Hugging Face's logo

ondevicellm
/
tinyllama_mole_sft_router05_lr1e-4_ep3

Text Generation
Transformers
TensorBoard
Safetensors
mixtralmole
alignment-handbook
trl
sft
Generated from Trainer
conversational
custom_code
Model card Files Metrics Community
hushell commited on
Commit
b2f9a13
•
1 Parent(s): d300b11

Model save

Browse files
Files changed (9) hide show
  1. README.md +99 -0
  2. all_results.json +13 -0
  3. eval_results.json +8 -0
  4. generation_config.json +7 -0
  5. model.safetensors +1 -1
  6. runs/Feb01_15-59-46_main1/events.out.tfevents.1706803248.main1.95696.0 +2 -2
  7. runs/Feb01_15-59-46_main1/events.out.tfevents.1706852763.main1.95696.1 +3 -0
  8. train_results.json +8 -0
  9. trainer_state.json +4412 -0
README.md ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: ondevicellm/tinyllama_mole_v1
3
+ tags:
4
+ - trl
5
+ - sft
6
+ - generated_from_trainer
7
+ datasets:
8
+ - generator
9
+ model-index:
10
+ - name: tinyllama_mole_sft_router05_lr1e-4_ep3
11
+ results: []
12
+ ---
13
+
14
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
15
+ should probably proofread and complete it, then remove this comment. -->
16
+
17
+ # tinyllama_mole_sft_router05_lr1e-4_ep3
18
+
19
+ This model is a fine-tuned version of [ondevicellm/tinyllama_mole_v1](https://huggingface.co/ondevicellm/tinyllama_mole_v1) on the generator dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: 2.1035
22
+
23
+ ## Model description
24
+
25
+ More information needed
26
+
27
+ ## Intended uses & limitations
28
+
29
+ More information needed
30
+
31
+ ## Training and evaluation data
32
+
33
+ More information needed
34
+
35
+ ## Training procedure
36
+
37
+ ### Training hyperparameters
38
+
39
+ The following hyperparameters were used during training:
40
+ - learning_rate: 0.0001
41
+ - train_batch_size: 16
42
+ - eval_batch_size: 8
43
+ - seed: 42
44
+ - distributed_type: multi-GPU
45
+ - num_devices: 4
46
+ - gradient_accumulation_steps: 2
47
+ - total_train_batch_size: 128
48
+ - total_eval_batch_size: 32
49
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
50
+ - lr_scheduler_type: cosine
51
+ - lr_scheduler_warmup_steps: 120
52
+ - num_epochs: 3
53
+
54
+ ### Training results
55
+
56
+ | Training Loss | Epoch | Step | Validation Loss |
57
+ |:-------------:|:-----:|:----:|:---------------:|
58
+ | 2.2617 | 0.09 | 100 | 2.2410 |
59
+ | 2.2246 | 0.18 | 200 | 2.2165 |
60
+ | 2.1994 | 0.26 | 300 | 2.1994 |
61
+ | 2.1767 | 0.35 | 400 | 2.1869 |
62
+ | 2.1532 | 0.44 | 500 | 2.1792 |
63
+ | 2.171 | 0.53 | 600 | 2.1717 |
64
+ | 2.1588 | 0.61 | 700 | 2.1645 |
65
+ | 2.145 | 0.7 | 800 | 2.1567 |
66
+ | 2.1366 | 0.79 | 900 | 2.1507 |
67
+ | 2.1219 | 0.88 | 1000 | 2.1450 |
68
+ | 2.1415 | 0.96 | 1100 | 2.1387 |
69
+ | 1.9765 | 1.05 | 1200 | 2.1446 |
70
+ | 1.9837 | 1.14 | 1300 | 2.1430 |
71
+ | 1.9952 | 1.23 | 1400 | 2.1388 |
72
+ | 1.9868 | 1.31 | 1500 | 2.1351 |
73
+ | 1.9864 | 1.4 | 1600 | 2.1316 |
74
+ | 1.987 | 1.49 | 1700 | 2.1263 |
75
+ | 1.9678 | 1.58 | 1800 | 2.1230 |
76
+ | 1.9827 | 1.66 | 1900 | 2.1164 |
77
+ | 1.9846 | 1.75 | 2000 | 2.1134 |
78
+ | 1.9694 | 1.84 | 2100 | 2.1068 |
79
+ | 1.9429 | 1.93 | 2200 | 2.1035 |
80
+ | 1.8079 | 2.01 | 2300 | 2.1369 |
81
+ | 1.8132 | 2.1 | 2400 | 2.1375 |
82
+ | 1.8043 | 2.19 | 2500 | 2.1360 |
83
+ | 1.7927 | 2.28 | 2600 | 2.1334 |
84
+ | 1.7935 | 2.37 | 2700 | 2.1335 |
85
+ | 1.7982 | 2.45 | 2800 | 2.1321 |
86
+ | 1.8029 | 2.54 | 2900 | 2.1311 |
87
+ | 1.7919 | 2.63 | 3000 | 2.1298 |
88
+ | 1.7953 | 2.72 | 3100 | 2.1287 |
89
+ | 1.798 | 2.8 | 3200 | 2.1280 |
90
+ | 1.7947 | 2.89 | 3300 | 2.1282 |
91
+ | 1.8015 | 2.98 | 3400 | 2.1283 |
92
+
93
+
94
+ ### Framework versions
95
+
96
+ - Transformers 4.37.0
97
+ - Pytorch 2.1.2+cu118
98
+ - Datasets 2.16.1
99
+ - Tokenizers 0.15.0
all_results.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "eval_loss": 2.10353946685791,
4
+ "eval_runtime": 435.0953,
5
+ "eval_samples": 23110,
6
+ "eval_samples_per_second": 37.155,
7
+ "eval_steps_per_second": 1.163,
8
+ "train_loss": 1.9906484538130407,
9
+ "train_runtime": 48039.6785,
10
+ "train_samples": 207865,
11
+ "train_samples_per_second": 9.121,
12
+ "train_steps_per_second": 0.071
13
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "eval_loss": 2.10353946685791,
4
+ "eval_runtime": 435.0953,
5
+ "eval_samples": 23110,
6
+ "eval_samples_per_second": 37.155,
7
+ "eval_steps_per_second": 1.163
8
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.37.0",
6
+ "use_cache": false
7
+ }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ce8e888a66e3fb5b5367b2cba3ae9d2df51bec8a78e6a1491fb4ead969d24bc4
3
  size 2223960880
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:769464401a1850ba38f4cc894660f8bc506d5e76713e49b17ffb0fbeaab97ac9
3
  size 2223960880
runs/Feb01_15-59-46_main1/events.out.tfevents.1706803248.main1.95696.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:028b25eed8fcf6a8c8225076fe2c324f5dca4806d64301e15fee8e8b1b8f33bc
3
- size 121168
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c70a7216d727c72baeba60d46edf2426b1c7dfdb3866160a1fc954dcfd4a565
3
+ size 122150
runs/Feb01_15-59-46_main1/events.out.tfevents.1706852763.main1.95696.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8ad70d3aab68e64431ff884e9afd776d397957be1588c2ac0e6cb41e596b28d
3
+ size 359
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "train_loss": 1.9906484538130407,
4
+ "train_runtime": 48039.6785,
5
+ "train_samples": 207865,
6
+ "train_samples_per_second": 9.121,
7
+ "train_steps_per_second": 0.071
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,4412 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 2.10353946685791,
3
+ "best_model_checkpoint": "data/tinyllama_mole_sft_router05_lr1e-4_ep3/checkpoint-2200",
4
+ "epoch": 2.9986859395532193,
5
+ "eval_steps": 100,
6
+ "global_step": 3423,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "learning_rate": 8.333333333333333e-07,
14
+ "loss": 3.7925,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.0,
19
+ "learning_rate": 4.166666666666667e-06,
20
+ "loss": 3.7302,
21
+ "step": 5
22
+ },
23
+ {
24
+ "epoch": 0.01,
25
+ "learning_rate": 8.333333333333334e-06,
26
+ "loss": 3.3746,
27
+ "step": 10
28
+ },
29
+ {
30
+ "epoch": 0.01,
31
+ "learning_rate": 1.25e-05,
32
+ "loss": 2.8214,
33
+ "step": 15
34
+ },
35
+ {
36
+ "epoch": 0.02,
37
+ "learning_rate": 1.6666666666666667e-05,
38
+ "loss": 2.6367,
39
+ "step": 20
40
+ },
41
+ {
42
+ "epoch": 0.02,
43
+ "learning_rate": 2.0833333333333336e-05,
44
+ "loss": 2.5137,
45
+ "step": 25
46
+ },
47
+ {
48
+ "epoch": 0.03,
49
+ "learning_rate": 2.5e-05,
50
+ "loss": 2.4115,
51
+ "step": 30
52
+ },
53
+ {
54
+ "epoch": 0.03,
55
+ "learning_rate": 2.916666666666667e-05,
56
+ "loss": 2.3625,
57
+ "step": 35
58
+ },
59
+ {
60
+ "epoch": 0.04,
61
+ "learning_rate": 3.3333333333333335e-05,
62
+ "loss": 2.3396,
63
+ "step": 40
64
+ },
65
+ {
66
+ "epoch": 0.04,
67
+ "learning_rate": 3.7500000000000003e-05,
68
+ "loss": 2.3126,
69
+ "step": 45
70
+ },
71
+ {
72
+ "epoch": 0.04,
73
+ "learning_rate": 4.166666666666667e-05,
74
+ "loss": 2.3046,
75
+ "step": 50
76
+ },
77
+ {
78
+ "epoch": 0.05,
79
+ "learning_rate": 4.5833333333333334e-05,
80
+ "loss": 2.2724,
81
+ "step": 55
82
+ },
83
+ {
84
+ "epoch": 0.05,
85
+ "learning_rate": 5e-05,
86
+ "loss": 2.2881,
87
+ "step": 60
88
+ },
89
+ {
90
+ "epoch": 0.06,
91
+ "learning_rate": 5.4166666666666664e-05,
92
+ "loss": 2.2671,
93
+ "step": 65
94
+ },
95
+ {
96
+ "epoch": 0.06,
97
+ "learning_rate": 5.833333333333334e-05,
98
+ "loss": 2.2738,
99
+ "step": 70
100
+ },
101
+ {
102
+ "epoch": 0.07,
103
+ "learning_rate": 6.25e-05,
104
+ "loss": 2.2463,
105
+ "step": 75
106
+ },
107
+ {
108
+ "epoch": 0.07,
109
+ "learning_rate": 6.666666666666667e-05,
110
+ "loss": 2.2509,
111
+ "step": 80
112
+ },
113
+ {
114
+ "epoch": 0.07,
115
+ "learning_rate": 7.083333333333334e-05,
116
+ "loss": 2.2554,
117
+ "step": 85
118
+ },
119
+ {
120
+ "epoch": 0.08,
121
+ "learning_rate": 7.500000000000001e-05,
122
+ "loss": 2.2522,
123
+ "step": 90
124
+ },
125
+ {
126
+ "epoch": 0.08,
127
+ "learning_rate": 7.916666666666666e-05,
128
+ "loss": 2.2278,
129
+ "step": 95
130
+ },
131
+ {
132
+ "epoch": 0.09,
133
+ "learning_rate": 8.333333333333334e-05,
134
+ "loss": 2.2617,
135
+ "step": 100
136
+ },
137
+ {
138
+ "epoch": 0.09,
139
+ "eval_loss": 2.2410387992858887,
140
+ "eval_runtime": 439.5124,
141
+ "eval_samples_per_second": 36.782,
