
"loss": 0.8021,
|
| 2378 |
+
"step": 3710
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 1.02,
|
| 2382 |
+
"learning_rate": 1.3297537493056842e-05,
|
| 2383 |
+
"loss": 0.793,
|
| 2384 |
+
"step": 3720
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 1.03,
|
| 2388 |
+
"learning_rate": 1.3279022403258657e-05,
|
| 2389 |
+
"loss": 0.7976,
|
| 2390 |
+
"step": 3730
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 1.03,
|
| 2394 |
+
"learning_rate": 1.3260507313460472e-05,
|
| 2395 |
+
"loss": 0.7978,
|
| 2396 |
+
"step": 3740
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 1.03,
|
| 2400 |
+
"learning_rate": 1.3241992223662285e-05,
|
| 2401 |
+
"loss": 0.8048,
|
| 2402 |
+
"step": 3750
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 1.03,
|
| 2406 |
+
"learning_rate": 1.32234771338641e-05,
|
| 2407 |
+
"loss": 0.814,
|
| 2408 |
+
"step": 3760
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 1.04,
|
| 2412 |
+
"learning_rate": 1.3204962044065914e-05,
|
| 2413 |
+
"loss": 0.8091,
|
| 2414 |
+
"step": 3770
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 1.04,
|
| 2418 |
+
"learning_rate": 1.3186446954267729e-05,
|
| 2419 |
+
"loss": 0.8128,
|
| 2420 |
+
"step": 3780
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 1.04,
|
| 2424 |
+
"learning_rate": 1.3167931864469544e-05,
|
| 2425 |
+
"loss": 0.8147,
|
| 2426 |
+
"step": 3790
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 1.05,
|
| 2430 |
+
"learning_rate": 1.3149416774671358e-05,
|
| 2431 |
+
"loss": 0.8067,
|
| 2432 |
+
"step": 3800
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 1.05,
|
| 2436 |
+
"eval_loss": 0.8047506809234619,
|
| 2437 |
+
"eval_runtime": 98.7957,
|
| 2438 |
+
"eval_samples_per_second": 10.122,
|
| 2439 |
+
"eval_steps_per_second": 0.638,
|
| 2440 |
+
"step": 3800
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 1.05,
|
| 2444 |
+
"learning_rate": 1.3130901684873171e-05,
|
| 2445 |
+
"loss": 0.7979,
|
| 2446 |
+
"step": 3810
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 1.05,
|
| 2450 |
+
"learning_rate": 1.3112386595074986e-05,
|
| 2451 |
+
"loss": 0.8154,
|
| 2452 |
+
"step": 3820
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 1.05,
|
| 2456 |
+
"learning_rate": 1.30938715052768e-05,
|
| 2457 |
+
"loss": 0.8133,
|
| 2458 |
+
"step": 3830
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 1.06,
|
| 2462 |
+
"learning_rate": 1.3075356415478617e-05,
|
| 2463 |
+
"loss": 0.7961,
|
| 2464 |
+
"step": 3840
|
| 2465 |
+
},
|
| 2466 |
+
{
|
| 2467 |
+
"epoch": 1.06,
|
| 2468 |
+
"learning_rate": 1.3056841325680432e-05,
|
| 2469 |
+
"loss": 0.8019,
|
| 2470 |
+
"step": 3850
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"epoch": 1.06,
|
| 2474 |
+
"learning_rate": 1.3038326235882246e-05,
|
| 2475 |
+
"loss": 0.7977,
|
| 2476 |
+
"step": 3860
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 1.06,
|
| 2480 |
+
"learning_rate": 1.3019811146084061e-05,
|
| 2481 |
+
"loss": 0.7995,
|
| 2482 |
+
"step": 3870
|
| 2483 |
+
},
|
| 2484 |
+
{
|
| 2485 |
+
"epoch": 1.07,
|
| 2486 |
+
"learning_rate": 1.3001296056285874e-05,
|
| 2487 |
+
"loss": 0.7997,
|
| 2488 |
+
"step": 3880
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 1.07,
|
| 2492 |
+
"learning_rate": 1.2982780966487689e-05,
|
| 2493 |
+
"loss": 0.8025,
|
| 2494 |
+
"step": 3890
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 1.07,
|
| 2498 |
+
"learning_rate": 1.2964265876689503e-05,
|
| 2499 |
+
"loss": 0.7911,
|
| 2500 |
+
"step": 3900
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 1.08,
|
| 2504 |
+
"learning_rate": 1.2945750786891318e-05,
|
| 2505 |
+
"loss": 0.8097,
|
| 2506 |
+
"step": 3910
|
| 2507 |
+
},
|
| 2508 |
+
{
|
| 2509 |
+
"epoch": 1.08,
|
| 2510 |
+
"learning_rate": 1.2927235697093133e-05,
|
| 2511 |
+
"loss": 0.7942,
|
| 2512 |
+
"step": 3920
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 1.08,
|
| 2516 |
+
"learning_rate": 1.2908720607294948e-05,
|
| 2517 |
+
"loss": 0.7999,
|
| 2518 |
+
"step": 3930
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 1.08,
|
| 2522 |
+
"learning_rate": 1.289020551749676e-05,
|
| 2523 |
+
"loss": 0.7935,
|
| 2524 |
+
"step": 3940
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 1.09,
|
| 2528 |
+
"learning_rate": 1.2871690427698575e-05,
|
| 2529 |
+
"loss": 0.8069,
|
| 2530 |
+
"step": 3950
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 1.09,
|
| 2534 |
+
"learning_rate": 1.285317533790039e-05,
|
| 2535 |
+
"loss": 0.8121,
|
| 2536 |
+
"step": 3960
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 1.09,
|
| 2540 |
+
"learning_rate": 1.2834660248102205e-05,
|
| 2541 |
+
"loss": 0.8206,
|
| 2542 |
+
"step": 3970
|
| 2543 |
+
},
|
| 2544 |
+
{
|
| 2545 |
+
"epoch": 1.1,
|
| 2546 |
+
"learning_rate": 1.281614515830402e-05,
|
| 2547 |
+
"loss": 0.7966,
|
| 2548 |
+
"step": 3980
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 1.1,
|
| 2552 |
+
"learning_rate": 1.2797630068505832e-05,
|
| 2553 |
+
"loss": 0.7969,
|
| 2554 |
+
"step": 3990
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 1.1,
|
| 2558 |
+
"learning_rate": 1.2779114978707647e-05,
|
| 2559 |
+
"loss": 0.8021,
|
| 2560 |
+
"step": 4000
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 1.1,
|
| 2564 |
+
"eval_loss": 0.8035591840744019,
|
| 2565 |
+
"eval_runtime": 98.7519,
|
| 2566 |
+
"eval_samples_per_second": 10.126,
|
| 2567 |
+
"eval_steps_per_second": 0.638,
|
| 2568 |
+
"step": 4000
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 1.1,
|
| 2572 |
+
"learning_rate": 1.2760599888909462e-05,
|
| 2573 |
+
"loss": 0.8029,
|
| 2574 |
+
"step": 4010
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 1.11,
|
| 2578 |
+
"learning_rate": 1.2742084799111276e-05,
|
| 2579 |
+
"loss": 0.8043,
|
| 2580 |
+
"step": 4020
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 1.11,
|
| 2584 |
+
"learning_rate": 1.2723569709313091e-05,
|
| 2585 |
+
"loss": 0.8049,
|
| 2586 |
+
"step": 4030
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 1.11,
|
| 2590 |
+
"learning_rate": 1.2705054619514906e-05,
|
| 2591 |
+
"loss": 0.8068,
|
| 2592 |
+
"step": 4040
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 1.11,
|
| 2596 |
+
"learning_rate": 1.2686539529716719e-05,
|
| 2597 |
+
"loss": 0.7908,
|
| 2598 |
+
"step": 4050
|
| 2599 |
+
},
|
| 2600 |
+
{
|
| 2601 |
+
"epoch": 1.12,
|
| 2602 |
+
"learning_rate": 1.2668024439918533e-05,
|
| 2603 |
+
"loss": 0.7989,
|
| 2604 |
+
"step": 4060
|
| 2605 |
+
},
|
| 2606 |
+
{
|
| 2607 |
+
"epoch": 1.12,
|
| 2608 |
+
"learning_rate": 1.2649509350120348e-05,
|
| 2609 |
+
"loss": 0.792,
|
| 2610 |
+
"step": 4070
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 1.12,
|
| 2614 |
+
"learning_rate": 1.2630994260322165e-05,
|
| 2615 |
+
"loss": 0.8028,
|
| 2616 |
+
"step": 4080
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 1.13,
|
| 2620 |
+
"learning_rate": 1.261247917052398e-05,
|
| 2621 |
+
"loss": 0.7886,
|
| 2622 |
+
"step": 4090
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 1.13,
|
| 2626 |
+
"learning_rate": 1.2593964080725794e-05,
|
| 2627 |
+
"loss": 0.807,
|
| 2628 |
+
"step": 4100
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 1.13,
|
| 2632 |
+
"learning_rate": 1.2575448990927609e-05,
|
| 2633 |
+
"loss": 0.8166,
|
| 2634 |
+
"step": 4110
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 1.13,
|
| 2638 |
+
"learning_rate": 1.2556933901129422e-05,
|
| 2639 |
+
"loss": 0.802,
|
| 2640 |
+
"step": 4120
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"epoch": 1.14,
|
| 2644 |
+
"learning_rate": 1.2538418811331236e-05,
|
| 2645 |
+
"loss": 0.8019,
|
| 2646 |
+
"step": 4130
|
| 2647 |
+
},
|
| 2648 |
+
{
|
| 2649 |
+
"epoch": 1.14,
|
| 2650 |
+
"learning_rate": 1.2519903721533051e-05,
|
| 2651 |
+
"loss": 0.8099,
|
| 2652 |
+
"step": 4140
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 1.14,
|
| 2656 |
+
"learning_rate": 1.2501388631734866e-05,
|
| 2657 |
+
"loss": 0.8052,
|
| 2658 |
+
"step": 4150
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 1.14,
|
| 2662 |
+
"learning_rate": 1.248287354193668e-05,
|
| 2663 |
+
"loss": 0.8015,
|
| 2664 |
+
"step": 4160
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 1.15,
|
| 2668 |
+
"learning_rate": 1.2464358452138493e-05,
|
| 2669 |
+
"loss": 0.797,
|
| 2670 |
+
"step": 4170
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 1.15,
|
| 2674 |
+
"learning_rate": 1.2445843362340308e-05,
|
| 2675 |
+
"loss": 0.7982,
|
| 2676 |
+
"step": 4180
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 1.15,
|
| 2680 |
+
"learning_rate": 1.2427328272542123e-05,
|
| 2681 |
+
"loss": 0.7919,
|
| 2682 |
+
"step": 4190
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 1.16,
|
| 2686 |
+
"learning_rate": 1.2408813182743937e-05,
|
| 2687 |
+
"loss": 0.7962,
|
| 2688 |
+
"step": 4200
|
| 2689 |
+
},
|
| 2690 |
+
{
|
| 2691 |
+
"epoch": 1.16,
|
| 2692 |
+
"eval_loss": 0.8028302192687988,
|
| 2693 |
+
"eval_runtime": 98.7487,
|
| 2694 |
+
"eval_samples_per_second": 10.127,
|
| 2695 |
+
"eval_steps_per_second": 0.638,
|
| 2696 |
+
"step": 4200
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 1.16,
|
| 2700 |
+
"learning_rate": 1.2390298092945752e-05,
|
| 2701 |
+
"loss": 0.7912,
|
| 2702 |
+
"step": 4210
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 1.16,
|
| 2706 |
+
"learning_rate": 1.2371783003147567e-05,
|
| 2707 |
+
"loss": 0.7894,
|
| 2708 |
+
"step": 4220
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 1.16,
|
| 2712 |
+
"learning_rate": 1.235326791334938e-05,
|
| 2713 |
+
"loss": 0.7847,
|
| 2714 |
+
"step": 4230
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 1.17,
|
| 2718 |
+
"learning_rate": 1.2334752823551195e-05,
|
| 2719 |
+
"loss": 0.8126,
|
| 2720 |
+
"step": 4240
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 1.17,
|
| 2724 |
+
"learning_rate": 1.231623773375301e-05,
|
| 2725 |
+
"loss": 0.7908,
|
| 2726 |
+
"step": 4250
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 1.17,
|
| 2730 |
+
"learning_rate": 1.2297722643954824e-05,
|
| 2731 |
+
"loss": 0.8084,
|
| 2732 |
+
"step": 4260
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 1.17,
|
| 2736 |
+
"learning_rate": 1.2279207554156639e-05,
|
| 2737 |
+
"loss": 0.7963,
|
| 2738 |
+
"step": 4270
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 1.18,
|
| 2742 |
+
"learning_rate": 1.2260692464358452e-05,
|
| 2743 |
+
"loss": 0.8031,
|
| 2744 |
+
"step": 4280
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 1.18,
|
| 2748 |
+
"learning_rate": 1.2242177374560266e-05,
|
| 2749 |
+
"loss": 0.7843,
|
| 2750 |
+
"step": 4290
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 1.18,
|
| 2754 |
+
"learning_rate": 1.2223662284762081e-05,
|
| 2755 |
+
"loss": 0.8006,
|
| 2756 |
+
"step": 4300
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 1.19,
|
| 2760 |
+
"learning_rate": 1.2205147194963896e-05,
|
| 2761 |
+
"loss": 0.8026,
|
| 2762 |
+
"step": 4310
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 1.19,
|
| 2766 |
+
"learning_rate": 1.2186632105165712e-05,
|
| 2767 |
+
"loss": 0.7994,
|
| 2768 |
+
"step": 4320
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 1.19,
|
| 2772 |
+
"learning_rate": 1.2168117015367527e-05,
|
| 2773 |
+
"loss": 0.8066,
|
| 2774 |
+
"step": 4330
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 1.19,
|
| 2778 |
+
"learning_rate": 1.2149601925569341e-05,
|
| 2779 |
+
"loss": 0.8063,
|
| 2780 |
+
"step": 4340
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 1.2,
|
| 2784 |
+
"learning_rate": 1.2131086835771156e-05,
|
| 2785 |
+
"loss": 0.8131,
|
| 2786 |
+
"step": 4350
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 1.2,
|
| 2790 |
+
"learning_rate": 1.2112571745972969e-05,
|
| 2791 |
+
"loss": 0.7985,
|
| 2792 |
+
"step": 4360
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 1.2,
|
| 2796 |
+
"learning_rate": 1.2094056656174784e-05,
|
| 2797 |
+
"loss": 0.8126,
|
| 2798 |
+
"step": 4370
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 1.21,
|
| 2802 |
+
"learning_rate": 1.2075541566376599e-05,
|
| 2803 |
+
"loss": 0.8088,
|
| 2804 |
+
"step": 4380
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 1.21,
|
| 2808 |
+
"learning_rate": 1.2057026476578413e-05,
|
| 2809 |
+
"loss": 0.7983,
|
| 2810 |
+
"step": 4390
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 1.21,
|
| 2814 |
+
"learning_rate": 1.2038511386780228e-05,
|
| 2815 |
+
"loss": 0.8059,
|
| 2816 |
+
"step": 4400
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 1.21,
|
| 2820 |
+
"eval_loss": 0.8016332983970642,
|
| 2821 |
+
"eval_runtime": 98.7551,
|
| 2822 |
+
"eval_samples_per_second": 10.126,
|
| 2823 |
+
"eval_steps_per_second": 0.638,
|
| 2824 |
+
"step": 4400
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 1.21,
|
| 2828 |
+
"learning_rate": 1.2019996296982041e-05,
|
| 2829 |
+
"loss": 0.7998,
|
| 2830 |
+
"step": 4410
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 1.22,
|
| 2834 |
+
"learning_rate": 1.2001481207183856e-05,
|
| 2835 |
+
"loss": 0.8014,
|
| 2836 |
+
"step": 4420
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"epoch": 1.22,
|
| 2840 |
+
"learning_rate": 1.198296611738567e-05,
|
| 2841 |
+
"loss": 0.7918,
|
| 2842 |
+
"step": 4430
|
| 2843 |
+
},
|
| 2844 |
+
{
|
| 2845 |
+
"epoch": 1.22,
|
| 2846 |
+
"learning_rate": 1.1964451027587485e-05,
|
| 2847 |
+
"loss": 0.8018,
|
| 2848 |
+
"step": 4440
|
| 2849 |
+
},
|
| 2850 |
+
{
|
| 2851 |
+
"epoch": 1.22,
|
| 2852 |
+
"learning_rate": 1.19459359377893e-05,
|
| 2853 |
+
"loss": 0.7924,
|
| 2854 |
+
"step": 4450
|
| 2855 |
+
},
|
| 2856 |
+
{
|
| 2857 |
+
"epoch": 1.23,
|
| 2858 |
+
"learning_rate": 1.1927420847991114e-05,
|
| 2859 |
+
"loss": 0.8068,
|
| 2860 |
+
"step": 4460
|
| 2861 |
+
},
|
| 2862 |
+
{
|
| 2863 |
+
"epoch": 1.23,
|
| 2864 |
+
"learning_rate": 1.1908905758192927e-05,
|
| 2865 |
+
"loss": 0.7883,
|
| 2866 |
+
"step": 4470
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 1.23,
|
| 2870 |
+
"learning_rate": 1.1890390668394742e-05,
|
| 2871 |
+
"loss": 0.7981,
|
| 2872 |
+
"step": 4480