142
+ "eval_steps_per_second": 1.151,
143
+ "step": 100
144
+ },
145
+ {
146
+ "epoch": 0.09,
147
+ "learning_rate": 8.75e-05,
148
+ "loss": 2.2391,
149
+ "step": 105
150
+ },
151
+ {
152
+ "epoch": 0.1,
153
+ "learning_rate": 9.166666666666667e-05,
154
+ "loss": 2.2334,
155
+ "step": 110
156
+ },
157
+ {
158
+ "epoch": 0.1,
159
+ "learning_rate": 9.583333333333334e-05,
160
+ "loss": 2.2407,
161
+ "step": 115
162
+ },
163
+ {
164
+ "epoch": 0.11,
165
+ "learning_rate": 0.0001,
166
+ "loss": 2.2282,
167
+ "step": 120
168
+ },
169
+ {
170
+ "epoch": 0.11,
171
+ "learning_rate": 9.999943459219818e-05,
172
+ "loss": 2.232,
173
+ "step": 125
174
+ },
175
+ {
176
+ "epoch": 0.11,
177
+ "learning_rate": 9.999773838158017e-05,
178
+ "loss": 2.2113,
179
+ "step": 130
180
+ },
181
+ {
182
+ "epoch": 0.12,
183
+ "learning_rate": 9.9994911406508e-05,
184
+ "loss": 2.2273,
185
+ "step": 135
186
+ },
187
+ {
188
+ "epoch": 0.12,
189
+ "learning_rate": 9.99909537309174e-05,
190
+ "loss": 2.2408,
191
+ "step": 140
192
+ },
193
+ {
194
+ "epoch": 0.13,
195
+ "learning_rate": 9.998586544431642e-05,
196
+ "loss": 2.2243,
197
+ "step": 145
198
+ },
199
+ {
200
+ "epoch": 0.13,
201
+ "learning_rate": 9.997964666178332e-05,
202
+ "loss": 2.2275,
203
+ "step": 150
204
+ },
205
+ {
206
+ "epoch": 0.14,
207
+ "learning_rate": 9.997229752396404e-05,
208
+ "loss": 2.2214,
209
+ "step": 155
210
+ },
211
+ {
212
+ "epoch": 0.14,
213
+ "learning_rate": 9.996381819706897e-05,
214
+ "loss": 2.2175,
215
+ "step": 160
216
+ },
217
+ {
218
+ "epoch": 0.14,
219
+ "learning_rate": 9.995420887286921e-05,
220
+ "loss": 2.2134,
221
+ "step": 165
222
+ },
223
+ {
224
+ "epoch": 0.15,
225
+ "learning_rate": 9.994346976869222e-05,
226
+ "loss": 2.219,
227
+ "step": 170
228
+ },
229
+ {
230
+ "epoch": 0.15,
231
+ "learning_rate": 9.993160112741697e-05,
232
+ "loss": 2.2295,
233
+ "step": 175
234
+ },
235
+ {
236
+ "epoch": 0.16,
237
+ "learning_rate": 9.991860321746833e-05,
238
+ "loss": 2.2077,
239
+ "step": 180
240
+ },
241
+ {
242
+ "epoch": 0.16,
243
+ "learning_rate": 9.990447633281108e-05,
244
+ "loss": 2.2074,
245
+ "step": 185
246
+ },
247
+ {
248
+ "epoch": 0.17,
249
+ "learning_rate": 9.988922079294327e-05,
250
+ "loss": 2.2145,
251
+ "step": 190
252
+ },
253
+ {
254
+ "epoch": 0.17,
255
+ "learning_rate": 9.987283694288896e-05,
256
+ "loss": 2.2054,
257
+ "step": 195
258
+ },
259
+ {
260
+ "epoch": 0.18,
261
+ "learning_rate": 9.985532515319038e-05,
262
+ "loss": 2.2246,
263
+ "step": 200
264
+ },
265
+ {
266
+ "epoch": 0.18,
267
+ "eval_loss": 2.21648907661438,
268
+ "eval_runtime": 475.9972,
269
+ "eval_samples_per_second": 33.962,
270
+ "eval_steps_per_second": 1.063,
271
+ "step": 200
272
+ },
273
+ {
274
+ "epoch": 0.18,
275
+ "learning_rate": 9.983668581989965e-05,
276
+ "loss": 2.2172,
277
+ "step": 205
278
+ },
279
+ {
280
+ "epoch": 0.18,
281
+ "learning_rate": 9.981691936456975e-05,
282
+ "loss": 2.1936,
283
+ "step": 210
284
+ },
285
+ {
286
+ "epoch": 0.19,
287
+ "learning_rate": 9.9796026234245e-05,
288
+ "loss": 2.2234,
289
+ "step": 215
290
+ },
291
+ {
292
+ "epoch": 0.19,
293
+ "learning_rate": 9.977400690145097e-05,
294
+ "loss": 2.2015,
295
+ "step": 220
296
+ },
297
+ {
298
+ "epoch": 0.2,
299
+ "learning_rate": 9.975086186418373e-05,
300
+ "loss": 2.1856,
301
+ "step": 225
302
+ },
303
+ {
304
+ "epoch": 0.2,
305
+ "learning_rate": 9.972659164589871e-05,
306
+ "loss": 2.2081,
307
+ "step": 230
308
+ },
309
+ {
310
+ "epoch": 0.21,
311
+ "learning_rate": 9.970119679549868e-05,
312
+ "loss": 2.2161,
313
+ "step": 235
314
+ },
315
+ {
316
+ "epoch": 0.21,
317
+ "learning_rate": 9.967467788732156e-05,
318
+ "loss": 2.1883,
319
+ "step": 240
320
+ },
321
+ {
322
+ "epoch": 0.21,
323
+ "learning_rate": 9.964703552112722e-05,
324
+ "loss": 2.2166,
325
+ "step": 245
326
+ },
327
+ {
328
+ "epoch": 0.22,
329
+ "learning_rate": 9.961827032208406e-05,
330
+ "loss": 2.1922,
331
+ "step": 250
332
+ },
333
+ {
334
+ "epoch": 0.22,
335
+ "learning_rate": 9.958838294075479e-05,
336
+ "loss": 2.2028,
337
+ "step": 255
338
+ },
339
+ {
340
+ "epoch": 0.23,
341
+ "learning_rate": 9.955737405308174e-05,
342
+ "loss": 2.2048,
343
+ "step": 260
344
+ },
345
+ {
346
+ "epoch": 0.23,
347
+ "learning_rate": 9.952524436037161e-05,
348
+ "loss": 2.1892,
349
+ "step": 265
350
+ },
351
+ {
352
+ "epoch": 0.24,
353
+ "learning_rate": 9.949199458927955e-05,
354
+ "loss": 2.1891,
355
+ "step": 270
356
+ },
357
+ {
358
+ "epoch": 0.24,
359
+ "learning_rate": 9.945762549179277e-05,
360
+ "loss": 2.2036,
361
+ "step": 275
362
+ },
363
+ {
364
+ "epoch": 0.25,
365
+ "learning_rate": 9.942213784521346e-05,
366
+ "loss": 2.1801,
367
+ "step": 280
368
+ },
369
+ {
370
+ "epoch": 0.25,
371
+ "learning_rate": 9.938553245214138e-05,
372
+ "loss": 2.2159,
373
+ "step": 285
374
+ },
375
+ {
376
+ "epoch": 0.25,
377
+ "learning_rate": 9.934781014045545e-05,
378
+ "loss": 2.1817,
379
+ "step": 290
380
+ },
381
+ {
382
+ "epoch": 0.26,
383
+ "learning_rate": 9.93089717632953e-05,
384
+ "loss": 2.2027,
385
+ "step": 295
386
+ },
387
+ {
388
+ "epoch": 0.26,
389
+ "learning_rate": 9.926901819904178e-05,
390
+ "loss": 2.1994,
391
+ "step": 300
392
+ },
393
+ {
394
+ "epoch": 0.26,
395
+ "eval_loss": 2.1994447708129883,
396
+ "eval_runtime": 435.4362,
397
+ "eval_samples_per_second": 37.126,
398
+ "eval_steps_per_second": 1.162,
399
+ "step": 300
400
+ },
401
+ {
402
+ "epoch": 0.27,
403
+ "learning_rate": 9.922795035129713e-05,
404
+ "loss": 2.199,
405
+ "step": 305
406
+ },
407
+ {
408
+ "epoch": 0.27,
409
+ "learning_rate": 9.918576914886465e-05,
410
+ "loss": 2.1699,
411
+ "step": 310
412
+ },
413
+ {
414
+ "epoch": 0.28,
415
+ "learning_rate": 9.914247554572757e-05,
416
+ "loss": 2.1873,
417
+ "step": 315
418
+ },
419
+ {
420
+ "epoch": 0.28,
421
+ "learning_rate": 9.909807052102751e-05,
422
+ "loss": 2.1832,
423
+ "step": 320
424
+ },
425
+ {
426
+ "epoch": 0.28,
427
+ "learning_rate": 9.905255507904238e-05,
428
+ "loss": 2.1833,
429
+ "step": 325
430
+ },
431
+ {
432
+ "epoch": 0.29,
433
+ "learning_rate": 9.900593024916365e-05,
434
+ "loss": 2.1766,
435
+ "step": 330
436
+ },
437
+ {
438
+ "epoch": 0.29,
439
+ "learning_rate": 9.895819708587295e-05,
440
+ "loss": 2.2268,
441
+ "step": 335
442
+ },
443
+ {
444
+ "epoch": 0.3,
445
+ "learning_rate": 9.890935666871847e-05,
446
+ "loss": 2.1813,
447
+ "step": 340
448
+ },
449
+ {
450
+ "epoch": 0.3,
451
+ "learning_rate": 9.885941010229027e-05,
452
+ "loss": 2.186,
453
+ "step": 345
454
+ },
455
+ {
456
+ "epoch": 0.31,
457
+ "learning_rate": 9.880835851619554e-05,
458
+ "loss": 2.1832,
459
+ "step": 350
460
+ },
461
+ {
462
+ "epoch": 0.31,
463
+ "learning_rate": 9.875620306503282e-05,
464
+ "loss": 2.1824,
465
+ "step": 355
466
+ },
467
+ {
468
+ "epoch": 0.32,
469
+ "learning_rate": 9.870294492836612e-05,
470
+ "loss": 2.1898,
471
+ "step": 360
472
+ },
473
+ {
474
+ "epoch": 0.32,
475
+ "learning_rate": 9.864858531069807e-05,
476
+ "loss": 2.1846,
477
+ "step": 365
478
+ },
479
+ {
480
+ "epoch": 0.32,
481
+ "learning_rate": 9.859312544144272e-05,
482
+ "loss": 2.1772,
483
+ "step": 370
484
+ },
485
+ {
486
+ "epoch": 0.33,
487
+ "learning_rate": 9.85365665748978e-05,
488
+ "loss": 2.1724,
489
+ "step": 375
490
+ },
491
+ {
492
+ "epoch": 0.33,
493
+ "learning_rate": 9.847890999021629e-05,
494
+ "loss": 2.1862,
495
+ "step": 380
496
+ },
497
+ {
498
+ "epoch": 0.34,
499
+ "learning_rate": 9.84201569913775e-05,
500
+ "loss": 2.1746,
501
+ "step": 385
502
+ },
503
+ {
504
+ "epoch": 0.34,
505
+ "learning_rate": 9.83603089071576e-05,
506
+ "loss": 2.1679,
507
+ "step": 390
508
+ },
509
+ {
510
+ "epoch": 0.35,
511
+ "learning_rate": 9.829936709109949e-05,
512
+ "loss": 2.1705,
513
+ "step": 395
514
+ },
515
+ {
516
+ "epoch": 0.35,
517
+ "learning_rate": 9.823733292148236e-05,
518
+ "loss": 2.1767,
519
+ "step": 400
520
+ },
521
+ {
522
+ "epoch": 0.35,
523
+ "eval_loss": 2.186896800994873,
524
+ "eval_runtime": 433.1626,
525
+ "eval_samples_per_second": 37.321,
526
+ "eval_steps_per_second": 1.168,
527
+ "step": 400
528
+ },
529
+ {
530
+ "epoch": 0.35,
531
+ "learning_rate": 9.817420780129031e-05,
532
+ "loss": 2.1981,
533
+ "step": 405
534
+ },
535
+ {
536
+ "epoch": 0.36,
537
+ "learning_rate": 9.810999315818077e-05,
538
+ "loss": 2.1918,
539
+ "step": 410
540
+ },
541
+ {
542
+ "epoch": 0.36,
543
+ "learning_rate": 9.804469044445215e-05,
544
+ "loss": 2.1802,
545
+ "step": 415
546
+ },
547
+ {
548
+ "epoch": 0.37,
549
+ "learning_rate": 9.797830113701101e-05,
550
+ "loss": 2.1797,
551
+ "step": 420
552
+ },
553
+ {
554
+ "epoch": 0.37,
555
+ "learning_rate": 9.791082673733864e-05,
556
+ "loss": 2.1725,
557
+ "step": 425
558
+ },
559
+ {
560
+ "epoch": 0.38,
561
+ "learning_rate": 9.784226877145712e-05,
562
+ "loss": 2.1803,
563
+ "step": 430
564
+ },
565
+ {
566
+ "epoch": 0.38,
567
+ "learning_rate": 9.77726287898948e-05,
568
+ "loss": 2.1648,
569
+ "step": 435
570
+ },
571
+ {
572
+ "epoch": 0.39,
573
+ "learning_rate": 9.770190836765123e-05,
574
+ "loss": 2.1775,
575
+ "step": 440
576
+ },
577
+ {
578
+ "epoch": 0.39,
579
+ "learning_rate": 9.763010910416155e-05,
580
+ "loss": 2.1689,
581
+ "step": 445
582
+ },
583
+ {
584
+ "epoch": 0.39,
585
+ "learning_rate": 9.755723262326031e-05,
586
+ "loss": 2.1968,
587
+ "step": 450
588
+ },
589
+ {
590
+ "epoch": 0.4,
591
+ "learning_rate": 9.748328057314476e-05,
592
+ "loss": 2.1676,
593
+ "step": 455
594
+ },
595
+ {
596
+ "epoch": 0.4,
597
+ "learning_rate": 9.740825462633752e-05,
598
+ "loss": 2.1657,
599
+ "step": 460
600
+ },
601
+ {
602
+ "epoch": 0.41,
603
+ "learning_rate": 9.733215647964884e-05,
604
+ "loss": 2.1778,
605
+ "step": 465
606
+ },
607
+ {
608
+ "epoch": 0.41,
609
+ "learning_rate": 9.725498785413813e-05,
610
+ "loss": 2.1802,
611
+ "step": 470
612
+ },
613
+ {
614
+ "epoch": 0.42,
615
+ "learning_rate": 9.717675049507514e-05,
616
+ "loss": 2.1739,
617
+ "step": 475
618
+ },
619
+ {
620
+ "epoch": 0.42,
621
+ "learning_rate": 9.709744617190038e-05,
622
+ "loss": 2.1821,
623
+ "step": 480
624
+ },
625
+ {
626
+ "epoch": 0.42,
627
+ "learning_rate": 9.701707667818515e-05,
628
+ "loss": 2.1494,
629
+ "step": 485
630
+ },
631
+ {
632
+ "epoch": 0.43,
633
+ "learning_rate": 9.693564383159103e-05,
634
+ "loss": 2.1817,
635
+ "step": 490
636
+ },
637
+ {
638
+ "epoch": 0.43,
639
+ "learning_rate": 9.685314947382868e-05,
640
+ "loss": 2.1639,
641
+ "step": 495
642
+ },
643
+ {
644
+ "epoch": 0.44,
645
+ "learning_rate": 9.676959547061624e-05,
646
+ "loss": 2.1532,
647
+ "step": 500
648
+ },
649
+ {
650
+ "epoch": 0.44,
651
+ "eval_loss": 2.179184913635254,
652
+ "eval_runtime": 432.5255,
653
+ "eval_samples_per_second": 37.376,
654
+ "eval_steps_per_second": 1.17,
655
+ "step": 500
656
+ },
657
+ {
658
+ "epoch": 0.44,
659
+ "learning_rate": 9.668498371163712e-05,
660
+ "loss": 2.1665,
661
+ "step": 505
662
+ },
663
+ {
664
+ "epoch": 0.45,
665
+ "learning_rate": 9.659931611049729e-05,
666
+ "loss": 2.1692,
667
+ "step": 510
668
+ },
669
+ {
670
+ "epoch": 0.45,
671
+ "learning_rate": 9.651259460468191e-05,
672
+ "loss": 2.1739,
673
+ "step": 515
674
+ },
675
+ {
676
+ "epoch": 0.46,
677
+ "learning_rate": 9.642482115551164e-05,
678
+ "loss": 2.1475,
679
+ "step": 520
680
+ },
681
+ {
682
+ "epoch": 0.46,
683
+ "learning_rate": 9.63359977480982e-05,
684
+ "loss": 2.1647,
685
+ "step": 525
686
+ },
687
+ {
688
+ "epoch": 0.46,
689
+ "learning_rate": 9.62461263912995e-05,
690
+ "loss": 2.159,
691
+ "step": 530
692
+ },
693
+ {
694
+ "epoch": 0.47,
695
+ "learning_rate": 9.615520911767418e-05,
696
+ "loss": 2.1532,
697
+ "step": 535
698
+ },
699
+ {
700
+ "epoch": 0.47,
701
+ "learning_rate": 9.606324798343567e-05,
702
+ "loss": 2.1686,
703
+ "step": 540
704
+ },
705
+ {
706
+ "epoch": 0.48,
707
+ "learning_rate": 9.597024506840568e-05,
708
+ "loss": 2.156,
709
+ "step": 545
710
+ },
711
+ {
712
+ "epoch": 0.48,
713
+ "learning_rate": 9.587620247596716e-05,
714
+ "loss": 2.1666,
715
+ "step": 550
716
+ },
717
+ {
718
+ "epoch": 0.49,
719
+ "learning_rate": 9.578112233301675e-05,
720
+ "loss": 2.1713,
721
+ "step": 555
722
+ },
723
+ {
724
+ "epoch": 0.49,
725
+ "learning_rate": 9.568500678991661e-05,
726
+ "loss": 2.1769,
727
+ "step": 560
728
+ },
729
+ {
730
+ "epoch": 0.49,
731
+ "learning_rate": 9.558785802044586e-05,
732
+ "loss": 2.1737,
733
+ "step": 565
734
+ },
735
+ {
736
+ "epoch": 0.5,
737
+ "learning_rate": 9.54896782217514e-05,
738
+ "loss": 2.1649,
739
+ "step": 570
740
+ },
741
+ {
742
+ "epoch": 0.5,
743
+ "learning_rate": 9.53904696142982e-05,
744
+ "loss": 2.1734,
745
+ "step": 575
746
+ },
747
+ {
748
+ "epoch": 0.51,
749
+ "learning_rate": 9.529023444181907e-05,
750
+ "loss": 2.1535,
751
+ "step": 580
752
+ },
753
+ {
754
+ "epoch": 0.51,
755
+ "learning_rate": 9.518897497126396e-05,
756
+ "loss": 2.1719,
757
+ "step": 585
758
+ },
759
+ {
760
+ "epoch": 0.52,
761
+ "learning_rate": 9.508669349274866e-05,
762
+ "loss": 2.1583,
763
+ "step": 590
764
+ },
765
+ {
766
+ "epoch": 0.52,
767
+ "learning_rate": 9.498339231950298e-05,
768
+ "loss": 2.183,
769
+ "step": 595
770
+ },
771
+ {
772
+ "epoch": 0.53,
773
+ "learning_rate": 9.487907378781853e-05,
774
+ "loss": 2.171,
775
+ "step": 600
776
+ },
777
+ {
778
+ "epoch": 0.53,
779
+ "eval_loss": 2.1716604232788086,
780
+ "eval_runtime": 434.3111,
781
+ "eval_samples_per_second": 37.222,
782
+ "eval_steps_per_second": 1.165,
783
+ "step": 600
784
+ },
785
+ {
786
+ "epoch": 0.53,
787
+ "learning_rate": 9.477374025699577e-05,
788
+ "loss": 2.143,
789
+ "step": 605
790
+ },
791
+ {
792
+ "epoch": 0.53,
793
+ "learning_rate": 9.466739410929069e-05,
794
+ "loss": 2.155,
795
+ "step": 610
796
+ },
797
+ {
798
+ "epoch": 0.54,
799
+ "learning_rate": 9.456003774986095e-05,
800
+ "loss": 2.1776,
801
+ "step": 615
802
+ },
803
+ {
804
+ "epoch": 0.54,
805
+ "learning_rate": 9.44516736067115e-05,
806
+ "loss": 2.1561,
807
+ "step": 620
808
+ },
809
+ {
810
+ "epoch": 0.55,
811
+ "learning_rate": 9.43423041306396e-05,
812
+ "loss": 2.1547,
813
+ "step": 625
814
+ },
815
+ {
816
+ "epoch": 0.55,
817
+ "learning_rate": 9.423193179517947e-05,
818
+ "loss": 2.1554,
819
+ "step": 630
820
+ },
821
+ {
822
+ "epoch": 0.56,
823
+ "learning_rate": 9.412055909654627e-05,
824
+ "loss": 2.1779,
825
+ "step": 635
826
+ },
827
+ {
828
+ "epoch": 0.56,
829
+ "learning_rate": 9.400818855357972e-05,
830
+ "loss": 2.1636,
831
+ "step": 640
832
+ },
833
+ {
834
+ "epoch": 0.57,
835
+ "learning_rate": 9.38948227076871e-05,
836
+ "loss": 2.1668,
837
+ "step": 645
838
+ },
839
+ {
840
+ "epoch": 0.57,
841
+ "learning_rate": 9.378046412278574e-05,
842
+ "loss": 2.1548,
843
+ "step": 650
844
+ },
845
+ {
846
+ "epoch": 0.57,
847
+ "learning_rate": 9.366511538524509e-05,
848
+ "loss": 2.1594,
849
+ "step": 655
850
+ },
851
+ {
852
+ "epoch": 0.58,
853
+ "learning_rate": 9.35487791038282e-05,
854
+ "loss": 2.1619,
855
+ "step": 660
856
+ },
857
+ {
858
+ "epoch": 0.58,
859
+ "learning_rate": 9.343145790963273e-05,
860
+ "loss": 2.1541,
861
+ "step": 665
862
+ },
863
+ {
864
+ "epoch": 0.59,
865
+ "learning_rate": 9.331315445603137e-05,
866
+ "loss": 2.1423,
867
+ "step": 670
868
+ },
869
+ {
870
+ "epoch": 0.59,
871
+ "learning_rate": 9.3193871418612e-05,
872
+ "loss": 2.165,
873
+ "step": 675
874
+ },
875
+ {
876
+ "epoch": 0.6,
877
+ "learning_rate": 9.3073611495117e-05,
878
+ "loss": 2.1749,
879
+ "step": 680
880
+ },
881
+ {
882
+ "epoch": 0.6,
883
+ "learning_rate": 9.295237740538232e-05,
884
+ "loss": 2.154,
885
+ "step": 685
886
+ },
887
+ {
888
+ "epoch": 0.6,
889
+ "learning_rate": 9.283017189127598e-05,
890
+ "loss": 2.1557,
891
+ "step": 690
892
+ },
893
+ {
894
+ "epoch": 0.61,
895
+ "learning_rate": 9.270699771663602e-05,
896
+ "loss": 2.1496,
897
+ "step": 695
898
+ },
899
+ {
900
+ "epoch": 0.61,
901
+ "learning_rate": 9.258285766720802e-05,
902
+ "loss": 2.1588,
903
+ "step": 700
904
+ },
905
+ {
906
+ "epoch": 0.61,
907
+ "eval_loss": 2.1645405292510986,
908
+ "eval_runtime": 433.8926,
909
+ "eval_samples_per_second": 37.258,
910
+ "eval_steps_per_second": 1.166,
911
+ "step": 700
912
+ },
913
+ {
914
+ "epoch": 0.62,
915
+ "learning_rate": 9.245775455058208e-05,
916
+ "loss": 2.1585,
917
+ "step": 705
918
+ },
919
+ {
920
+ "epoch": 0.62,
921
+ "learning_rate": 9.23316911961293e-05,
922
+ "loss": 2.1316,
923
+ "step": 710
924
+ },
925
+ {
926
+ "epoch": 0.63,
927
+ "learning_rate": 9.220467045493787e-05,
928
+ "loss": 2.1385,
929
+ "step": 715
930
+ },
931
+ {
932
+ "epoch": 0.63,
933
+ "learning_rate": 9.207669519974851e-05,
934
+ "loss": 2.1692,
935
+ "step": 720
936
+ },
937
+ {
938
+ "epoch": 0.64,
939
+ "learning_rate": 9.194776832488952e-05,
940
+ "loss": 2.1358,
941
+ "step": 725
942
+ },
943
+ {
944
+ "epoch": 0.64,
945
+ "learning_rate": 9.181789274621133e-05,
946
+ "loss": 2.1739,
947
+ "step": 730
948
+ },
949
+ {
950
+ "epoch": 0.64,
951
+ "learning_rate": 9.168707140102057e-05,