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 1.24,
|
| 2876 |
+
"learning_rate": 1.1871875578596557e-05,
|
| 2877 |
+
"loss": 0.807,
|
| 2878 |
+
"step": 4490
|
| 2879 |
+
},
|
| 2880 |
+
{
|
| 2881 |
+
"epoch": 1.24,
|
| 2882 |
+
"learning_rate": 1.1853360488798371e-05,
|
| 2883 |
+
"loss": 0.7953,
|
| 2884 |
+
"step": 4500
|
| 2885 |
+
},
|
| 2886 |
+
{
|
| 2887 |
+
"epoch": 1.24,
|
| 2888 |
+
"learning_rate": 1.1834845399000186e-05,
|
| 2889 |
+
"loss": 0.7949,
|
| 2890 |
+
"step": 4510
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 1.24,
|
| 2894 |
+
"learning_rate": 1.1816330309201999e-05,
|
| 2895 |
+
"loss": 0.8015,
|
| 2896 |
+
"step": 4520
|
| 2897 |
+
},
|
| 2898 |
+
{
|
| 2899 |
+
"epoch": 1.25,
|
| 2900 |
+
"learning_rate": 1.1797815219403814e-05,
|
| 2901 |
+
"loss": 0.7967,
|
| 2902 |
+
"step": 4530
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"epoch": 1.25,
|
| 2906 |
+
"learning_rate": 1.1779300129605629e-05,
|
| 2907 |
+
"loss": 0.782,
|
| 2908 |
+
"step": 4540
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 1.25,
|
| 2912 |
+
"learning_rate": 1.1760785039807443e-05,
|
| 2913 |
+
"loss": 0.8061,
|
| 2914 |
+
"step": 4550
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 1.25,
|
| 2918 |
+
"learning_rate": 1.1742269950009258e-05,
|
| 2919 |
+
"loss": 0.7991,
|
| 2920 |
+
"step": 4560
|
| 2921 |
+
},
|
| 2922 |
+
{
|
| 2923 |
+
"epoch": 1.26,
|
| 2924 |
+
"learning_rate": 1.1723754860211074e-05,
|
| 2925 |
+
"loss": 0.8064,
|
| 2926 |
+
"step": 4570
|
| 2927 |
+
},
|
| 2928 |
+
{
|
| 2929 |
+
"epoch": 1.26,
|
| 2930 |
+
"learning_rate": 1.1705239770412889e-05,
|
| 2931 |
+
"loss": 0.8071,
|
| 2932 |
+
"step": 4580
|
| 2933 |
+
},
|
| 2934 |
+
{
|
| 2935 |
+
"epoch": 1.26,
|
| 2936 |
+
"learning_rate": 1.1686724680614702e-05,
|
| 2937 |
+
"loss": 0.8122,
|
| 2938 |
+
"step": 4590
|
| 2939 |
+
},
|
| 2940 |
+
{
|
| 2941 |
+
"epoch": 1.27,
|
| 2942 |
+
"learning_rate": 1.1668209590816517e-05,
|
| 2943 |
+
"loss": 0.8063,
|
| 2944 |
+
"step": 4600
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"epoch": 1.27,
|
| 2948 |
+
"eval_loss": 0.8010302186012268,
|
| 2949 |
+
"eval_runtime": 98.6771,
|
| 2950 |
+
"eval_samples_per_second": 10.134,
|
| 2951 |
+
"eval_steps_per_second": 0.638,
|
| 2952 |
+
"step": 4600
|
| 2953 |
+
},
|
| 2954 |
+
{
|
| 2955 |
+
"epoch": 1.27,
|
| 2956 |
+
"learning_rate": 1.1649694501018331e-05,
|
| 2957 |
+
"loss": 0.8053,
|
| 2958 |
+
"step": 4610
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 1.27,
|
| 2962 |
+
"learning_rate": 1.1631179411220146e-05,
|
| 2963 |
+
"loss": 0.8058,
|
| 2964 |
+
"step": 4620
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 1.27,
|
| 2968 |
+
"learning_rate": 1.161266432142196e-05,
|
| 2969 |
+
"loss": 0.7906,
|
| 2970 |
+
"step": 4630
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 1.28,
|
| 2974 |
+
"learning_rate": 1.1594149231623775e-05,
|
| 2975 |
+
"loss": 0.7955,
|
| 2976 |
+
"step": 4640
|
| 2977 |
+
},
|
| 2978 |
+
{
|
| 2979 |
+
"epoch": 1.28,
|
| 2980 |
+
"learning_rate": 1.1575634141825588e-05,
|
| 2981 |
+
"loss": 0.7967,
|
| 2982 |
+
"step": 4650
|
| 2983 |
+
},
|
| 2984 |
+
{
|
| 2985 |
+
"epoch": 1.28,
|
| 2986 |
+
"learning_rate": 1.1557119052027403e-05,
|
| 2987 |
+
"loss": 0.8047,
|
| 2988 |
+
"step": 4660
|
| 2989 |
+
},
|
| 2990 |
+
{
|
| 2991 |
+
"epoch": 1.28,
|
| 2992 |
+
"learning_rate": 1.1538603962229218e-05,
|
| 2993 |
+
"loss": 0.7835,
|
| 2994 |
+
"step": 4670
|
| 2995 |
+
},
|
| 2996 |
+
{
|
| 2997 |
+
"epoch": 1.29,
|
| 2998 |
+
"learning_rate": 1.1520088872431033e-05,
|
| 2999 |
+
"loss": 0.8087,
|
| 3000 |
+
"step": 4680
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 1.29,
|
| 3004 |
+
"learning_rate": 1.1501573782632847e-05,
|
| 3005 |
+
"loss": 0.787,
|
| 3006 |
+
"step": 4690
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 1.29,
|
| 3010 |
+
"learning_rate": 1.148305869283466e-05,
|
| 3011 |
+
"loss": 0.804,
|
| 3012 |
+
"step": 4700
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 1.3,
|
| 3016 |
+
"learning_rate": 1.1464543603036475e-05,
|
| 3017 |
+
"loss": 0.8023,
|
| 3018 |
+
"step": 4710
|
| 3019 |
+
},
|
| 3020 |
+
{
|
| 3021 |
+
"epoch": 1.3,
|
| 3022 |
+
"learning_rate": 1.144602851323829e-05,
|
| 3023 |
+
"loss": 0.8001,
|
| 3024 |
+
"step": 4720
|
| 3025 |
+
},
|
| 3026 |
+
{
|
| 3027 |
+
"epoch": 1.3,
|
| 3028 |
+
"learning_rate": 1.1427513423440104e-05,
|
| 3029 |
+
"loss": 0.8026,
|
| 3030 |
+
"step": 4730
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 1.3,
|
| 3034 |
+
"learning_rate": 1.1408998333641919e-05,
|
| 3035 |
+
"loss": 0.7916,
|
| 3036 |
+
"step": 4740
|
| 3037 |
+
},
|
| 3038 |
+
{
|
| 3039 |
+
"epoch": 1.31,
|
| 3040 |
+
"learning_rate": 1.1390483243843734e-05,
|
| 3041 |
+
"loss": 0.7939,
|
| 3042 |
+
"step": 4750
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 1.31,
|
| 3046 |
+
"learning_rate": 1.1371968154045547e-05,
|
| 3047 |
+
"loss": 0.8032,
|
| 3048 |
+
"step": 4760
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 1.31,
|
| 3052 |
+
"learning_rate": 1.1353453064247361e-05,
|
| 3053 |
+
"loss": 0.8013,
|
| 3054 |
+
"step": 4770
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 1.32,
|
| 3058 |
+
"learning_rate": 1.1334937974449176e-05,
|
| 3059 |
+
"loss": 0.8026,
|
| 3060 |
+
"step": 4780
|
| 3061 |
+
},
|
| 3062 |
+
{
|
| 3063 |
+
"epoch": 1.32,
|
| 3064 |
+
"learning_rate": 1.131642288465099e-05,
|
| 3065 |
+
"loss": 0.793,
|
| 3066 |
+
"step": 4790
|
| 3067 |
+
},
|
| 3068 |
+
{
|
| 3069 |
+
"epoch": 1.32,
|
| 3070 |
+
"learning_rate": 1.1297907794852805e-05,
|
| 3071 |
+
"loss": 0.7993,
|
| 3072 |
+
"step": 4800
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 1.32,
|
| 3076 |
+
"eval_loss": 0.8001218438148499,
|
| 3077 |
+
"eval_runtime": 98.8463,
|
| 3078 |
+
"eval_samples_per_second": 10.117,
|
| 3079 |
+
"eval_steps_per_second": 0.637,
|
| 3080 |
+
"step": 4800
|
| 3081 |
+
},
|
| 3082 |
+
{
|
| 3083 |
+
"epoch": 1.32,
|
| 3084 |
+
"learning_rate": 1.1279392705054622e-05,
|
| 3085 |
+
"loss": 0.8054,
|
| 3086 |
+
"step": 4810
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 1.33,
|
| 3090 |
+
"learning_rate": 1.1260877615256437e-05,
|
| 3091 |
+
"loss": 0.792,
|
| 3092 |
+
"step": 4820
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 1.33,
|
| 3096 |
+
"learning_rate": 1.124236252545825e-05,
|
| 3097 |
+
"loss": 0.7997,
|
| 3098 |
+
"step": 4830
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 1.33,
|
| 3102 |
+
"learning_rate": 1.1223847435660064e-05,
|
| 3103 |
+
"loss": 0.7973,
|
| 3104 |
+
"step": 4840
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 1.33,
|
| 3108 |
+
"learning_rate": 1.1205332345861879e-05,
|
| 3109 |
+
"loss": 0.7994,
|
| 3110 |
+
"step": 4850
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 1.34,
|
| 3114 |
+
"learning_rate": 1.1186817256063694e-05,
|
| 3115 |
+
"loss": 0.7923,
|
| 3116 |
+
"step": 4860
|
| 3117 |
+
},
|
| 3118 |
+
{
|
| 3119 |
+
"epoch": 1.34,
|
| 3120 |
+
"learning_rate": 1.1168302166265508e-05,
|
| 3121 |
+
"loss": 0.7959,
|
| 3122 |
+
"step": 4870
|
| 3123 |
+
},
|
| 3124 |
+
{
|
| 3125 |
+
"epoch": 1.34,
|
| 3126 |
+
"learning_rate": 1.1149787076467323e-05,
|
| 3127 |
+
"loss": 0.794,
|
| 3128 |
+
"step": 4880
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 1.35,
|
| 3132 |
+
"learning_rate": 1.1131271986669136e-05,
|
| 3133 |
+
"loss": 0.796,
|
| 3134 |
+
"step": 4890
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 1.35,
|
| 3138 |
+
"learning_rate": 1.111275689687095e-05,
|
| 3139 |
+
"loss": 0.8023,
|
| 3140 |
+
"step": 4900
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 1.35,
|
| 3144 |
+
"learning_rate": 1.1094241807072765e-05,
|
| 3145 |
+
"loss": 0.7972,
|
| 3146 |
+
"step": 4910
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 1.35,
|
| 3150 |
+
"learning_rate": 1.107572671727458e-05,
|
| 3151 |
+
"loss": 0.811,
|
| 3152 |
+
"step": 4920
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 1.36,
|
| 3156 |
+
"learning_rate": 1.1057211627476395e-05,
|
| 3157 |
+
"loss": 0.7931,
|
| 3158 |
+
"step": 4930
|
| 3159 |
+
},
|
| 3160 |
+
{
|
| 3161 |
+
"epoch": 1.36,
|
| 3162 |
+
"learning_rate": 1.1038696537678208e-05,
|
| 3163 |
+
"loss": 0.7933,
|
| 3164 |
+
"step": 4940
|
| 3165 |
+
},
|
| 3166 |
+
{
|
| 3167 |
+
"epoch": 1.36,
|
| 3168 |
+
"learning_rate": 1.1020181447880022e-05,
|
| 3169 |
+
"loss": 0.7961,
|
| 3170 |
+
"step": 4950
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 1.36,
|
| 3174 |
+
"learning_rate": 1.1001666358081837e-05,
|
| 3175 |
+
"loss": 0.8124,
|
| 3176 |
+
"step": 4960
|
| 3177 |
+
},
|
| 3178 |
+
{
|
| 3179 |
+
"epoch": 1.37,
|
| 3180 |
+
"learning_rate": 1.0983151268283652e-05,
|
| 3181 |
+
"loss": 0.7983,
|
| 3182 |
+
"step": 4970
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 1.37,
|
| 3186 |
+
"learning_rate": 1.0964636178485467e-05,
|
| 3187 |
+
"loss": 0.7932,
|
| 3188 |
+
"step": 4980
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 1.37,
|
| 3192 |
+
"learning_rate": 1.0946121088687281e-05,
|
| 3193 |
+
"loss": 0.8005,
|
| 3194 |
+
"step": 4990
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 1.38,
|
| 3198 |
+
"learning_rate": 1.0927605998889094e-05,
|
| 3199 |
+
"loss": 0.7956,
|
| 3200 |
+
"step": 5000
|
| 3201 |
+
},
|
| 3202 |
+
{
|
| 3203 |
+
"epoch": 1.38,
|
| 3204 |
+
"eval_loss": 0.7999187111854553,
|
| 3205 |
+
"eval_runtime": 98.8229,
|
| 3206 |
+
"eval_samples_per_second": 10.119,
|
| 3207 |
+
"eval_steps_per_second": 0.638,
|
| 3208 |
+
"step": 5000
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 1.38,
|
| 3212 |
+
"learning_rate": 1.0909090909090909e-05,
|
| 3213 |
+
"loss": 0.795,
|
| 3214 |
+
"step": 5010
|
| 3215 |
+
},
|
| 3216 |
+
{
|
| 3217 |
+
"epoch": 1.38,
|
| 3218 |
+
"learning_rate": 1.0890575819292724e-05,
|
| 3219 |
+
"loss": 0.7998,
|
| 3220 |
+
"step": 5020
|
| 3221 |
+
},
|
| 3222 |
+
{
|
| 3223 |
+
"epoch": 1.38,
|
| 3224 |
+
"learning_rate": 1.0872060729494538e-05,
|
| 3225 |
+
"loss": 0.8049,
|
| 3226 |
+
"step": 5030
|
| 3227 |
+
},
|
| 3228 |
+
{
|
| 3229 |
+
"epoch": 1.39,
|
| 3230 |
+
"learning_rate": 1.0853545639696353e-05,
|
| 3231 |
+
"loss": 0.7942,
|
| 3232 |
+
"step": 5040
|
| 3233 |
+
},
|
| 3234 |
+
{
|
| 3235 |
+
"epoch": 1.39,
|
| 3236 |
+
"learning_rate": 1.083503054989817e-05,
|
| 3237 |
+
"loss": 0.7899,
|
| 3238 |
+
"step": 5050
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 1.39,
|
| 3242 |
+
"learning_rate": 1.0816515460099984e-05,
|
| 3243 |
+
"loss": 0.806,
|
| 3244 |
+
"step": 5060
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 1.39,
|
| 3248 |
+
"learning_rate": 1.0798000370301797e-05,
|
| 3249 |
+
"loss": 0.8026,
|
| 3250 |
+
"step": 5070
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 1.4,
|
| 3254 |
+
"learning_rate": 1.0779485280503612e-05,
|
| 3255 |
+
"loss": 0.7914,
|
| 3256 |
+
"step": 5080
|
| 3257 |
+
},
|
| 3258 |
+
{
|
| 3259 |
+
"epoch": 1.4,
|
| 3260 |
+
"learning_rate": 1.0760970190705426e-05,
|
| 3261 |
+
"loss": 0.7993,
|
| 3262 |
+
"step": 5090
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 1.4,
|
| 3266 |
+
"learning_rate": 1.0742455100907241e-05,
|
| 3267 |
+
"loss": 0.8064,
|
| 3268 |
+
"step": 5100
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 1.41,
|
| 3272 |
+
"learning_rate": 1.0723940011109056e-05,
|
| 3273 |
+
"loss": 0.7989,
|
| 3274 |
+
"step": 5110
|
| 3275 |
+
},
|
| 3276 |
+
{
|
| 3277 |
+
"epoch": 1.41,
|
| 3278 |
+
"learning_rate": 1.0705424921310869e-05,
|
| 3279 |
+
"loss": 0.8054,
|
| 3280 |
+
"step": 5120
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 1.41,
|
| 3284 |
+
"learning_rate": 1.0686909831512684e-05,
|
| 3285 |
+
"loss": 0.8007,
|
| 3286 |
+
"step": 5130
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 1.41,
|
| 3290 |
+
"learning_rate": 1.0668394741714498e-05,
|
| 3291 |
+
"loss": 0.7882,
|
| 3292 |
+
"step": 5140
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 1.42,
|
| 3296 |
+
"learning_rate": 1.0649879651916313e-05,
|
| 3297 |
+
"loss": 0.8137,
|
| 3298 |
+
"step": 5150
|
| 3299 |
+
},
|
| 3300 |
+
{
|
| 3301 |
+
"epoch": 1.42,
|
| 3302 |
+
"learning_rate": 1.0631364562118128e-05,
|
| 3303 |
+
"loss": 0.8,
|
| 3304 |
+
"step": 5160
|
| 3305 |
+
},
|
| 3306 |
+
{
|
| 3307 |
+
"epoch": 1.42,
|
| 3308 |
+
"learning_rate": 1.0612849472319942e-05,
|
| 3309 |
+
"loss": 0.7873,
|
| 3310 |
+
"step": 5170
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 1.43,
|
| 3314 |
+
"learning_rate": 1.0594334382521755e-05,
|
| 3315 |
+
"loss": 0.8002,
|
| 3316 |
+
"step": 5180
|
| 3317 |
+
},
|
| 3318 |
+
{
|
| 3319 |
+
"epoch": 1.43,
|
| 3320 |
+
"learning_rate": 1.057581929272357e-05,
|
| 3321 |
+
"loss": 0.8025,
|
| 3322 |
+
"step": 5190
|
| 3323 |
+
},
|
| 3324 |
+
{
|
| 3325 |
+
"epoch": 1.43,
|
| 3326 |
+
"learning_rate": 1.0557304202925385e-05,
|
| 3327 |
+
"loss": 0.8128,
|
| 3328 |
+
"step": 5200
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 1.43,
|
| 3332 |
+
"eval_loss": 0.7986959218978882,
|
| 3333 |
+
"eval_runtime": 98.7666,
|
| 3334 |
+
"eval_samples_per_second": 10.125,
|
| 3335 |
+
"eval_steps_per_second": 0.638,
|
| 3336 |
+
"step": 5200
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 1.43,
|
| 3340 |
+
"learning_rate": 1.05387891131272e-05,
|
| 3341 |
+
"loss": 0.7914,
|
| 3342 |
+
"step": 5210
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 1.44,
|
| 3346 |
+
"learning_rate": 1.0520274023329014e-05,
|
| 3347 |
+
"loss": 0.8034,
|
| 3348 |
+
"step": 5220
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 1.44,
|
| 3352 |
+
"learning_rate": 1.0501758933530827e-05,
|
| 3353 |
+
"loss": 0.7978,
|
| 3354 |
+
"step": 5230
|
| 3355 |
+
},
|
| 3356 |
+
{
|
| 3357 |
+
"epoch": 1.44,
|
| 3358 |
+
"learning_rate": 1.0483243843732642e-05,
|
| 3359 |
+
"loss": 0.798,
|
| 3360 |
+
"step": 5240
|
| 3361 |
+
},
|
| 3362 |
+
{
|
| 3363 |
+
"epoch": 1.44,
|
| 3364 |
+
"learning_rate": 1.0464728753934456e-05,
|
| 3365 |
+
"loss": 0.7996,
|
| 3366 |
+
"step": 5250
|
| 3367 |
+
},
|
| 3368 |
+
{
|
| 3369 |
+
"epoch": 1.45,
|
| 3370 |
+
"learning_rate": 1.0446213664136271e-05,
|