952
+ "loss": 2.1616,
953
+ "step": 735
954
+ },
955
+ {
956
+ "epoch": 0.65,
957
+ "learning_rate": 9.155530724801363e-05,
958
+ "loss": 2.1517,
959
+ "step": 740
960
+ },
961
+ {
962
+ "epoch": 0.65,
963
+ "learning_rate": 9.142260326720969e-05,
964
+ "loss": 2.1642,
965
+ "step": 745
966
+ },
967
+ {
968
+ "epoch": 0.66,
969
+ "learning_rate": 9.128896245988338e-05,
970
+ "loss": 2.1499,
971
+ "step": 750
972
+ },
973
+ {
974
+ "epoch": 0.66,
975
+ "learning_rate": 9.115438784849692e-05,
976
+ "loss": 2.1646,
977
+ "step": 755
978
+ },
979
+ {
980
+ "epoch": 0.67,
981
+ "learning_rate": 9.101888247663173e-05,
982
+ "loss": 2.1492,
983
+ "step": 760
984
+ },
985
+ {
986
+ "epoch": 0.67,
987
+ "learning_rate": 9.088244940891957e-05,
988
+ "loss": 2.1331,
989
+ "step": 765
990
+ },
991
+ {
992
+ "epoch": 0.67,
993
+ "learning_rate": 9.074509173097328e-05,
994
+ "loss": 2.137,
995
+ "step": 770
996
+ },
997
+ {
998
+ "epoch": 0.68,
999
+ "learning_rate": 9.060681254931698e-05,
1000
+ "loss": 2.1401,
1001
+ "step": 775
1002
+ },
1003
+ {
1004
+ "epoch": 0.68,
1005
+ "learning_rate": 9.046761499131578e-05,
1006
+ "loss": 2.134,
1007
+ "step": 780
1008
+ },
1009
+ {
1010
+ "epoch": 0.69,
1011
+ "learning_rate": 9.032750220510509e-05,
1012
+ "loss": 2.1545,
1013
+ "step": 785
1014
+ },
1015
+ {
1016
+ "epoch": 0.69,
1017
+ "learning_rate": 9.018647735951943e-05,
1018
+ "loss": 2.1285,
1019
+ "step": 790
1020
+ },
1021
+ {
1022
+ "epoch": 0.7,
1023
+ "learning_rate": 9.00445436440207e-05,
1024
+ "loss": 2.1547,
1025
+ "step": 795
1026
+ },
1027
+ {
1028
+ "epoch": 0.7,
1029
+ "learning_rate": 8.990170426862612e-05,
1030
+ "loss": 2.145,
1031
+ "step": 800
1032
+ },
1033
+ {
1034
+ "epoch": 0.7,
1035
+ "eval_loss": 2.156743049621582,
1036
+ "eval_runtime": 439.2176,
1037
+ "eval_samples_per_second": 36.806,
1038
+ "eval_steps_per_second": 1.152,
1039
+ "step": 800
1040
+ },
1041
+ {
1042
+ "epoch": 0.71,
1043
+ "learning_rate": 8.975796246383556e-05,
1044
+ "loss": 2.128,
1045
+ "step": 805
1046
+ },
1047
+ {
1048
+ "epoch": 0.71,
1049
+ "learning_rate": 8.961332148055855e-05,
1050
+ "loss": 2.1334,
1051
+ "step": 810
1052
+ },
1053
+ {
1054
+ "epoch": 0.71,
1055
+ "learning_rate": 8.946778459004068e-05,
1056
+ "loss": 2.1313,
1057
+ "step": 815
1058
+ },
1059
+ {
1060
+ "epoch": 0.72,
1061
+ "learning_rate": 8.932135508378972e-05,
1062
+ "loss": 2.1222,
1063
+ "step": 820
1064
+ },
1065
+ {
1066
+ "epoch": 0.72,
1067
+ "learning_rate": 8.917403627350106e-05,
1068
+ "loss": 2.1448,
1069
+ "step": 825
1070
+ },
1071
+ {
1072
+ "epoch": 0.73,
1073
+ "learning_rate": 8.902583149098289e-05,
1074
+ "loss": 2.141,
1075
+ "step": 830
1076
+ },
1077
+ {
1078
+ "epoch": 0.73,
1079
+ "learning_rate": 8.887674408808082e-05,
1080
+ "loss": 2.166,
1081
+ "step": 835
1082
+ },
1083
+ {
1084
+ "epoch": 0.74,
1085
+ "learning_rate": 8.872677743660209e-05,
1086
+ "loss": 2.1459,
1087
+ "step": 840
1088
+ },
1089
+ {
1090
+ "epoch": 0.74,
1091
+ "learning_rate": 8.857593492823929e-05,
1092
+ "loss": 2.1573,
1093
+ "step": 845
1094
+ },
1095
+ {
1096
+ "epoch": 0.74,
1097
+ "learning_rate": 8.842421997449366e-05,
1098
+ "loss": 2.1362,
1099
+ "step": 850
1100
+ },
1101
+ {
1102
+ "epoch": 0.75,
1103
+ "learning_rate": 8.827163600659791e-05,
1104
+ "loss": 2.1476,
1105
+ "step": 855
1106
+ },
1107
+ {
1108
+ "epoch": 0.75,
1109
+ "learning_rate": 8.811818647543874e-05,
1110
+ "loss": 2.1424,
1111
+ "step": 860
1112
+ },
1113
+ {
1114
+ "epoch": 0.76,
1115
+ "learning_rate": 8.796387485147857e-05,
1116
+ "loss": 2.1579,
1117
+ "step": 865
1118
+ },
1119
+ {
1120
+ "epoch": 0.76,
1121
+ "learning_rate": 8.780870462467729e-05,
1122
+ "loss": 2.1499,
1123
+ "step": 870
1124
+ },
1125
+ {
1126
+ "epoch": 0.77,
1127
+ "learning_rate": 8.765267930441314e-05,
1128
+ "loss": 2.1309,
1129
+ "step": 875
1130
+ },
1131
+ {
1132
+ "epoch": 0.77,
1133
+ "learning_rate": 8.749580241940347e-05,
1134
+ "loss": 2.1447,
1135
+ "step": 880
1136
+ },
1137
+ {
1138
+ "epoch": 0.78,
1139
+ "learning_rate": 8.733807751762486e-05,
1140
+ "loss": 2.1302,
1141
+ "step": 885
1142
+ },
1143
+ {
1144
+ "epoch": 0.78,
1145
+ "learning_rate": 8.717950816623293e-05,
1146
+ "loss": 2.1313,
1147
+ "step": 890
1148
+ },
1149
+ {
1150
+ "epoch": 0.78,
1151
+ "learning_rate": 8.702009795148159e-05,
1152
+ "loss": 2.1356,
1153
+ "step": 895
1154
+ },
1155
+ {
1156
+ "epoch": 0.79,
1157
+ "learning_rate": 8.685985047864204e-05,
1158
+ "loss": 2.1366,
1159
+ "step": 900
1160
+ },
1161
+ {
1162
+ "epoch": 0.79,
1163
+ "eval_loss": 2.150692939758301,
1164
+ "eval_runtime": 443.8324,
1165
+ "eval_samples_per_second": 36.424,
1166
+ "eval_steps_per_second": 1.14,
1167
+ "step": 900
1168
+ },
1169
+ {
1170
+ "epoch": 0.79,
1171
+ "learning_rate": 8.66987693719211e-05,
1172
+ "loss": 2.1463,
1173
+ "step": 905
1174
+ },
1175
+ {
1176
+ "epoch": 0.8,
1177
+ "learning_rate": 8.653685827437936e-05,
1178
+ "loss": 2.1394,
1179
+ "step": 910
1180
+ },
1181
+ {
1182
+ "epoch": 0.8,
1183
+ "learning_rate": 8.637412084784873e-05,
1184
+ "loss": 2.1548,
1185
+ "step": 915
1186
+ },
1187
+ {
1188
+ "epoch": 0.81,
1189
+ "learning_rate": 8.621056077284964e-05,
1190
+ "loss": 2.129,
1191
+ "step": 920
1192
+ },
1193
+ {
1194
+ "epoch": 0.81,
1195
+ "learning_rate": 8.604618174850777e-05,
1196
+ "loss": 2.1367,
1197
+ "step": 925
1198
+ },
1199
+ {
1200
+ "epoch": 0.81,
1201
+ "learning_rate": 8.588098749247046e-05,
1202
+ "loss": 2.1304,
1203
+ "step": 930
1204
+ },
1205
+ {
1206
+ "epoch": 0.82,
1207
+ "learning_rate": 8.571498174082253e-05,
1208
+ "loss": 2.1517,
1209
+ "step": 935
1210
+ },
1211
+ {
1212
+ "epoch": 0.82,
1213
+ "learning_rate": 8.554816824800189e-05,
1214
+ "loss": 2.1461,
1215
+ "step": 940
1216
+ },
1217
+ {
1218
+ "epoch": 0.83,
1219
+ "learning_rate": 8.538055078671454e-05,
1220
+ "loss": 2.144,
1221
+ "step": 945
1222
+ },
1223
+ {
1224
+ "epoch": 0.83,
1225
+ "learning_rate": 8.521213314784931e-05,
1226
+ "loss": 2.1571,
1227
+ "step": 950
1228
+ },
1229
+ {
1230
+ "epoch": 0.84,
1231
+ "learning_rate": 8.504291914039204e-05,
1232
+ "loss": 2.1504,
1233
+ "step": 955
1234
+ },
1235
+ {
1236
+ "epoch": 0.84,
1237
+ "learning_rate": 8.487291259133956e-05,
1238
+ "loss": 2.1348,
1239
+ "step": 960
1240
+ },
1241
+ {
1242
+ "epoch": 0.85,
1243
+ "learning_rate": 8.470211734561305e-05,
1244
+ "loss": 2.1241,
1245
+ "step": 965
1246
+ },
1247
+ {
1248
+ "epoch": 0.85,
1249
+ "learning_rate": 8.453053726597103e-05,
1250
+ "loss": 2.1386,
1251
+ "step": 970
1252
+ },
1253
+ {
1254
+ "epoch": 0.85,
1255
+ "learning_rate": 8.43581762329222e-05,
1256
+ "loss": 2.1515,
1257
+ "step": 975
1258
+ },
1259
+ {
1260
+ "epoch": 0.86,
1261
+ "learning_rate": 8.418503814463741e-05,
1262
+ "loss": 2.1459,
1263
+ "step": 980
1264
+ },
1265
+ {
1266
+ "epoch": 0.86,
1267
+ "learning_rate": 8.401112691686175e-05,
1268
+ "loss": 2.1281,
1269
+ "step": 985
1270
+ },
1271
+ {
1272
+ "epoch": 0.87,
1273
+ "learning_rate": 8.383644648282578e-05,
1274
+ "loss": 2.1318,
1275
+ "step": 990
1276
+ },
1277
+ {
1278
+ "epoch": 0.87,
1279
+ "learning_rate": 8.366100079315672e-05,
1280
+ "loss": 2.1453,
1281
+ "step": 995
1282
+ },
1283
+ {
1284
+ "epoch": 0.88,
1285
+ "learning_rate": 8.348479381578904e-05,
1286
+ "loss": 2.1219,
1287
+ "step": 1000
1288
+ },
1289
+ {
1290
+ "epoch": 0.88,
1291
+ "eval_loss": 2.144993782043457,
1292
+ "eval_runtime": 448.6393,
1293
+ "eval_samples_per_second": 36.033,
1294
+ "eval_steps_per_second": 1.128,
1295
+ "step": 1000
1296
+ },
1297
+ {
1298
+ "epoch": 0.88,
1299
+ "learning_rate": 8.330782953587475e-05,
1300
+ "loss": 2.1227,
1301
+ "step": 1005
1302
+ },
1303
+ {
1304
+ "epoch": 0.88,
1305
+ "learning_rate": 8.31301119556932e-05,
1306
+ "loss": 2.1257,
1307
+ "step": 1010
1308
+ },
1309
+ {
1310
+ "epoch": 0.89,
1311
+ "learning_rate": 8.295164509456065e-05,
1312
+ "loss": 2.1231,
1313
+ "step": 1015
1314
+ },
1315
+ {
1316
+ "epoch": 0.89,
1317
+ "learning_rate": 8.277243298873936e-05,
1318
+ "loss": 2.1199,
1319
+ "step": 1020
1320
+ },
1321
+ {
1322
+ "epoch": 0.9,
1323
+ "learning_rate": 8.25924796913462e-05,
1324
+ "loss": 2.1376,
1325
+ "step": 1025
1326
+ },
1327
+ {
1328
+ "epoch": 0.9,
1329
+ "learning_rate": 8.241178927226112e-05,
1330
+ "loss": 2.1166,
1331
+ "step": 1030
1332
+ },
1333
+ {
1334
+ "epoch": 0.91,
1335
+ "learning_rate": 8.223036581803501e-05,
1336
+ "loss": 2.1259,
1337
+ "step": 1035
1338
+ },