| 3371 |
+
"loss": 0.7986,
|
| 3372 |
+
"step": 5260
|
| 3373 |
+
},
|
| 3374 |
+
{
|
| 3375 |
+
"epoch": 1.45,
|
| 3376 |
+
"learning_rate": 1.0427698574338086e-05,
|
| 3377 |
+
"loss": 0.8057,
|
| 3378 |
+
"step": 5270
|
| 3379 |
+
},
|
| 3380 |
+
{
|
| 3381 |
+
"epoch": 1.45,
|
| 3382 |
+
"learning_rate": 1.04091834845399e-05,
|
| 3383 |
+
"loss": 0.7974,
|
| 3384 |
+
"step": 5280
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 1.46,
|
| 3388 |
+
"learning_rate": 1.0390668394741717e-05,
|
| 3389 |
+
"loss": 0.8033,
|
| 3390 |
+
"step": 5290
|
| 3391 |
+
},
|
| 3392 |
+
{
|
| 3393 |
+
"epoch": 1.46,
|
| 3394 |
+
"learning_rate": 1.0372153304943532e-05,
|
| 3395 |
+
"loss": 0.8002,
|
| 3396 |
+
"step": 5300
|
| 3397 |
+
},
|
| 3398 |
+
{
|
| 3399 |
+
"epoch": 1.46,
|
| 3400 |
+
"learning_rate": 1.0353638215145345e-05,
|
| 3401 |
+
"loss": 0.8013,
|
| 3402 |
+
"step": 5310
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 1.46,
|
| 3406 |
+
"learning_rate": 1.033512312534716e-05,
|
| 3407 |
+
"loss": 0.7841,
|
| 3408 |
+
"step": 5320
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 1.47,
|
| 3412 |
+
"learning_rate": 1.0316608035548974e-05,
|
| 3413 |
+
"loss": 0.7902,
|
| 3414 |
+
"step": 5330
|
| 3415 |
+
},
|
| 3416 |
+
{
|
| 3417 |
+
"epoch": 1.47,
|
| 3418 |
+
"learning_rate": 1.0298092945750789e-05,
|
| 3419 |
+
"loss": 0.8043,
|
| 3420 |
+
"step": 5340
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 1.47,
|
| 3424 |
+
"learning_rate": 1.0279577855952603e-05,
|
| 3425 |
+
"loss": 0.7913,
|
| 3426 |
+
"step": 5350
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 1.47,
|
| 3430 |
+
"learning_rate": 1.0261062766154416e-05,
|
| 3431 |
+
"loss": 0.792,
|
| 3432 |
+
"step": 5360
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 1.48,
|
| 3436 |
+
"learning_rate": 1.0242547676356231e-05,
|
| 3437 |
+
"loss": 0.7938,
|
| 3438 |
+
"step": 5370
|
| 3439 |
+
},
|
| 3440 |
+
{
|
| 3441 |
+
"epoch": 1.48,
|
| 3442 |
+
"learning_rate": 1.0224032586558046e-05,
|
| 3443 |
+
"loss": 0.8014,
|
| 3444 |
+
"step": 5380
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 1.48,
|
| 3448 |
+
"learning_rate": 1.020551749675986e-05,
|
| 3449 |
+
"loss": 0.7939,
|
| 3450 |
+
"step": 5390
|
| 3451 |
+
},
|
| 3452 |
+
{
|
| 3453 |
+
"epoch": 1.49,
|
| 3454 |
+
"learning_rate": 1.0187002406961675e-05,
|
| 3455 |
+
"loss": 0.793,
|
| 3456 |
+
"step": 5400
|
| 3457 |
+
},
|
| 3458 |
+
{
|
| 3459 |
+
"epoch": 1.49,
|
| 3460 |
+
"eval_loss": 0.7980086207389832,
|
| 3461 |
+
"eval_runtime": 98.7281,
|
| 3462 |
+
"eval_samples_per_second": 10.129,
|
| 3463 |
+
"eval_steps_per_second": 0.638,
|
| 3464 |
+
"step": 5400
|
| 3465 |
+
},
|
| 3466 |
+
{
|
| 3467 |
+
"epoch": 1.49,
|
| 3468 |
+
"learning_rate": 1.016848731716349e-05,
|
| 3469 |
+
"loss": 0.7972,
|
| 3470 |
+
"step": 5410
|
| 3471 |
+
},
|
| 3472 |
+
{
|
| 3473 |
+
"epoch": 1.49,
|
| 3474 |
+
"learning_rate": 1.0149972227365303e-05,
|
| 3475 |
+
"loss": 0.7937,
|
| 3476 |
+
"step": 5420
|
| 3477 |
+
},
|
| 3478 |
+
{
|
| 3479 |
+
"epoch": 1.49,
|
| 3480 |
+
"learning_rate": 1.0131457137567118e-05,
|
| 3481 |
+
"loss": 0.7918,
|
| 3482 |
+
"step": 5430
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 1.5,
|
| 3486 |
+
"learning_rate": 1.0112942047768932e-05,
|
| 3487 |
+
"loss": 0.816,
|
| 3488 |
+
"step": 5440
|
| 3489 |
+
},
|
| 3490 |
+
{
|
| 3491 |
+
"epoch": 1.5,
|
| 3492 |
+
"learning_rate": 1.0094426957970747e-05,
|
| 3493 |
+
"loss": 0.7963,
|
| 3494 |
+
"step": 5450
|
| 3495 |
+
},
|
| 3496 |
+
{
|
| 3497 |
+
"epoch": 1.5,
|
| 3498 |
+
"learning_rate": 1.0075911868172562e-05,
|
| 3499 |
+
"loss": 0.8012,
|
| 3500 |
+
"step": 5460
|
| 3501 |
+
},
|
| 3502 |
+
{
|
| 3503 |
+
"epoch": 1.51,
|
| 3504 |
+
"learning_rate": 1.0057396778374375e-05,
|
| 3505 |
+
"loss": 0.801,
|
| 3506 |
+
"step": 5470
|
| 3507 |
+
},
|
| 3508 |
+
{
|
| 3509 |
+
"epoch": 1.51,
|
| 3510 |
+
"learning_rate": 1.003888168857619e-05,
|
| 3511 |
+
"loss": 0.7947,
|
| 3512 |
+
"step": 5480
|
| 3513 |
+
},
|
| 3514 |
+
{
|
| 3515 |
+
"epoch": 1.51,
|
| 3516 |
+
"learning_rate": 1.0020366598778004e-05,
|
| 3517 |
+
"loss": 0.8022,
|
| 3518 |
+
"step": 5490
|
| 3519 |
+
},
|
| 3520 |
+
{
|
| 3521 |
+
"epoch": 1.51,
|
| 3522 |
+
"learning_rate": 1.0001851508979819e-05,
|
| 3523 |
+
"loss": 0.7928,
|
| 3524 |
+
"step": 5500
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 1.52,
|
| 3528 |
+
"learning_rate": 9.983336419181633e-06,
|
| 3529 |
+
"loss": 0.7971,
|
| 3530 |
+
"step": 5510
|
| 3531 |
+
},
|
| 3532 |
+
{
|
| 3533 |
+
"epoch": 1.52,
|
| 3534 |
+
"learning_rate": 9.964821329383448e-06,
|
| 3535 |
+
"loss": 0.8007,
|
| 3536 |
+
"step": 5520
|
| 3537 |
+
},
|
| 3538 |
+
{
|
| 3539 |
+
"epoch": 1.52,
|
| 3540 |
+
"learning_rate": 9.946306239585263e-06,
|
| 3541 |
+
"loss": 0.796,
|
| 3542 |
+
"step": 5530
|
| 3543 |
+
},
|
| 3544 |
+
{
|
| 3545 |
+
"epoch": 1.52,
|
| 3546 |
+
"learning_rate": 9.927791149787077e-06,
|
| 3547 |
+
"loss": 0.7948,
|
| 3548 |
+
"step": 5540
|
| 3549 |
+
},
|
| 3550 |
+
{
|
| 3551 |
+
"epoch": 1.53,
|
| 3552 |
+
"learning_rate": 9.909276059988892e-06,
|
| 3553 |
+
"loss": 0.7928,
|
| 3554 |
+
"step": 5550
|
| 3555 |
+
},
|
| 3556 |
+
{
|
| 3557 |
+
"epoch": 1.53,
|
| 3558 |
+
"learning_rate": 9.890760970190705e-06,
|
| 3559 |
+
"loss": 0.7943,
|
| 3560 |
+
"step": 5560
|
| 3561 |
+
},
|
| 3562 |
+
{
|
| 3563 |
+
"epoch": 1.53,
|
| 3564 |
+
"learning_rate": 9.87224588039252e-06,
|
| 3565 |
+
"loss": 0.7896,
|
| 3566 |
+
"step": 5570
|
| 3567 |
+
},
|
| 3568 |
+
{
|
| 3569 |
+
"epoch": 1.54,
|
| 3570 |
+
"learning_rate": 9.853730790594335e-06,
|
| 3571 |
+
"loss": 0.7975,
|
| 3572 |
+
"step": 5580
|
| 3573 |
+
},
|
| 3574 |
+
{
|
| 3575 |
+
"epoch": 1.54,
|
| 3576 |
+
"learning_rate": 9.835215700796151e-06,
|
| 3577 |
+
"loss": 0.7944,
|
| 3578 |
+
"step": 5590
|
| 3579 |
+
},
|
| 3580 |
+
{
|
| 3581 |
+
"epoch": 1.54,
|
| 3582 |
+
"learning_rate": 9.816700610997964e-06,
|
| 3583 |
+
"loss": 0.7968,
|
| 3584 |
+
"step": 5600
|
| 3585 |
+
},
|
| 3586 |
+
{
|
| 3587 |
+
"epoch": 1.54,
|
| 3588 |
+
"eval_loss": 0.7978225350379944,
|
| 3589 |
+
"eval_runtime": 98.8162,
|
| 3590 |
+
"eval_samples_per_second": 10.12,
|
| 3591 |
+
"eval_steps_per_second": 0.638,
|
| 3592 |
+
"step": 5600
|
| 3593 |
+
},
|
| 3594 |
+
{
|
| 3595 |
+
"epoch": 1.54,
|
| 3596 |
+
"learning_rate": 9.798185521199779e-06,
|
| 3597 |
+
"loss": 0.7939,
|
| 3598 |
+
"step": 5610
|
| 3599 |
+
},
|
| 3600 |
+
{
|
| 3601 |
+
"epoch": 1.55,
|
| 3602 |
+
"learning_rate": 9.779670431401593e-06,
|
| 3603 |
+
"loss": 0.7958,
|
| 3604 |
+
"step": 5620
|
| 3605 |
+
},
|
| 3606 |
+
{
|
| 3607 |
+
"epoch": 1.55,
|
| 3608 |
+
"learning_rate": 9.761155341603408e-06,
|
| 3609 |
+
"loss": 0.8013,
|
| 3610 |
+
"step": 5630
|
| 3611 |
+
},
|
| 3612 |
+
{
|
| 3613 |
+
"epoch": 1.55,
|
| 3614 |
+
"learning_rate": 9.742640251805223e-06,
|
| 3615 |
+
"loss": 0.8005,
|
| 3616 |
+
"step": 5640
|
| 3617 |
+
},
|
| 3618 |
+
{
|
| 3619 |
+
"epoch": 1.55,
|
| 3620 |
+
"learning_rate": 9.724125162007036e-06,
|