1339
+ {
1340
+ "epoch": 0.91,
1341
+ "learning_rate": 8.204821343179737e-05,
1342
+ "loss": 2.1397,
1343
+ "step": 1040
1344
+ },
1345
+ {
1346
+ "epoch": 0.92,
1347
+ "learning_rate": 8.186533623316334e-05,
1348
+ "loss": 2.138,
1349
+ "step": 1045
1350
+ },
1351
+ {
1352
+ "epoch": 0.92,
1353
+ "learning_rate": 8.16817383581408e-05,
1354
+ "loss": 2.1282,
1355
+ "step": 1050
1356
+ },
1357
+ {
1358
+ "epoch": 0.92,
1359
+ "learning_rate": 8.149742395903653e-05,
1360
+ "loss": 2.1212,
1361
+ "step": 1055
1362
+ },
1363
+ {
1364
+ "epoch": 0.93,
1365
+ "learning_rate": 8.131239720436252e-05,
1366
+ "loss": 2.1642,
1367
+ "step": 1060
1368
+ },
1369
+ {
1370
+ "epoch": 0.93,
1371
+ "learning_rate": 8.11266622787416e-05,
1372
+ "loss": 2.1365,
1373
+ "step": 1065
1374
+ },
1375
+ {
1376
+ "epoch": 0.94,
1377
+ "learning_rate": 8.09402233828128e-05,
1378
+ "loss": 2.127,
1379
+ "step": 1070
1380
+ },
1381
+ {
1382
+ "epoch": 0.94,
1383
+ "learning_rate": 8.075308473313636e-05,
1384
+ "loss": 2.1207,
1385
+ "step": 1075
1386
+ },
1387
+ {
1388
+ "epoch": 0.95,
1389
+ "learning_rate": 8.056525056209841e-05,
1390
+ "loss": 2.131,
1391
+ "step": 1080
1392
+ },
1393
+ {
1394
+ "epoch": 0.95,
1395
+ "learning_rate": 8.037672511781516e-05,
1396
+ "loss": 2.1154,
1397
+ "step": 1085
1398
+ },
1399
+ {
1400
+ "epoch": 0.95,
1401
+ "learning_rate": 8.018751266403691e-05,
1402
+ "loss": 2.1133,
1403
+ "step": 1090
1404
+ },
1405
+ {
1406
+ "epoch": 0.96,
1407
+ "learning_rate": 7.999761748005156e-05,
1408
+ "loss": 2.1324,
1409
+ "step": 1095
1410
+ },
1411
+ {
1412
+ "epoch": 0.96,
1413
+ "learning_rate": 7.980704386058783e-05,
1414
+ "loss": 2.1415,
1415
+ "step": 1100
1416
+ },
1417
+ {
1418
+ "epoch": 0.96,
1419
+ "eval_loss": 2.1387038230895996,
1420
+ "eval_runtime": 444.7162,
1421
+ "eval_samples_per_second": 36.351,
1422
+ "eval_steps_per_second": 1.138,
1423
+ "step": 1100
1424
+ },
1425
+ {
1426
+ "epoch": 0.97,
1427
+ "learning_rate": 7.961579611571819e-05,
1428
+ "loss": 2.1378,
1429
+ "step": 1105
1430
+ },
1431
+ {
1432
+ "epoch": 0.97,
1433
+ "learning_rate": 7.94238785707613e-05,
1434
+ "loss": 2.1092,
1435
+ "step": 1110
1436
+ },
1437
+ {
1438
+ "epoch": 0.98,
1439
+ "learning_rate": 7.923129556618428e-05,
1440
+ "loss": 2.103,
1441
+ "step": 1115
1442
+ },
1443
+ {
1444
+ "epoch": 0.98,
1445
+ "learning_rate": 7.903805145750445e-05,
1446
+ "loss": 2.1229,
1447
+ "step": 1120
1448
+ },
1449
+ {
1450
+ "epoch": 0.99,
1451
+ "learning_rate": 7.884415061519086e-05,
1452
+ "loss": 2.1163,
1453
+ "step": 1125
1454
+ },
1455
+ {
1456
+ "epoch": 0.99,
1457
+ "learning_rate": 7.864959742456549e-05,
1458
+ "loss": 2.1291,
1459
+ "step": 1130
1460
+ },
1461
+ {
1462
+ "epoch": 0.99,
1463
+ "learning_rate": 7.845439628570402e-05,
1464
+ "loss": 2.1214,
1465
+ "step": 1135
1466
+ },
1467
+ {
1468
+ "epoch": 1.0,
1469
+ "learning_rate": 7.82585516133363e-05,
1470
+ "loss": 2.1326,
1471
+ "step": 1140
1472
+ },
1473
+ {
1474
+ "epoch": 1.0,
1475
+ "learning_rate": 7.806206783674657e-05,
1476
+ "loss": 2.0425,
1477
+ "step": 1145
1478
+ },
1479
+ {
1480
+ "epoch": 1.01,
1481
+ "learning_rate": 7.786494939967324e-05,
1482
+ "loss": 1.9622,
1483
+ "step": 1150
1484
+ },
1485
+ {
1486
+ "epoch": 1.01,
1487
+ "learning_rate": 7.76672007602084e-05,
1488
+ "loss": 1.9902,
1489
+ "step": 1155
1490
+ },
1491
+ {
1492
+ "epoch": 1.02,
1493
+ "learning_rate": 7.746882639069699e-05,
1494
+ "loss": 1.999,
1495
+ "step": 1160
1496
+ },
1497
+ {
1498
+ "epoch": 1.02,
1499
+ "learning_rate": 7.726983077763566e-05,
1500
+ "loss": 1.979,
1501
+ "step": 1165
1502
+ },
1503
+ {
1504
+ "epoch": 1.02,
1505
+ "learning_rate": 7.70702184215713e-05,
1506
+ "loss": 1.9913,
1507
+ "step": 1170
1508
+ },
1509
+ {
1510
+ "epoch": 1.03,
1511
+ "learning_rate": 7.686999383699924e-05,
1512
+ "loss": 1.9861,
1513
+ "step": 1175
1514
+ },
1515
+ {
1516
+ "epoch": 1.03,
1517
+ "learning_rate": 7.666916155226116e-05,
1518
+ "loss": 1.9913,
1519
+ "step": 1180
1520
+ },
1521
+ {
1522
+ "epoch": 1.04,
1523
+ "learning_rate": 7.646772610944271e-05,
1524
+ "loss": 1.9892,
1525
+ "step": 1185
1526
+ },
1527
+ {
1528
+ "epoch": 1.04,
1529
+ "learning_rate": 7.62656920642707e-05,
1530
+ "loss": 1.9885,
1531
+ "step": 1190
1532
+ },
1533
+ {
1534
+ "epoch": 1.05,
1535
+ "learning_rate": 7.606306398601017e-05,
1536
+ "loss": 1.9982,
1537
+ "step": 1195
1538
+ },
1539
+ {
1540
+ "epoch": 1.05,
1541
+ "learning_rate": 7.585984645736096e-05,
1542
+ "loss": 1.9765,
1543
+ "step": 1200
1544
+ },
1545
+ {
1546
+ "epoch": 1.05,
1547
+ "eval_loss": 2.144597053527832,
1548
+ "eval_runtime": 441.7594,
1549
+ "eval_samples_per_second": 36.595,
1550
+ "eval_steps_per_second": 1.145,
1551
+ "step": 1200
1552
+ },
1553
+ {
1554
+ "epoch": 1.06,
1555
+ "learning_rate": 7.56560440743541e-05,
1556
+ "loss": 2.0011,
1557
+ "step": 1205
1558
+ },
1559
+ {
1560
+ "epoch": 1.06,
1561
+ "learning_rate": 7.545166144624792e-05,
1562
+ "loss": 1.9738,
1563
+ "step": 1210
1564
+ },
1565
+ {
1566
+ "epoch": 1.06,
1567
+ "learning_rate": 7.52467031954237e-05,
1568
+ "loss": 1.9889,
1569
+ "step": 1215
1570
+ },
1571
+ {
1572
+ "epoch": 1.07,
1573
+ "learning_rate": 7.504117395728121e-05,
1574
+ "loss": 1.986,
1575
+ "step": 1220
1576
+ },
1577
+ {
1578
+ "epoch": 1.07,
1579
+ "learning_rate": 7.483507838013384e-05,
1580
+ "loss": 1.9961,
1581
+ "step": 1225
1582
+ },
1583
+ {
1584
+ "epoch": 1.08,
1585
+ "learning_rate": 7.462842112510346e-05,
1586
+ "loss": 1.9755,
1587
+ "step": 1230
1588
+ },
1589
+ {
1590
+ "epoch": 1.08,
1591
+ "learning_rate": 7.442120686601507e-05,
1592
+ "loss": 1.9793,
1593
+ "step": 1235
1594
+ },
1595
+ {
1596
+ "epoch": 1.09,
1597
+ "learning_rate": 7.421344028929102e-05,
1598
+ "loss": 1.986,
1599
+ "step": 1240
1600
+ },
1601
+ {
1602
+ "epoch": 1.09,
1603
+ "learning_rate": 7.4005126093845e-05,
1604
+ "loss": 1.9839,
1605
+ "step": 1245
1606
+ },
1607
+ {
1608
+ "epoch": 1.1,
1609
+ "learning_rate": 7.379626899097592e-05,
1610
+ "loss": 1.9917,
1611
+ "step": 1250
1612
+ },
1613
+ {
1614
+ "epoch": 1.1,
1615
+ "learning_rate": 7.358687370426117e-05,
1616
+ "loss": 2.0014,
1617
+ "step": 1255
1618
+ },
1619
+ {
1620
+ "epoch": 1.1,
1621
+ "learning_rate": 7.33769449694499e-05,
1622
+ "loss": 1.9931,
1623
+ "step": 1260
1624
+ },
1625
+ {
1626
+ "epoch": 1.11,
1627
+ "learning_rate": 7.31664875343559e-05,
1628
+ "loss": 1.9819,
1629
+ "step": 1265
1630
+ },
1631
+ {
1632
+ "epoch": 1.11,
1633
+ "learning_rate": 7.29555061587502e-05,
1634
+ "loss": 1.9744,
1635
+ "step": 1270
1636
+ },
1637
+ {
1638
+ "epoch": 1.12,
1639
+ "learning_rate": 7.274400561425341e-05,
1640
+ "loss": 1.9886,
1641
+ "step": 1275
1642
+ },
1643
+ {
1644
+ "epoch": 1.12,
1645
+ "learning_rate": 7.253199068422789e-05,
1646
+ "loss": 1.9983,
1647
+ "step": 1280
1648
+ },
1649
+ {
1650
+ "epoch": 1.13,
1651
+ "learning_rate": 7.231946616366943e-05,
1652
+ "loss": 1.9928,
1653
+ "step": 1285
1654
+ },
1655
+ {
1656
+ "epoch": 1.13,
1657
+ "learning_rate": 7.210643685909891e-05,
1658
+ "loss": 1.9915,
1659
+ "step": 1290
1660
+ },
1661
+ {
1662
+ "epoch": 1.13,
1663
+ "learning_rate": 7.189290758845355e-05,
1664
+ "loss": 1.9745,
1665
+ "step": 1295
1666
+ },
1667
+ {
1668
+ "epoch": 1.14,
1669
+ "learning_rate": 7.167888318097802e-05,
1670
+ "loss": 1.9837,
1671
+ "step": 1300
1672
+ },
1673
+ {
1674
+ "epoch": 1.14,
1675
+ "eval_loss": 2.1429882049560547,
1676
+ "eval_runtime": 442.1904,
1677
+ "eval_samples_per_second": 36.559,
1678
+ "eval_steps_per_second": 1.144,
1679
+ "step": 1300
1680
+ },
1681
+ {
1682
+ "epoch": 1.14,
1683
+ "learning_rate": 7.146436847711506e-05,
1684
+ "loss": 1.9898,
1685
+ "step": 1305
1686
+ },
1687
+ {
1688
+ "epoch": 1.15,
1689
+ "learning_rate": 7.124936832839621e-05,
1690
+ "loss": 1.9762,
1691
+ "step": 1310
1692
+ },
1693
+ {
1694
+ "epoch": 1.15,
1695
+ "learning_rate": 7.103388759733188e-05,
1696
+ "loss": 1.9928,
1697
+ "step": 1315
1698
+ },
1699
+ {
1700
+ "epoch": 1.16,
1701
+ "learning_rate": 7.081793115730153e-05,
1702
+ "loss": 1.9774,
1703
+ "step": 1320
1704
+ },
1705
+ {
1706
+ "epoch": 1.16,
1707
+ "learning_rate": 7.060150389244343e-05,
1708
+ "loss": 1.9905,
1709
+ "step": 1325
1710
+ },
1711
+ {
1712
+ "epoch": 1.17,
1713
+ "learning_rate": 7.038461069754414e-05,
1714
+ "loss": 1.9779,
1715
+ "step": 1330
1716
+ },
1717
+ {
1718
+ "epoch": 1.17,
1719
+ "learning_rate": 7.016725647792782e-05,
1720
+ "loss": 1.9917,
1721
+ "step": 1335
1722
+ },
1723
+ {
1724
+ "epoch": 1.17,
1725
+ "learning_rate": 6.994944614934535e-05,
1726
+ "loss": 1.9901,
1727
+ "step": 1340
1728
+ },
1729
+ {
1730
+ "epoch": 1.18,
1731
+ "learning_rate": 6.97311846378631e-05,
1732
+ "loss": 2.011,
1733
+ "step": 1345
1734
+ },
1735
+ {
1736
+ "epoch": 1.18,
1737
+ "learning_rate": 6.951247687975152e-05,
1738
+ "loss": 1.9745,
1739
+ "step": 1350
1740
+ },
1741
+ {
1742
+ "epoch": 1.19,
1743
+ "learning_rate": 6.929332782137348e-05,
1744
+ "loss": 2.0014,
1745
+ "step": 1355
1746
+ },
1747
+ {
1748
+ "epoch": 1.19,
1749
+ "learning_rate": 6.907374241907256e-05,
1750
+ "loss": 1.9771,
1751
+ "step": 1360
1752
+ },
1753
+ {
1754
+ "epoch": 1.2,
1755
+ "learning_rate": 6.885372563906067e-05,
1756
+ "loss": 1.9963,
1757
+ "step": 1365
1758
+ },
1759
+ {
1760
+ "epoch": 1.2,
1761
+ "learning_rate": 6.863328245730598e-05,
1762
+ "loss": 1.9887,
1763
+ "step": 1370
1764
+ },
1765
+ {
1766
+ "epoch": 1.2,
1767
+ "learning_rate": 6.841241785942032e-05,
1768
+ "loss": 1.9948,
1769
+ "step": 1375
1770
+ },
1771
+ {
1772
+ "epoch": 1.21,
1773
+ "learning_rate": 6.819113684054634e-05,
1774
+ "loss": 1.9912,
1775
+ "step": 1380
1776
+ },
1777
+ {
1778
+ "epoch": 1.21,
1779
+ "learning_rate": 6.79694444052446e-05,
1780
+ "loss": 1.9855,
1781
+ "step": 1385
1782
+ },
1783
+ {
1784
+ "epoch": 1.22,
1785
+ "learning_rate": 6.774734556738044e-05,
1786
+ "loss": 1.9882,
1787
+ "step": 1390
1788
+ },
1789
+ {
1790
+ "epoch": 1.22,
1791
+ "learning_rate": 6.752484535001045e-05,
1792
+ "loss": 1.9877,
1793
+ "step": 1395
1794
+ },
1795
+ {
1796
+ "epoch": 1.23,
1797
+ "learning_rate": 6.7301948785269e-05,
1798
+ "loss": 1.9952,
1799
+ "step": 1400
1800
+ },
1801
+ {
1802
+ "epoch": 1.23,
1803
+ "eval_loss": 2.1388282775878906,
1804
+ "eval_runtime": 440.5731,
1805
+ "eval_samples_per_second": 36.693,
1806
+ "eval_steps_per_second": 1.149,
1807
+ "step": 1400
1808
+ },
1809
+ {
1810
+ "epoch": 1.23,
1811
+ "learning_rate": 6.707866091425437e-05,
1812
+ "loss": 1.9885,
1813
+ "step": 1405
1814
+ },
1815
+ {
1816
+ "epoch": 1.24,
1817
+ "learning_rate": 6.685498678691471e-05,
1818
+ "loss": 2.0204,
1819
+ "step": 1410
1820
+ },
1821
+ {
1822
+ "epoch": 1.24,
1823
+ "learning_rate": 6.663093146193389e-05,
1824
+ "loss": 1.9935,
1825
+ "step": 1415
1826
+ },
1827
+ {
1828
+ "epoch": 1.24,
1829
+ "learning_rate": 6.640650000661708e-05,
1830
+ "loss": 2.0011,
1831
+ "step": 1420
1832
+ },
1833
+ {
1834
+ "epoch": 1.25,
1835
+ "learning_rate": 6.618169749677608e-05,
1836
+ "loss": 1.9993,
1837
+ "step": 1425
1838
+ },
1839
+ {
1840
+ "epoch": 1.25,
1841
+ "learning_rate": 6.595652901661464e-05,
1842
+ "loss": 1.9818,
1843
+ "step": 1430
1844
+ },
1845
+ {
1846
+ "epoch": 1.26,
1847
+ "learning_rate": 6.573099965861337e-05,
1848
+ "loss": 1.9964,
1849
+ "step": 1435
1850
+ },
1851
+ {
1852
+ "epoch": 1.26,
1853
+ "learning_rate": 6.550511452341459e-05,
1854
+ "loss": 1.9809,
1855
+ "step": 1440
1856
+ },
1857
+ {
1858
+ "epoch": 1.27,
1859
+ "learning_rate": 6.527887871970703e-05,
1860
+ "loss": 1.992,
1861
+ "step": 1445
1862
+ },
1863
+ {
1864
+ "epoch": 1.27,
1865
+ "learning_rate": 6.505229736411022e-05,
1866
+ "loss": 1.9682,
1867
+ "step": 1450
1868
+ },
1869
+ {
1870
+ "epoch": 1.27,
1871
+ "learning_rate": 6.482537558105885e-05,
1872
+ "loss": 1.9893,
1873
+ "step": 1455
1874
+ },
1875
+ {
1876
+ "epoch": 1.28,
1877
+ "learning_rate": 6.459811850268672e-05,
1878
+ "loss": 1.996,
1879
+ "step": 1460
1880
+ },
1881
+ {
1882
+ "epoch": 1.28,
1883
+ "learning_rate": 6.437053126871085e-05,
1884
+ "loss": 1.9839,
1885
+ "step": 1465
1886
+ },
1887
+ {
1888
+ "epoch": 1.29,
1889
+ "learning_rate": 6.414261902631515e-05,
1890
+ "loss": 1.9839,
1891
+ "step": 1470
1892
+ },
1893
+ {
1894
+ "epoch": 1.29,
1895
+ "learning_rate": 6.391438693003403e-05,
1896
+ "loss": 1.9891,
1897
+ "step": 1475
1898
+ },
1899
+ {
1900
+ "epoch": 1.3,
1901
+ "learning_rate": 6.368584014163581e-05,
1902
+ "loss": 1.9768,
1903
+ "step": 1480
1904
+ },
1905
+ {
1906
+ "epoch": 1.3,
1907
+ "learning_rate": 6.345698383000596e-05,
1908
+ "loss": 1.971,
1909
+ "step": 1485
1910
+ },
1911
+ {
1912
+ "epoch": 1.31,
1913
+ "learning_rate": 6.322782317103026e-05,
1914
+ "loss": 1.9826,
1915
+ "step": 1490
1916
+ },
1917
+ {
1918
+ "epoch": 1.31,
1919
+ "learning_rate": 6.299836334747768e-05,
1920
+ "loss": 1.9897,
1921
+ "step": 1495
1922
+ },
1923
+ {
1924
+ "epoch": 1.31,
1925
+ "learning_rate": 6.276860954888322e-05,
1926
+ "loss": 1.9868,
1927
+ "step": 1500
1928
+ },
1929
+ {
1930
+ "epoch": 1.31,
1931
+ "eval_loss": 2.1350769996643066,
1932
+ "eval_runtime": 441.8563,
1933
+ "eval_samples_per_second": 36.587,
1934
+ "eval_steps_per_second": 1.145,
1935
+ "step": 1500
1936
+ },
1937
+ {
1938
+ "epoch": 1.32,
1939
+ "learning_rate": 6.253856697143043e-05,
1940
+ "loss": 1.9757,
1941
+ "step": 1505
1942
+ },
1943
+ {
1944
+ "epoch": 1.32,
1945
+ "learning_rate": 6.23082408178341e-05,
1946
+ "loss": 1.9987,
1947
+ "step": 1510
1948
+ },
1949
+ {
1950
+ "epoch": 1.33,
1951
+ "learning_rate": 6.207763629722236e-05,
1952
+ "loss": 2.0011,
1953
+ "step": 1515
1954
+ },
1955
+ {
1956
+ "epoch": 1.33,
1957
+ "learning_rate": 6.184675862501901e-05,
1958
+ "loss": 2.0104,
1959
+ "step": 1520
1960
+ },
1961
+ {
1962
+ "epoch": 1.34,
1963
+ "learning_rate": 6.161561302282555e-05,
1964
+ "loss": 1.9858,
1965
+ "step": 1525
1966
+ },
1967
+ {
1968
+ "epoch": 1.34,
1969
+ "learning_rate": 6.138420471830307e-05,
1970
+ "loss": 1.9906,
1971
+ "step": 1530
1972
+ },
1973
+ {
1974
+ "epoch": 1.34,
1975
+ "learning_rate": 6.115253894505396e-05,
1976
+ "loss": 1.9911,
1977
+ "step": 1535
1978
+ },
1979
+ {
1980
+ "epoch": 1.35,
1981
+ "learning_rate": 6.092062094250367e-05,
1982
+ "loss": 1.9828,
1983
+ "step": 1540
1984
+ },
1985
+ {
1986
+ "epoch": 1.35,
1987
+ "learning_rate": 6.068845595578212e-05,
1988
+ "loss": 1.9726,
1989
+ "step": 1545
1990
+ },
1991
+ {
1992
+ "epoch": 1.36,
1993
+ "learning_rate": 6.0456049235605093e-05,
1994
+ "loss": 1.985,
1995
+ "step": 1550
1996
+ },
1997
+ {
1998
+ "epoch": 1.36,
1999
+ "learning_rate": 6.0223406038155514e-05,
2000
+ "loss": 1.9846,
2001
+ "step": 1555
2002
+ },
2003
+ {
2004
+ "epoch": 1.37,
2005
+ "learning_rate": 5.999053162496453e-05,
2006
+ "loss": 2.0014,
2007
+ "step": 1560
2008
+ },
2009
+ {
2010
+ "epoch": 1.37,
2011
+ "learning_rate": 5.975743126279254e-05,
2012
+ "loss": 1.9893,
2013
+ "step": 1565
2014
+ },
2015
+ {
2016
+ "epoch": 1.38,
2017
+ "learning_rate": 5.9524110223510076e-05,
2018
+ "loss": 1.9907,
2019
+ "step": 1570
2020
+ },
2021
+ {
2022
+ "epoch": 1.38,
2023
+ "learning_rate": 5.9290573783978585e-05,
2024
+ "loss": 1.9987,
2025
+ "step": 1575
2026
+ },
2027
+ {
2028
+ "epoch": 1.38,
2029
+ "learning_rate": 5.9056827225931064e-05,
2030
+ "loss": 1.9957,
2031
+ "step": 1580
2032
+ },
2033
+ {
2034
+ "epoch": 1.39,
2035
+ "learning_rate": 5.882287583585262e-05,
2036
+ "loss": 1.9827,
2037
+ "step": 1585
2038
+ },
2039
+ {
2040
+ "epoch": 1.39,
2041
+ "learning_rate": 5.858872490486089e-05,
2042
+ "loss": 1.9926,
2043
+ "step": 1590
2044
+ },
2045
+ {
2046
+ "epoch": 1.4,
2047
+ "learning_rate": 5.835437972858641e-05,
2048
+ "loss": 1.9924,
2049
+ "step": 1595
2050
+ },
2051
+ {
2052
+ "epoch": 1.4,
2053
+ "learning_rate": 5.8119845607052815e-05,
2054
+ "loss": 1.9864,
2055
+ "step": 1600
2056
+ },
2057
+ {
2058
+ "epoch": 1.4,
2059
+ "eval_loss": 2.131560802459717,
2060
+ "eval_runtime": 441.1354,
2061
+ "eval_samples_per_second": 36.646,
2062
+ "eval_steps_per_second": 1.147,
2063
+ "step": 1600
2064
+ },
2065
+ {
2066
+ "epoch": 1.41,
2067
+ "learning_rate": 5.7885127844556975e-05,
2068
+ "loss": 1.9917,
2069
+ "step": 1605
2070
+ },
2071
+ {
2072
+ "epoch": 1.41,
2073
+ "learning_rate": 5.7650231749549074e-05,
2074
+ "loss": 1.9837,
2075
+ "step": 1610
2076
+ },
2077
+ {
2078
+ "epoch": 1.41,
2079
+ "learning_rate": 5.741516263451251e-05,
2080
+ "loss": 1.9881,
2081
+ "step": 1615
2082
+ },
2083
+ {
2084
+ "epoch": 1.42,
2085
+ "learning_rate": 5.717992581584373e-05,
2086
+ "loss": 1.9762,
2087
+ "step": 1620
2088
+ },
2089
+ {
2090
+ "epoch": 1.42,
2091
+ "learning_rate": 5.694452661373203e-05,
2092
+ "loss": 1.9964,
2093
+ "step": 1625
2094
+ },
2095
+ {
2096
+ "epoch": 1.43,
2097
+ "learning_rate": 5.670897035203924e-05,
2098
+ "loss": 1.9896,
2099
+ "step": 1630
2100
+ },
2101
+ {
2102
+ "epoch": 1.43,
2103
+ "learning_rate": 5.647326235817928e-05,