| 3621 |
+
"loss": 0.809,
|
| 3622 |
+
"step": 5650
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 1.56,
|
| 3626 |
+
"learning_rate": 9.70561007220885e-06,
|
| 3627 |
+
"loss": 0.8062,
|
| 3628 |
+
"step": 5660
|
| 3629 |
+
},
|
| 3630 |
+
{
|
| 3631 |
+
"epoch": 1.56,
|
| 3632 |
+
"learning_rate": 9.687094982410665e-06,
|
| 3633 |
+
"loss": 0.7933,
|
| 3634 |
+
"step": 5670
|
| 3635 |
+
},
|
| 3636 |
+
{
|
| 3637 |
+
"epoch": 1.56,
|
| 3638 |
+
"learning_rate": 9.66857989261248e-06,
|
| 3639 |
+
"loss": 0.7945,
|
| 3640 |
+
"step": 5680
|
| 3641 |
+
},
|
| 3642 |
+
{
|
| 3643 |
+
"epoch": 1.57,
|
| 3644 |
+
"learning_rate": 9.650064802814294e-06,
|
| 3645 |
+
"loss": 0.8052,
|
| 3646 |
+
"step": 5690
|
| 3647 |
+
},
|
| 3648 |
+
{
|
| 3649 |
+
"epoch": 1.57,
|
| 3650 |
+
"learning_rate": 9.631549713016109e-06,
|
| 3651 |
+
"loss": 0.8031,
|
| 3652 |
+
"step": 5700
|
| 3653 |
+
},
|
| 3654 |
+
{
|
| 3655 |
+
"epoch": 1.57,
|
| 3656 |
+
"learning_rate": 9.613034623217924e-06,
|
| 3657 |
+
"loss": 0.786,
|
| 3658 |
+
"step": 5710
|
| 3659 |
+
},
|
| 3660 |
+
{
|
| 3661 |
+
"epoch": 1.57,
|
| 3662 |
+
"learning_rate": 9.594519533419739e-06,
|
| 3663 |
+
"loss": 0.7982,
|
| 3664 |
+
"step": 5720
|
| 3665 |
+
},
|
| 3666 |
+
{
|
| 3667 |
+
"epoch": 1.58,
|
| 3668 |
+
"learning_rate": 9.576004443621553e-06,
|
| 3669 |
+
"loss": 0.7946,
|
| 3670 |
+
"step": 5730
|
| 3671 |
+
},
|
| 3672 |
+
{
|
| 3673 |
+
"epoch": 1.58,
|
| 3674 |
+
"learning_rate": 9.557489353823368e-06,
|
| 3675 |
+
"loss": 0.7991,
|
| 3676 |
+
"step": 5740
|
| 3677 |
+
},
|
| 3678 |
+
{
|
| 3679 |
+
"epoch": 1.58,
|
| 3680 |
+
"learning_rate": 9.538974264025181e-06,
|
| 3681 |
+
"loss": 0.7918,
|
| 3682 |
+
"step": 5750
|
| 3683 |
+
},
|
| 3684 |
+
{
|
| 3685 |
+
"epoch": 1.58,
|
| 3686 |
+
"learning_rate": 9.520459174226996e-06,
|
| 3687 |
+
"loss": 0.7963,
|
| 3688 |
+
"step": 5760
|
| 3689 |
+
},
|
| 3690 |
+
{
|
| 3691 |
+
"epoch": 1.59,
|
| 3692 |
+
"learning_rate": 9.50194408442881e-06,
|
| 3693 |
+
"loss": 0.7871,
|
| 3694 |
+
"step": 5770
|
| 3695 |
+
},
|
| 3696 |
+
{
|
| 3697 |
+
"epoch": 1.59,
|
| 3698 |
+
"learning_rate": 9.483428994630625e-06,
|
| 3699 |
+
"loss": 0.7944,
|
| 3700 |
+
"step": 5780
|
| 3701 |
+
},
|
| 3702 |
+
{
|
| 3703 |
+
"epoch": 1.59,
|
| 3704 |
+
"learning_rate": 9.46491390483244e-06,
|
| 3705 |
+
"loss": 0.7986,
|
| 3706 |
+
"step": 5790
|
| 3707 |
+
},
|
| 3708 |
+
{
|
| 3709 |
+
"epoch": 1.6,
|
| 3710 |
+
"learning_rate": 9.446398815034253e-06,
|
| 3711 |
+
"loss": 0.7943,
|
| 3712 |
+
"step": 5800
|
| 3713 |
+
},
|
| 3714 |
+
{
|
| 3715 |
+
"epoch": 1.6,
|
| 3716 |
+
"eval_loss": 0.7971388101577759,
|
| 3717 |
+
"eval_runtime": 98.783,
|
| 3718 |
+
"eval_samples_per_second": 10.123,
|
| 3719 |
+
"eval_steps_per_second": 0.638,
|
| 3720 |
+
"step": 5800
|
| 3721 |
+
},
|
| 3722 |
+
{
|
| 3723 |
+
"epoch": 1.6,
|
| 3724 |
+
"learning_rate": 9.427883725236067e-06,
|
| 3725 |
+
"loss": 0.804,
|
| 3726 |
+
"step": 5810
|
| 3727 |
+
},
|
| 3728 |
+
{
|
| 3729 |
+
"epoch": 1.6,
|
| 3730 |
+
"learning_rate": 9.409368635437882e-06,
|
| 3731 |
+
"loss": 0.8041,
|
| 3732 |
+
"step": 5820
|
| 3733 |
+
},
|
| 3734 |
+
{
|
| 3735 |
+
"epoch": 1.6,
|
| 3736 |
+
"learning_rate": 9.390853545639698e-06,
|
| 3737 |
+
"loss": 0.7908,
|
| 3738 |
+
"step": 5830
|
| 3739 |
+
},
|
| 3740 |
+
{
|
| 3741 |
+
"epoch": 1.61,
|
| 3742 |
+
"learning_rate": 9.372338455841511e-06,
|
| 3743 |
+
"loss": 0.8059,
|
| 3744 |
+
"step": 5840
|
| 3745 |
+
},
|
| 3746 |
+
{
|
| 3747 |
+
"epoch": 1.61,
|
| 3748 |
+
"learning_rate": 9.353823366043326e-06,
|
| 3749 |
+
"loss": 0.8027,
|
| 3750 |
+
"step": 5850
|
| 3751 |
+
},
|
| 3752 |
+
{
|
| 3753 |
+
"epoch": 1.61,
|
| 3754 |
+
"learning_rate": 9.33530827624514e-06,
|
| 3755 |
+
"loss": 0.7851,
|
| 3756 |
+
"step": 5860
|
| 3757 |
+
},
|
| 3758 |
+
{
|
| 3759 |
+
"epoch": 1.62,
|
| 3760 |
+
"learning_rate": 9.316793186446956e-06,
|
| 3761 |
+
"loss": 0.7932,
|
| 3762 |
+
"step": 5870
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 1.62,
|
| 3766 |
+
"learning_rate": 9.29827809664877e-06,
|
| 3767 |
+
"loss": 0.7911,
|
| 3768 |
+
"step": 5880
|
| 3769 |
+
},
|
| 3770 |
+
{
|
| 3771 |
+
"epoch": 1.62,
|
| 3772 |
+
"learning_rate": 9.279763006850583e-06,
|
| 3773 |
+
"loss": 0.7917,
|
| 3774 |
+
"step": 5890
|
| 3775 |
+
},
|
| 3776 |
+
{
|
| 3777 |
+
"epoch": 1.62,
|
| 3778 |
+
"learning_rate": 9.261247917052398e-06,
|
| 3779 |
+
"loss": 0.7987,
|
| 3780 |
+
"step": 5900
|
| 3781 |
+
},
|
| 3782 |
+
{
|
| 3783 |
+
"epoch": 1.63,
|
| 3784 |
+
"learning_rate": 9.242732827254213e-06,
|
| 3785 |
+
"loss": 0.796,
|
| 3786 |
+
"step": 5910
|
| 3787 |
+
},
|
| 3788 |
+
{
|
| 3789 |
+
"epoch": 1.63,
|
| 3790 |
+
"learning_rate": 9.224217737456027e-06,
|
| 3791 |
+
"loss": 0.7934,
|
| 3792 |
+
"step": 5920
|
| 3793 |
+
},
|
| 3794 |
+
{
|
| 3795 |
+
"epoch": 1.63,
|
| 3796 |
+
"learning_rate": 9.205702647657842e-06,
|
| 3797 |
+
"loss": 0.7984,
|
| 3798 |
+
"step": 5930
|
| 3799 |
+
},
|
| 3800 |
+
{
|
| 3801 |
+
"epoch": 1.63,
|
| 3802 |
+
"learning_rate": 9.187187557859657e-06,
|
| 3803 |
+
"loss": 0.7887,
|
| 3804 |
+
"step": 5940
|
| 3805 |
+
},
|
| 3806 |
+
{
|
| 3807 |
+
"epoch": 1.64,
|
| 3808 |
+
"learning_rate": 9.168672468061471e-06,
|
| 3809 |
+
"loss": 0.8099,
|
| 3810 |
+
"step": 5950
|
| 3811 |
+
},
|
| 3812 |
+
{
|
| 3813 |
+
"epoch": 1.64,
|
| 3814 |
+
"learning_rate": 9.150157378263286e-06,
|
| 3815 |
+
"loss": 0.7949,
|
| 3816 |
+
"step": 5960
|
| 3817 |
+
},
|
| 3818 |
+
{
|
| 3819 |
+
"epoch": 1.64,
|
| 3820 |
+
"learning_rate": 9.1316422884651e-06,
|
| 3821 |
+
"loss": 0.7951,
|
| 3822 |
+
"step": 5970
|
| 3823 |
+
},
|
| 3824 |
+
{
|
| 3825 |
+
"epoch": 1.65,
|
| 3826 |
+
"learning_rate": 9.113127198666914e-06,
|
| 3827 |
+
"loss": 0.79,
|
| 3828 |
+
"step": 5980
|
| 3829 |
+
},
|
| 3830 |
+
{
|
| 3831 |
+
"epoch": 1.65,
|
| 3832 |
+
"learning_rate": 9.094612108868728e-06,
|
| 3833 |
+
"loss": 0.776,
|
| 3834 |
+
"step": 5990
|
| 3835 |
+
},
|
| 3836 |
+
{
|
| 3837 |
+
"epoch": 1.65,
|
| 3838 |
+
"learning_rate": 9.076097019070543e-06,
|
| 3839 |
+
"loss": 0.7867,
|
| 3840 |
+
"step": 6000
|
| 3841 |
+
},
|
| 3842 |
+
{
|
| 3843 |
+
"epoch": 1.65,
|
| 3844 |
+
"eval_loss": 0.7963413596153259,
|
| 3845 |
+
"eval_runtime": 98.6932,
|
| 3846 |
+
"eval_samples_per_second": 10.132,
|
| 3847 |
+
"eval_steps_per_second": 0.638,
|
| 3848 |
+
"step": 6000
|
| 3849 |
+
},
|
| 3850 |
+
{
|
| 3851 |
+
"epoch": 1.65,
|
| 3852 |
+
"learning_rate": 9.057581929272358e-06,
|
| 3853 |
+
"loss": 0.7967,
|
| 3854 |
+
"step": 6010
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 1.66,
|
| 3858 |
+
"learning_rate": 9.039066839474173e-06,
|
| 3859 |
+
"loss": 0.7858,
|
| 3860 |
+
"step": 6020
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 1.66,
|
| 3864 |
+
"learning_rate": 9.020551749675987e-06,
|
| 3865 |
+
"loss": 0.803,
|
| 3866 |
+
"step": 6030
|
| 3867 |
+
},
|
| 3868 |
+
{
|
| 3869 |
+
"epoch": 1.66,
|
| 3870 |
+
"learning_rate": 9.0020366598778e-06,
|
| 3871 |
+
"loss": 0.8034,
|
| 3872 |
+