2104
+ "loss": 1.9837,
2105
+ "step": 1635
2106
+ },
2107
+ {
2108
+ "epoch": 1.44,
2109
+ "learning_rate": 5.6237407962997704e-05,
2110
+ "loss": 1.9859,
2111
+ "step": 1640
2112
+ },
2113
+ {
2114
+ "epoch": 1.44,
2115
+ "learning_rate": 5.6001412500651104e-05,
2116
+ "loss": 1.987,
2117
+ "step": 1645
2118
+ },
2119
+ {
2120
+ "epoch": 1.45,
2121
+ "learning_rate": 5.5765281308486514e-05,
2122
+ "loss": 1.9873,
2123
+ "step": 1650
2124
+ },
2125
+ {
2126
+ "epoch": 1.45,
2127
+ "learning_rate": 5.5529019726920675e-05,
2128
+ "loss": 1.978,
2129
+ "step": 1655
2130
+ },
2131
+ {
2132
+ "epoch": 1.45,
2133
+ "learning_rate": 5.5292633099319224e-05,
2134
+ "loss": 1.9852,
2135
+ "step": 1660
2136
+ },
2137
+ {
2138
+ "epoch": 1.46,
2139
+ "learning_rate": 5.505612677187592e-05,
2140
+ "loss": 1.9987,
2141
+ "step": 1665
2142
+ },
2143
+ {
2144
+ "epoch": 1.46,
2145
+ "learning_rate": 5.481950609349166e-05,
2146
+ "loss": 1.9821,
2147
+ "step": 1670
2148
+ },
2149
+ {
2150
+ "epoch": 1.47,
2151
+ "learning_rate": 5.458277641565357e-05,
2152
+ "loss": 1.9828,
2153
+ "step": 1675
2154
+ },
2155
+ {
2156
+ "epoch": 1.47,
2157
+ "learning_rate": 5.434594309231389e-05,
2158
+ "loss": 1.9924,
2159
+ "step": 1680
2160
+ },
2161
+ {
2162
+ "epoch": 1.48,
2163
+ "learning_rate": 5.410901147976898e-05,
2164
+ "loss": 1.9699,
2165
+ "step": 1685
2166
+ },
2167
+ {
2168
+ "epoch": 1.48,
2169
+ "learning_rate": 5.387198693653813e-05,
2170
+ "loss": 1.9827,
2171
+ "step": 1690
2172
+ },
2173
+ {
2174
+ "epoch": 1.48,
2175
+ "learning_rate": 5.3634874823242386e-05,
2176
+ "loss": 1.9694,
2177
+ "step": 1695
2178
+ },
2179
+ {
2180
+ "epoch": 1.49,
2181
+ "learning_rate": 5.3397680502483306e-05,
2182
+ "loss": 1.987,
2183
+ "step": 1700
2184
+ },
2185
+ {
2186
+ "epoch": 1.49,
2187
+ "eval_loss": 2.1262922286987305,
2188
+ "eval_runtime": 438.0155,
2189
+ "eval_samples_per_second": 36.907,
2190
+ "eval_steps_per_second": 1.155,
2191
+ "step": 1700
2192
+ },
2193
+ {
2194
+ "epoch": 1.49,
2195
+ "learning_rate": 5.316040933872165e-05,
2196
+ "loss": 1.9838,
2197
+ "step": 1705
2198
+ },
2199
+ {
2200
+ "epoch": 1.5,
2201
+ "learning_rate": 5.292306669815611e-05,
2202
+ "loss": 1.9873,
2203
+ "step": 1710
2204
+ },
2205
+ {
2206
+ "epoch": 1.5,
2207
+ "learning_rate": 5.268565794860193e-05,
2208
+ "loss": 1.9737,
2209
+ "step": 1715
2210
+ },
2211
+ {
2212
+ "epoch": 1.51,
2213
+ "learning_rate": 5.244818845936944e-05,
2214
+ "loss": 1.997,
2215
+ "step": 1720
2216
+ },
2217
+ {
2218
+ "epoch": 1.51,
2219
+ "learning_rate": 5.221066360114275e-05,
2220
+ "loss": 1.9945,
2221
+ "step": 1725
2222
+ },
2223
+ {
2224
+ "epoch": 1.52,
2225
+ "learning_rate": 5.197308874585818e-05,
2226
+ "loss": 1.9926,
2227
+ "step": 1730
2228
+ },
2229
+ {
2230
+ "epoch": 1.52,
2231
+ "learning_rate": 5.173546926658277e-05,
2232
+ "loss": 1.9759,
2233
+ "step": 1735
2234
+ },
2235
+ {
2236
+ "epoch": 1.52,
2237
+ "learning_rate": 5.1497810537392844e-05,
2238
+ "loss": 1.9713,
2239
+ "step": 1740
2240
+ },
2241
+ {
2242
+ "epoch": 1.53,
2243
+ "learning_rate": 5.1260117933252384e-05,
2244
+ "loss": 1.9984,
2245
+ "step": 1745
2246
+ },
2247
+ {
2248
+ "epoch": 1.53,
2249
+ "learning_rate": 5.1022396829891484e-05,
2250
+ "loss": 1.972,
2251
+ "step": 1750
2252
+ },
2253
+ {
2254
+ "epoch": 1.54,
2255
+ "learning_rate": 5.0784652603684824e-05,
2256
+ "loss": 1.9857,
2257
+ "step": 1755
2258
+ },
2259
+ {
2260
+ "epoch": 1.54,
2261
+ "learning_rate": 5.0546890631530006e-05,
2262
+ "loss": 1.9922,
2263
+ "step": 1760
2264
+ },
2265
+ {
2266
+ "epoch": 1.55,
2267
+ "learning_rate": 5.0309116290726e-05,
2268
+ "loss": 1.9865,
2269
+ "step": 1765
2270
+ },
2271
+ {
2272
+ "epoch": 1.55,
2273
+ "learning_rate": 5.007133495885149e-05,
2274
+ "loss": 1.9732,
2275
+ "step": 1770
2276
+ },
2277
+ {
2278
+ "epoch": 1.55,
2279
+ "learning_rate": 4.9833552013643294e-05,
2280
+ "loss": 1.9873,
2281
+ "step": 1775
2282
+ },
2283
+ {
2284
+ "epoch": 1.56,
2285
+ "learning_rate": 4.959577283287471e-05,
2286
+ "loss": 1.9925,
2287
+ "step": 1780
2288
+ },
2289
+ {
2290
+ "epoch": 1.56,
2291
+ "learning_rate": 4.935800279423386e-05,
2292
+ "loss": 1.984,
2293
+ "step": 1785
2294
+ },
2295
+ {
2296
+ "epoch": 1.57,
2297
+ "learning_rate": 4.91202472752022e-05,
2298
+ "loss": 1.9769,
2299
+ "step": 1790
2300
+ },
2301
+ {
2302
+ "epoch": 1.57,
2303
+ "learning_rate": 4.888251165293267e-05,
2304
+ "loss": 1.984,
2305
+ "step": 1795
2306
+ },
2307
+ {
2308
+ "epoch": 1.58,
2309
+ "learning_rate": 4.8644801304128374e-05,
2310
+ "loss": 1.9678,
2311
+ "step": 1800
2312
+ },
2313
+ {
2314
+ "epoch": 1.58,
2315
+ "eval_loss": 2.1230146884918213,
2316
+ "eval_runtime": 438.1696,
2317
+ "eval_samples_per_second": 36.894,
2318
+ "eval_steps_per_second": 1.155,
2319
+ "step": 1800
2320
+ },
2321
+ {
2322
+ "epoch": 1.58,
2323
+ "learning_rate": 4.840712160492067e-05,
2324
+ "loss": 1.9793,
2325
+ "step": 1805
2326
+ },
2327
+ {
2328
+ "epoch": 1.59,
2329
+ "learning_rate": 4.8169477930747874e-05,
2330
+ "loss": 1.9652,
2331
+ "step": 1810
2332
+ },
2333
+ {
2334
+ "epoch": 1.59,
2335
+ "learning_rate": 4.7931875656233435e-05,
2336
+ "loss": 1.9659,
2337
+ "step": 1815
2338
+ },
2339
+ {
2340
+ "epoch": 1.59,
2341
+ "learning_rate": 4.769432015506456e-05,
2342
+ "loss": 1.9991,
2343
+ "step": 1820
2344
+ },
2345
+ {
2346
+ "epoch": 1.6,
2347
+ "learning_rate": 4.7456816799870607e-05,
2348
+ "loss": 1.9674,
2349
+ "step": 1825
2350
+ },
2351
+ {
2352
+ "epoch": 1.6,
2353
+ "learning_rate": 4.721937096210156e-05,
2354
+ "loss": 1.9692,
2355
+ "step": 1830
2356
+ },
2357
+ {
2358
+ "epoch": 1.61,
2359
+ "learning_rate": 4.698198801190659e-05,
2360
+ "loss": 1.9937,
2361
+ "step": 1835
2362
+ },
2363
+ {
2364
+ "epoch": 1.61,
2365
+ "learning_rate": 4.67446733180126e-05,
2366
+ "loss": 1.9792,
2367
+ "step": 1840
2368
+ },
2369
+ {
2370
+ "epoch": 1.62,
2371
+ "learning_rate": 4.6507432247602715e-05,
2372
+ "loss": 1.9883,
2373
+ "step": 1845
2374
+ },
2375
+ {
2376
+ "epoch": 1.62,
2377
+ "learning_rate": 4.627027016619508e-05,
2378
+ "loss": 1.9606,
2379
+ "step": 1850
2380
+ },
2381
+ {
2382
+ "epoch": 1.63,
2383
+ "learning_rate": 4.603319243752132e-05,
2384
+ "loss": 1.9699,
2385
+ "step": 1855
2386
+ },
2387
+ {
2388
+ "epoch": 1.63,
2389
+ "learning_rate": 4.57962044234053e-05,
2390
+ "loss": 1.9701,
2391
+ "step": 1860
2392
+ },
2393
+ {
2394
+ "epoch": 1.63,
2395
+ "learning_rate": 4.555931148364195e-05,
2396
+ "loss": 1.9763,
2397
+ "step": 1865
2398
+ },
2399
+ {
2400
+ "epoch": 1.64,
2401
+ "learning_rate": 4.5322518975875875e-05,
2402
+ "loss": 1.978,
2403
+ "step": 1870
2404
+ },
2405
+ {
2406
+ "epoch": 1.64,
2407
+ "learning_rate": 4.5085832255480386e-05,
2408
+ "loss": 1.9652,
2409
+ "step": 1875
2410
+ },
2411
+ {
2412
+ "epoch": 1.65,
2413
+ "learning_rate": 4.4849256675436165e-05,
2414
+ "loss": 1.9806,
2415
+ "step": 1880
2416
+ },
2417
+ {
2418
+ "epoch": 1.65,
2419
+ "learning_rate": 4.4612797586210394e-05,
2420
+ "loss": 1.9621,
2421
+ "step": 1885
2422
+ },
2423
+ {
2424
+ "epoch": 1.66,
2425
+ "learning_rate": 4.4376460335635594e-05,
2426
+ "loss": 1.9786,
2427
+ "step": 1890
2428
+ },
2429
+ {
2430
+ "epoch": 1.66,
2431
+ "learning_rate": 4.414025026878882e-05,
2432
+ "loss": 1.9622,
2433
+ "step": 1895
2434
+ },
2435
+ {
2436
+ "epoch": 1.66,
2437
+ "learning_rate": 4.3904172727870606e-05,
2438
+ "loss": 1.9827,
2439
+ "step": 1900
2440
+ },
2441
+ {
2442
+ "epoch": 1.66,
2443
+ "eval_loss": 2.1164419651031494,
2444
+ "eval_runtime": 440.4446,
2445
+ "eval_samples_per_second": 36.704,
2446
+ "eval_steps_per_second": 1.149,
2447
+ "step": 1900
2448
+ },
2449
+ {
2450
+ "epoch": 1.67,
2451
+ "learning_rate": 4.366823305208435e-05,
2452
+ "loss": 1.9846,
2453
+ "step": 1905
2454
+ },
2455
+ {
2456
+ "epoch": 1.67,
2457
+ "learning_rate": 4.343243657751533e-05,
2458
+ "loss": 1.9577,
2459
+ "step": 1910
2460
+ },
2461
+ {
2462
+ "epoch": 1.68,
2463
+ "learning_rate": 4.3196788637010254e-05,
2464
+ "loss": 1.9577,
2465
+ "step": 1915
2466
+ },
2467
+ {
2468
+ "epoch": 1.68,
2469
+ "learning_rate": 4.2961294560056445e-05,
2470
+ "loss": 1.9646,
2471
+ "step": 1920
2472
+ },
2473
+ {
2474
+ "epoch": 1.69,
2475
+ "learning_rate": 4.272595967266146e-05,
2476
+ "loss": 1.9788,
2477
+ "step": 1925
2478
+ },
2479
+ {
2480
+ "epoch": 1.69,
2481
+ "learning_rate": 4.249078929723256e-05,
2482
+ "loss": 1.9553,
2483
+ "step": 1930
2484
+ },