"step": 6040
|
| 3873 |
+
},
|
| 3874 |
+
{
|
| 3875 |
+
"epoch": 1.66,
|
| 3876 |
+
"learning_rate": 8.983521570079615e-06,
|
| 3877 |
+
"loss": 0.7844,
|
| 3878 |
+
"step": 6050
|
| 3879 |
+
},
|
| 3880 |
+
{
|
| 3881 |
+
"epoch": 1.67,
|
| 3882 |
+
"learning_rate": 8.96500648028143e-06,
|
| 3883 |
+
"loss": 0.7867,
|
| 3884 |
+
"step": 6060
|
| 3885 |
+
},
|
| 3886 |
+
{
|
| 3887 |
+
"epoch": 1.67,
|
| 3888 |
+
"learning_rate": 8.946491390483244e-06,
|
| 3889 |
+
"loss": 0.8031,
|
| 3890 |
+
"step": 6070
|
| 3891 |
+
},
|
| 3892 |
+
{
|
| 3893 |
+
"epoch": 1.67,
|
| 3894 |
+
"learning_rate": 8.927976300685059e-06,
|
| 3895 |
+
"loss": 0.7979,
|
| 3896 |
+
"step": 6080
|
| 3897 |
+
},
|
| 3898 |
+
{
|
| 3899 |
+
"epoch": 1.68,
|
| 3900 |
+
"learning_rate": 8.909461210886874e-06,
|
| 3901 |
+
"loss": 0.7784,
|
| 3902 |
+
"step": 6090
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 1.68,
|
| 3906 |
+
"learning_rate": 8.890946121088688e-06,
|
| 3907 |
+
"loss": 0.7909,
|
| 3908 |
+
"step": 6100
|
| 3909 |
+
},
|
| 3910 |
+
{
|
| 3911 |
+
"epoch": 1.68,
|
| 3912 |
+
"learning_rate": 8.872431031290503e-06,
|
| 3913 |
+
"loss": 0.8128,
|
| 3914 |
+
"step": 6110
|
| 3915 |
+
},
|
| 3916 |
+
{
|
| 3917 |
+
"epoch": 1.68,
|
| 3918 |
+
"learning_rate": 8.853915941492318e-06,
|
| 3919 |
+
"loss": 0.7901,
|
| 3920 |
+
"step": 6120
|
| 3921 |
+
},
|
| 3922 |
+
{
|
| 3923 |
+
"epoch": 1.69,
|
| 3924 |
+
"learning_rate": 8.83540085169413e-06,
|
| 3925 |
+
"loss": 0.808,
|
| 3926 |
+
"step": 6130
|
| 3927 |
+
},
|
| 3928 |
+
{
|
| 3929 |
+
"epoch": 1.69,
|
| 3930 |
+
"learning_rate": 8.816885761895945e-06,
|
| 3931 |
+
"loss": 0.7872,
|
| 3932 |
+
"step": 6140
|
| 3933 |
+
},
|
| 3934 |
+
{
|
| 3935 |
+
"epoch": 1.69,
|
| 3936 |
+
"learning_rate": 8.79837067209776e-06,
|
| 3937 |
+
"loss": 0.7927,
|
| 3938 |
+
"step": 6150
|
| 3939 |
+
},
|
| 3940 |
+
{
|
| 3941 |
+
"epoch": 1.69,
|
| 3942 |
+
"learning_rate": 8.779855582299575e-06,
|
| 3943 |
+
"loss": 0.7958,
|
| 3944 |
+
"step": 6160
|
| 3945 |
+
},
|
| 3946 |
+
{
|
| 3947 |
+
"epoch": 1.7,
|
| 3948 |
+
"learning_rate": 8.76134049250139e-06,
|
| 3949 |
+
"loss": 0.7989,
|
| 3950 |
+
"step": 6170
|
| 3951 |
+
},
|
| 3952 |
+
{
|
| 3953 |
+
"epoch": 1.7,
|
| 3954 |
+
"learning_rate": 8.742825402703203e-06,
|
| 3955 |
+
"loss": 0.8054,
|
| 3956 |
+
"step": 6180
|
| 3957 |
+
},
|
| 3958 |
+
{
|
| 3959 |
+
"epoch": 1.7,
|
| 3960 |
+
"learning_rate": 8.724310312905019e-06,
|
| 3961 |
+
"loss": 0.7951,
|
| 3962 |
+
"step": 6190
|
| 3963 |
+
},
|
| 3964 |
+
{
|
| 3965 |
+
"epoch": 1.71,
|
| 3966 |
+
"learning_rate": 8.705795223106834e-06,
|
| 3967 |
+
"loss": 0.7843,
|
| 3968 |
+
"step": 6200
|
| 3969 |
+
},
|
| 3970 |
+
{
|
| 3971 |
+
"epoch": 1.71,
|
| 3972 |
+
"eval_loss": 0.7958487272262573,
|
| 3973 |
+
"eval_runtime": 98.7031,
|
| 3974 |
+
"eval_samples_per_second": 10.131,
|
| 3975 |
+
"eval_steps_per_second": 0.638,
|
| 3976 |
+
"step": 6200
|
| 3977 |
+
},
|
| 3978 |
+
{
|
| 3979 |
+
"epoch": 1.71,
|
| 3980 |
+
"learning_rate": 8.687280133308648e-06,
|
| 3981 |
+
"loss": 0.7896,
|
| 3982 |
+
"step": 6210
|
| 3983 |
+
},
|
| 3984 |
+
{
|
| 3985 |
+
"epoch": 1.71,
|
| 3986 |
+
"learning_rate": 8.668765043510461e-06,
|
| 3987 |
+
"loss": 0.7814,
|
| 3988 |
+
"step": 6220
|
| 3989 |
+
},
|
| 3990 |
+
{
|
| 3991 |
+
"epoch": 1.71,
|
| 3992 |
+
"learning_rate": 8.650249953712276e-06,
|
| 3993 |
+
"loss": 0.8004,
|
| 3994 |
+
"step": 6230
|
| 3995 |
+
},
|
| 3996 |
+
{
|
| 3997 |
+
"epoch": 1.72,
|
| 3998 |
+
"learning_rate": 8.63173486391409e-06,
|
| 3999 |
+
"loss": 0.7786,
|
| 4000 |
+
"step": 6240
|
| 4001 |
+
},
|
| 4002 |
+
{
|
| 4003 |
+
"epoch": 1.72,
|
| 4004 |
+
"learning_rate": 8.613219774115905e-06,
|
| 4005 |
+
"loss": 0.7991,
|
| 4006 |
+
"step": 6250
|
| 4007 |
+
},
|
| 4008 |
+
{
|
| 4009 |
+
"epoch": 1.72,
|
| 4010 |
+
"learning_rate": 8.59470468431772e-06,
|
| 4011 |
+
"loss": 0.7918,
|
| 4012 |
+
"step": 6260
|
| 4013 |
+
},
|
| 4014 |
+
{
|
| 4015 |
+
"epoch": 1.73,
|
| 4016 |
+
"learning_rate": 8.576189594519535e-06,
|
| 4017 |
+
"loss": 0.7778,
|
| 4018 |
+
"step": 6270
|
| 4019 |
+
},
|
| 4020 |
+
{
|
| 4021 |
+
"epoch": 1.73,
|
| 4022 |
+
"learning_rate": 8.557674504721348e-06,
|
| 4023 |
+
"loss": 0.7909,
|
| 4024 |
+
"step": 6280
|
| 4025 |
+
},
|
| 4026 |
+
{
|
| 4027 |
+
"epoch": 1.73,
|
| 4028 |
+
"learning_rate": 8.539159414923162e-06,
|
| 4029 |
+
"loss": 0.792,
|
| 4030 |
+
"step": 6290
|
| 4031 |
+
},
|
| 4032 |
+
{
|
| 4033 |
+
"epoch": 1.73,
|
| 4034 |
+
"learning_rate": 8.520644325124977e-06,
|
| 4035 |
+
"loss": 0.8067,
|
| 4036 |
+
"step": 6300
|
| 4037 |
+
},
|
| 4038 |
+
{
|
| 4039 |
+
"epoch": 1.74,
|
| 4040 |
+
"learning_rate": 8.502129235326792e-06,
|
| 4041 |
+
"loss": 0.7965,
|
| 4042 |
+
"step": 6310
|
| 4043 |
+
},
|
| 4044 |
+
{
|
| 4045 |
+
"epoch": 1.74,
|
| 4046 |
+
"learning_rate": 8.483614145528607e-06,
|
| 4047 |
+
"loss": 0.7958,
|
| 4048 |
+
"step": 6320
|
| 4049 |
+
},
|
| 4050 |
+
{
|
| 4051 |
+
"epoch": 1.74,
|
| 4052 |
+
"learning_rate": 8.465099055730421e-06,
|
| 4053 |
+
"loss": 0.777,
|
| 4054 |
+
"step": 6330
|
| 4055 |
+
},
|
| 4056 |
+
{
|
| 4057 |
+
"epoch": 1.74,
|
| 4058 |
+
"learning_rate": 8.446583965932236e-06,
|
| 4059 |
+
"loss": 0.7934,
|
| 4060 |
+
"step": 6340
|
| 4061 |
+
},
|
| 4062 |
+
{
|
| 4063 |
+
"epoch": 1.75,
|
| 4064 |
+
"learning_rate": 8.42806887613405e-06,
|
| 4065 |
+
"loss": 0.7899,
|
| 4066 |
+
"step": 6350
|
| 4067 |
+
},
|
| 4068 |
+
{
|
| 4069 |
+
"epoch": 1.75,
|
| 4070 |
+
"learning_rate": 8.409553786335865e-06,
|
| 4071 |
+
"loss": 0.7933,
|
| 4072 |
+
"step": 6360
|
| 4073 |
+
},
|
| 4074 |
+
{
|
| 4075 |
+
"epoch": 1.75,
|
| 4076 |
+
"learning_rate": 8.391038696537678e-06,
|
| 4077 |
+
"loss": 0.7898,
|
| 4078 |
+
"step": 6370
|
| 4079 |
+
},
|
| 4080 |
+
{
|
| 4081 |
+
"epoch": 1.76,
|
| 4082 |
+
"learning_rate": 8.372523606739493e-06,
|
| 4083 |
+
"loss": 0.7749,
|
| 4084 |
+
"step": 6380
|
| 4085 |
+
},
|
| 4086 |
+
{
|
| 4087 |
+
"epoch": 1.76,
|
| 4088 |
+
"learning_rate": 8.354008516941308e-06,
|
| 4089 |
+
"loss": 0.7951,
|
| 4090 |
+
"step": 6390
|
| 4091 |
+
},
|
| 4092 |
+
{
|
| 4093 |
+
"epoch": 1.76,
|
| 4094 |
+
"learning_rate": 8.335493427143122e-06,
|
| 4095 |
+
"loss": 0.79,
|
| 4096 |
+
"step": 6400
|
| 4097 |
+
},
|
| 4098 |
+
{
|
| 4099 |
+
"epoch": 1.76,
|
| 4100 |
+
"eval_loss": 0.7957026362419128,
|
| 4101 |
+
"eval_runtime": 98.7285,
|
| 4102 |
+
"eval_samples_per_second": 10.129,
|
| 4103 |
+
"eval_steps_per_second": 0.638,
|
| 4104 |
+
"step": 6400
|
| 4105 |
+
},
|
| 4106 |
+
{
|
| 4107 |
+
"epoch": 1.76,
|
| 4108 |
+
"learning_rate": 8.316978337344937e-06,
|
| 4109 |
+
"loss": 0.8007,
|
| 4110 |
+
"step": 6410
|
| 4111 |
+
},
|
| 4112 |
+
{
|
| 4113 |
+
"epoch": 1.77,
|
| 4114 |
+
"learning_rate": 8.29846324754675e-06,
|
| 4115 |
+
"loss": 0.791,
|
| 4116 |
+
"step": 6420
|
| 4117 |
+
},
|
| 4118 |
+
{
|
| 4119 |
+
"epoch": 1.77,
|
| 4120 |
+
"learning_rate": 8.279948157748566e-06,
|
| 4121 |
+
"loss": 0.8045,
|
| 4122 |
+
"step": 6430
|
| 4123 |
+
},
|
| 4124 |
+
{
|
| 4125 |
+
"epoch": 1.77,
|
| 4126 |
+
"learning_rate": 8.261433067950381e-06,
|
| 4127 |
+
"loss": 0.7969,
|
| 4128 |
+
"step": 6440
|
| 4129 |
+
},
|
| 4130 |
+
{
|
| 4131 |
+
"epoch": 1.77,
|
| 4132 |
+
"learning_rate": 8.242917978152196e-06,
|
| 4133 |
+
"loss": 0.7826,
|
| 4134 |
+
"step": 6450
|
| 4135 |
+
},
|
| 4136 |
+
{
|
| 4137 |
+
"epoch": 1.78,
|
| 4138 |
+
"learning_rate": 8.224402888354009e-06,
|
| 4139 |
+
"loss": 0.7792,
|
| 4140 |
+
"step": 6460
|
| 4141 |
+
},
|
| 4142 |
+
{
|
| 4143 |
+
"epoch": 1.78,
|
| 4144 |
+
"learning_rate": 8.205887798555824e-06,
|
| 4145 |
+
"loss": 0.7953,
|
| 4146 |
+
"step": 6470
|
| 4147 |
+
},
|
| 4148 |
+
{
|
| 4149 |
+
"epoch": 1.78,
|
| 4150 |
+
"learning_rate": 8.187372708757638e-06,
|
| 4151 |
+
"loss": 0.7822,
|
| 4152 |
+
"step": 6480
|
| 4153 |
+
},
|
| 4154 |
+
{
|
| 4155 |
+
"epoch": 1.79,
|
| 4156 |
+
"learning_rate": 8.168857618959453e-06,
|
| 4157 |
+
"loss": 0.79,
|
| 4158 |
+
"step": 6490
|
| 4159 |
+
},
|
| 4160 |
+
{
|
| 4161 |
+
"epoch": 1.79,
|
| 4162 |
+
"learning_rate": 8.150342529161268e-06,
|
| 4163 |
+
"loss": 0.7902,
|
| 4164 |
+
"step": 6500
|
| 4165 |
+
},
|
| 4166 |
+
{
|
| 4167 |
+
"epoch": 1.79,
|
| 4168 |
+
"learning_rate": 8.13182743936308e-06,
|
| 4169 |
+
"loss": 0.7979,
|
| 4170 |
+
"step": 6510
|
| 4171 |
+
},
|
| 4172 |
+
{
|
| 4173 |
+
"epoch": 1.79,
|
| 4174 |
+
"learning_rate": 8.113312349564895e-06,
|
| 4175 |
+
"loss": 0.8036,
|
| 4176 |
+
"step": 6520
|
| 4177 |
+
},
|
| 4178 |
+
{
|
| 4179 |
+
"epoch": 1.8,
|
| 4180 |
+
"learning_rate": 8.09479725976671e-06,
|
| 4181 |
+
"loss": 0.7917,
|
| 4182 |
+
"step": 6530
|
| 4183 |
+
},
|
| 4184 |
+
{
|
| 4185 |
+
"epoch": 1.8,
|
| 4186 |
+
"learning_rate": 8.076282169968525e-06,
|
| 4187 |
+
"loss": 0.7659,
|
| 4188 |
+
"step": 6540
|
| 4189 |
+
},
|
| 4190 |
+
{
|
| 4191 |
+
"epoch": 1.8,
|
| 4192 |
+
"learning_rate": 8.05776708017034e-06,
|
| 4193 |
+
"loss": 0.8012,
|
| 4194 |
+
"step": 6550
|
| 4195 |
+
},
|
| 4196 |
+
{
|
| 4197 |
+
"epoch": 1.8,
|
| 4198 |
+
"learning_rate": 8.039251990372154e-06,
|
| 4199 |
+
"loss": 0.7836,
|
| 4200 |
+
"step": 6560
|
| 4201 |
+
},
|
| 4202 |
+
{
|
| 4203 |
+
"epoch": 1.81,
|
| 4204 |
+
"learning_rate": 8.020736900573969e-06,
|
| 4205 |
+
"loss": 0.7888,
|
| 4206 |
+
"step": 6570
|
| 4207 |
+
},
|
| 4208 |
+
{
|
| 4209 |
+
"epoch": 1.81,
|
| 4210 |
+
"learning_rate": 8.002221810775783e-06,
|
| 4211 |
+
"loss": 0.7869,
|
| 4212 |
+
"step": 6580
|
| 4213 |
+
},
|
| 4214 |
+
{
|
| 4215 |
+
"epoch": 1.81,
|
| 4216 |
+
"learning_rate": 7.983706720977598e-06,
|
| 4217 |
+
"loss": 0.7913,
|
| 4218 |
+
"step": 6590
|
| 4219 |
+
},
|
| 4220 |
+
{
|
| 4221 |
+
"epoch": 1.82,
|
| 4222 |
+
"learning_rate": 7.965191631179411e-06,
|
| 4223 |
+
"loss": 0.8022,
|
| 4224 |
+
"step": 6600
|
| 4225 |
+
},
|
| 4226 |
+
{
|
| 4227 |
+
"epoch": 1.82,
|
| 4228 |
+
"eval_loss": 0.7951002717018127,
|
| 4229 |
+
"eval_runtime": 98.6725,
|
| 4230 |
+
"eval_samples_per_second": 10.135,
|
| 4231 |
+
"eval_steps_per_second": 0.638,
|
| 4232 |
+
"step": 6600
|
| 4233 |
+
},
|
| 4234 |
+
{
|
| 4235 |
+
"epoch": 1.82,
|
| 4236 |
+
"learning_rate": 7.948528050361045e-06,
|
| 4237 |
+
"loss": 0.7867,
|
| 4238 |
+
"step": 6610
|
| 4239 |
+
},
|
| 4240 |
+
{
|
| 4241 |
+
"epoch": 1.82,
|
| 4242 |
+
"learning_rate": 7.93001296056286e-06,
|
| 4243 |
+
"loss": 0.7974,
|
| 4244 |
+
"step": 6620
|
| 4245 |
+
},
|
| 4246 |
+
{
|
| 4247 |
+
"epoch": 1.82,
|
| 4248 |
+
"learning_rate": 7.911497870764675e-06,
|
| 4249 |
+
"loss": 0.7959,
|
| 4250 |
+
"step": 6630
|
| 4251 |
+
},
|
| 4252 |
+
{
|
| 4253 |
+
"epoch": 1.83,
|
| 4254 |
+
"learning_rate": 7.892982780966488e-06,
|
| 4255 |
+
"loss": 0.79,
|
| 4256 |
+
"step": 6640
|
| 4257 |
+
},
|
| 4258 |
+
{
|
| 4259 |
+
"epoch": 1.83,
|
| 4260 |
+
"learning_rate": 7.874467691168302e-06,
|
| 4261 |
+
"loss": 0.7942,
|
| 4262 |
+
"step": 6650
|
| 4263 |
+
},
|
| 4264 |
+
{
|
| 4265 |
+
"epoch": 1.83,
|
| 4266 |
+
"learning_rate": 7.855952601370117e-06,
|
| 4267 |
+
"loss": 0.7984,
|
| 4268 |
+
"step": 6660
|
| 4269 |
+
},
|
| 4270 |
+
{
|
| 4271 |
+
"epoch": 1.84,
|
| 4272 |
+
"learning_rate": 7.837437511571932e-06,
|
| 4273 |
+
"loss": 0.785,
|
| 4274 |
+
"step": 6670
|
| 4275 |
+
},
|
| 4276 |
+
{
|
| 4277 |
+
"epoch": 1.84,
|
| 4278 |
+
"learning_rate": 7.818922421773747e-06,
|
| 4279 |
+
"loss": 0.7959,
|
| 4280 |
+
"step": 6680
|
| 4281 |
+
},
|
| 4282 |
+
{
|
| 4283 |
+
"epoch": 1.84,
|
| 4284 |
+
"learning_rate": 7.80040733197556e-06,
|
| 4285 |
+
"loss": 0.7774,
|
| 4286 |
+
"step": 6690
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 1.84,
|
| 4290 |
+
"learning_rate": 7.781892242177376e-06,
|
| 4291 |
+
"loss": 0.7914,
|
| 4292 |
+
"step": 6700
|
| 4293 |
+
},
|
| 4294 |
+
{
|
| 4295 |
+
"epoch": 1.85,
|
| 4296 |
+
"learning_rate": 7.76337715237919e-06,
|
| 4297 |
+
"loss": 0.7791,
|
| 4298 |
+
"step": 6710
|
| 4299 |
+
},
|
| 4300 |
+
{
|
| 4301 |
+
"epoch": 1.85,
|
| 4302 |
+
"learning_rate": 7.744862062581005e-06,
|
| 4303 |
+
"loss": 0.7986,
|
| 4304 |
+
"step": 6720
|
| 4305 |
+
},
|
| 4306 |
+
{
|
| 4307 |
+
"epoch": 1.85,
|
| 4308 |
+
"learning_rate": 7.726346972782818e-06,
|
| 4309 |
+
"loss": 0.7937,
|
| 4310 |
+
"step": 6730
|
| 4311 |
+
},
|
| 4312 |
+
{
|
| 4313 |
+
"epoch": 1.85,
|
| 4314 |
+
"learning_rate": 7.707831882984633e-06,
|
| 4315 |
+
"loss": 0.7949,
|
| 4316 |
+
"step": 6740
|
| 4317 |
+
},
|
| 4318 |
+
{
|
| 4319 |
+
"epoch": 1.86,
|
| 4320 |
+
"learning_rate": 7.689316793186448e-06,
|
| 4321 |
+
"loss": 0.7932,
|
| 4322 |
+
"step": 6750
|
| 4323 |
+
},
|
| 4324 |
+
{
|
| 4325 |
+
"epoch": 1.86,
|
| 4326 |
+
"learning_rate": 7.670801703388262e-06,
|
| 4327 |
+
"loss": 0.7988,
|
| 4328 |
+
"step": 6760
|
| 4329 |
+
},
|
| 4330 |
+
{
|
| 4331 |
+
"epoch": 1.86,
|
| 4332 |
+
"learning_rate": 7.652286613590077e-06,
|
| 4333 |
+
"loss": 0.7852,
|
| 4334 |
+
"step": 6770
|
| 4335 |
+
},
|
| 4336 |
+
{
|
| 4337 |
+
"epoch": 1.87,
|
| 4338 |
+
"learning_rate": 7.63377152379189e-06,
|
| 4339 |
+
"loss": 0.7922,
|
| 4340 |
+
"step": 6780
|
| 4341 |
+
},
|
| 4342 |
+
{
|
| 4343 |
+
"epoch": 1.87,
|
| 4344 |
+
"learning_rate": 7.615256433993706e-06,
|
| 4345 |
+
"loss": 0.7936,
|
| 4346 |
+
"step": 6790
|
| 4347 |
+
},
|
| 4348 |
+
{
|
| 4349 |
+
"epoch": 1.87,
|
| 4350 |
+
"learning_rate": 7.5967413441955194e-06,
|
| 4351 |
+
"loss": 0.7794,
|
| 4352 |
+
"step": 6800
|
| 4353 |
+
},
|
| 4354 |
+
{
|
| 4355 |
+
"epoch": 1.87,
|
| 4356 |
+
"eval_loss": 0.7945918440818787,
|
| 4357 |
+
"eval_runtime": 98.7262,
|
| 4358 |
+
"eval_samples_per_second": 10.129,
|
| 4359 |
+
"eval_steps_per_second": 0.638,
|
| 4360 |
+
"step": 6800
|
| 4361 |
+
}
|
| 4362 |
+
],
|
| 4363 |
+
"max_steps": 10902,
|
| 4364 |
+
"num_train_epochs": 3,
|
| 4365 |
+
"total_flos": 2.0184019484077457e+19,
|
| 4366 |
+
"trial_name": null,
|
| 4367 |
+
"trial_params": null
|
| 4368 |
+
}
|
checkpoint-6800/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:061bef34d08900ea9fa9968114112b4c9df622a8ab74fd0a9a3c90bc3e259f6c
|
| 3 |
+
size 3579
|