2485
+ {
2486
+ "epoch": 1.7,
2487
+ "learning_rate": 4.225578875245632e-05,
2488
+ "loss": 1.9815,
2489
+ "step": 1935
2490
+ },
2491
+ {
2492
+ "epoch": 1.7,
2493
+ "learning_rate": 4.202096335317841e-05,
2494
+ "loss": 1.9916,
2495
+ "step": 1940
2496
+ },
2497
+ {
2498
+ "epoch": 1.7,
2499
+ "learning_rate": 4.1786318410283345e-05,
2500
+ "loss": 1.9741,
2501
+ "step": 1945
2502
+ },
2503
+ {
2504
+ "epoch": 1.71,
2505
+ "learning_rate": 4.1551859230574375e-05,
2506
+ "loss": 1.964,
2507
+ "step": 1950
2508
+ },
2509
+ {
2510
+ "epoch": 1.71,
2511
+ "learning_rate": 4.131759111665349e-05,
2512
+ "loss": 1.9781,
2513
+ "step": 1955
2514
+ },
2515
+ {
2516
+ "epoch": 1.72,
2517
+ "learning_rate": 4.108351936680145e-05,
2518
+ "loss": 1.9615,
2519
+ "step": 1960
2520
+ },
2521
+ {
2522
+ "epoch": 1.72,
2523
+ "learning_rate": 4.0849649274857994e-05,
2524
+ "loss": 1.9783,
2525
+ "step": 1965
2526
+ },
2527
+ {
2528
+ "epoch": 1.73,
2529
+ "learning_rate": 4.061598613010213e-05,
2530
+ "loss": 1.9516,
2531
+ "step": 1970
2532
+ },
2533
+ {
2534
+ "epoch": 1.73,
2535
+ "learning_rate": 4.0382535217132414e-05,
2536
+ "loss": 1.9868,
2537
+ "step": 1975
2538
+ },
2539
+ {
2540
+ "epoch": 1.73,
2541
+ "learning_rate": 4.01493018157476e-05,
2542
+ "loss": 1.9674,
2543
+ "step": 1980
2544
+ },
2545
+ {
2546
+ "epoch": 1.74,
2547
+ "learning_rate": 3.991629120082703e-05,
2548
+ "loss": 1.9773,
2549
+ "step": 1985
2550
+ },
2551
+ {
2552
+ "epoch": 1.74,
2553
+ "learning_rate": 3.968350864221154e-05,
2554
+ "loss": 1.9787,
2555
+ "step": 1990
2556
+ },
2557
+ {
2558
+ "epoch": 1.75,
2559
+ "learning_rate": 3.945095940458407e-05,
2560
+ "loss": 1.9728,
2561
+ "step": 1995
2562
+ },
2563
+ {
2564
+ "epoch": 1.75,
2565
+ "learning_rate": 3.921864874735079e-05,
2566
+ "loss": 1.9846,
2567
+ "step": 2000
2568
+ },
2569
+ {
2570
+ "epoch": 1.75,
2571
+ "eval_loss": 2.113431930541992,
2572
+ "eval_runtime": 438.4756,
2573
+ "eval_samples_per_second": 36.869,
2574
+ "eval_steps_per_second": 1.154,
2575
+ "step": 2000
2576
+ },
2577
+ {
2578
+ "epoch": 1.76,
2579
+ "learning_rate": 3.8986581924522007e-05,
2580
+ "loss": 1.9738,
2581
+ "step": 2005
2582
+ },
2583
+ {
2584
+ "epoch": 1.76,
2585
+ "learning_rate": 3.875476418459339e-05,
2586
+ "loss": 1.9683,
2587
+ "step": 2010
2588
+ },
2589
+ {
2590
+ "epoch": 1.77,
2591
+ "learning_rate": 3.8523200770427326e-05,
2592
+ "loss": 1.9774,
2593
+ "step": 2015
2594
+ },
2595
+ {
2596
+ "epoch": 1.77,
2597
+ "learning_rate": 3.829189691913421e-05,
2598
+ "loss": 1.9757,
2599
+ "step": 2020
2600
+ },
2601
+ {
2602
+ "epoch": 1.77,
2603
+ "learning_rate": 3.806085786195417e-05,
2604
+ "loss": 1.9761,
2605
+ "step": 2025
2606
+ },
2607
+ {
2608
+ "epoch": 1.78,
2609
+ "learning_rate": 3.7830088824138585e-05,
2610
+ "loss": 1.9855,
2611
+ "step": 2030
2612
+ },
2613
+ {
2614
+ "epoch": 1.78,
2615
+ "learning_rate": 3.759959502483207e-05,
2616
+ "loss": 1.9749,
2617
+ "step": 2035
2618
+ },
2619
+ {
2620
+ "epoch": 1.79,
2621
+ "learning_rate": 3.7369381676954284e-05,
2622
+ "loss": 1.9515,
2623
+ "step": 2040
2624
+ },
2625
+ {
2626
+ "epoch": 1.79,
2627
+ "learning_rate": 3.713945398708217e-05,
2628
+ "loss": 1.965,
2629
+ "step": 2045
2630
+ },
2631
+ {
2632
+ "epoch": 1.8,
2633
+ "learning_rate": 3.690981715533211e-05,
2634
+ "loss": 1.9592,
2635
+ "step": 2050
2636
+ },
2637
+ {
2638
+ "epoch": 1.8,
2639
+ "learning_rate": 3.668047637524237e-05,
2640
+ "loss": 1.9767,
2641
+ "step": 2055
2642
+ },
2643
+ {
2644
+ "epoch": 1.8,
2645
+ "learning_rate": 3.645143683365555e-05,
2646
+ "loss": 1.9668,
2647
+ "step": 2060
2648
+ },
2649
+ {
2650
+ "epoch": 1.81,
2651
+ "learning_rate": 3.622270371060147e-05,
2652
+ "loss": 1.9478,
2653
+ "step": 2065
2654
+ },
2655
+ {
2656
+ "epoch": 1.81,
2657
+ "learning_rate": 3.5994282179179784e-05,
2658
+ "loss": 1.9671,
2659
+ "step": 2070
2660
+ },
2661
+ {
2662
+ "epoch": 1.82,
2663
+ "learning_rate": 3.576617740544312e-05,
2664
+ "loss": 1.9796,
2665
+ "step": 2075
2666
+ },
2667
+ {
2668
+ "epoch": 1.82,
2669
+ "learning_rate": 3.5538394548280255e-05,
2670
+ "loss": 1.98,
2671
+ "step": 2080
2672
+ },
2673
+ {
2674
+ "epoch": 1.83,
2675
+ "learning_rate": 3.531093875929934e-05,
2676
+ "loss": 1.9799,
2677
+ "step": 2085
2678
+ },
2679
+ {
2680
+ "epoch": 1.83,
2681
+ "learning_rate": 3.5083815182711524e-05,
2682
+ "loss": 1.9712,
2683
+ "step": 2090
2684
+ },
2685
+ {
2686
+ "epoch": 1.84,
2687
+ "learning_rate": 3.485702895521444e-05,
2688
+ "loss": 1.9687,
2689
+ "step": 2095
2690
+ },
2691
+ {
2692
+ "epoch": 1.84,
2693
+ "learning_rate": 3.463058520587625e-05,
2694
+ "loss": 1.9694,
2695
+ "step": 2100
2696
+ },
2697
+ {
2698
+ "epoch": 1.84,
2699
+ "eval_loss": 2.1068317890167236,
2700
+ "eval_runtime": 436.7843,
2701
+ "eval_samples_per_second": 37.011,
2702
+ "eval_steps_per_second": 1.158,
2703
+ "step": 2100
2704
+ },
2705
+ {
2706
+ "epoch": 1.84,
2707
+ "learning_rate": 3.440448905601939e-05,
2708
+ "loss": 1.971,
2709
+ "step": 2105
2710
+ },
2711
+ {
2712
+ "epoch": 1.85,
2713
+ "learning_rate": 3.417874561910498e-05,
2714
+ "loss": 1.9587,
2715
+ "step": 2110
2716
+ },
2717
+ {
2718
+ "epoch": 1.85,
2719
+ "learning_rate": 3.395336000061702e-05,
2720
+ "loss": 1.9869,
2721
+ "step": 2115
2722
+ },
2723
+ {
2724
+ "epoch": 1.86,
2725
+ "learning_rate": 3.372833729794703e-05,
2726
+ "loss": 1.9621,
2727
+ "step": 2120
2728
+ },
2729
+ {
2730
+ "epoch": 1.86,
2731
+ "learning_rate": 3.350368260027862e-05,
2732
+ "loss": 1.9603,
2733
+ "step": 2125
2734
+ },
2735
+ {
2736
+ "epoch": 1.87,
2737
+ "learning_rate": 3.32794009884726e-05,
2738
+ "loss": 1.9715,
2739
+ "step": 2130
2740
+ },
2741
+ {
2742
+ "epoch": 1.87,
2743
+ "learning_rate": 3.305549753495186e-05,
2744
+ "loss": 1.9782,
2745
+ "step": 2135
2746
+ },
2747
+ {
2748
+ "epoch": 1.87,
2749
+ "learning_rate": 3.283197730358678e-05,
2750
+ "loss": 1.9622,
2751
+ "step": 2140
2752
+ },
2753
+ {
2754
+ "epoch": 1.88,
2755
+ "learning_rate": 3.260884534958068e-05,
2756
+ "loss": 1.9574,
2757
+ "step": 2145
2758
+ },
2759
+ {
2760
+ "epoch": 1.88,
2761
+ "learning_rate": 3.2386106719355444e-05,
2762
+ "loss": 1.9677,
2763
+ "step": 2150
2764
+ },
2765
+ {
2766
+ "epoch": 1.89,
2767
+ "learning_rate": 3.216376645043747e-05,
2768
+ "loss": 1.9754,
2769
+ "step": 2155
2770
+ },
2771
+ {
2772
+ "epoch": 1.89,
2773
+ "learning_rate": 3.194182957134365e-05,
2774
+ "loss": 1.9524,
2775
+ "step": 2160
2776
+ },
2777
+ {
2778
+ "epoch": 1.9,
2779
+ "learning_rate": 3.172030110146771e-05,
2780
+ "loss": 1.9704,
2781
+ "step": 2165
2782
+ },
2783
+ {
2784
+ "epoch": 1.9,
2785
+ "learning_rate": 3.149918605096666e-05,
2786
+ "loss": 1.9553,
2787
+ "step": 2170
2788
+ },
2789
+ {
2790
+ "epoch": 1.91,
2791
+ "learning_rate": 3.127848942064747e-05,
2792
+ "loss": 1.9811,
2793
+ "step": 2175
2794
+ },
2795
+ {
2796
+ "epoch": 1.91,
2797
+ "learning_rate": 3.1058216201854007e-05,
2798
+ "loss": 1.9698,
2799
+ "step": 2180
2800
+ },
2801
+ {
2802
+ "epoch": 1.91,
2803
+ "learning_rate": 3.083837137635416e-05,
2804
+ "loss": 1.9662,
2805
+ "step": 2185
2806
+ },
2807
+ {
2808
+ "epoch": 1.92,
2809
+ "learning_rate": 3.0618959916227056e-05,
2810
+ "loss": 1.9589,
2811
+ "step": 2190
2812
+ },
2813
+ {
2814
+ "epoch": 1.92,
2815
+ "learning_rate": 3.0399986783750806e-05,
2816
+ "loss": 1.9703,
2817
+ "step": 2195
2818
+ },
2819
+ {
2820
+ "epoch": 1.93,
2821
+ "learning_rate": 3.0181456931290064e-05,
2822
+ "loss": 1.9429,
2823
+ "step": 2200
2824
+ },
2825
+ {
2826
+ "epoch": 1.93,
2827
+ "eval_loss": 2.10353946685791,
2828
+ "eval_runtime": 434.9488,
2829
+ "eval_samples_per_second": 37.168,
2830
+ "eval_steps_per_second": 1.163,
2831
+ "step": 2200
2832
+ },
2833
+ {
2834
+ "epoch": 1.93,
2835
+ "learning_rate": 2.9963375301184204e-05,
2836
+ "loss": 1.9823,
2837
+ "step": 2205
2838
+ },
2839
+ {
2840
+ "epoch": 1.94,
2841
+ "learning_rate": 2.9745746825635402e-05,
2842
+ "loss": 1.9587,
2843
+ "step": 2210
2844
+ },
2845
+ {
2846
+ "epoch": 1.94,
2847
+ "learning_rate": 2.952857642659721e-05,
2848
+ "loss": 1.9622,
2849
+ "step": 2215
2850
+ },
2851
+ {
2852
+ "epoch": 1.94,
2853
+ "learning_rate": 2.9311869015663125e-05,
2854
+ "loss": 1.9736,
2855
+ "step": 2220
2856
+ },
2857
+ {
2858
+ "epoch": 1.95,
2859
+ "learning_rate": 2.9095629493955573e-05,
2860
+ "loss": 1.9581,
2861
+ "step": 2225
2862
+ },
2863
+ {
2864
+ "epoch": 1.95,
2865
+ "learning_rate": 2.8879862752015075e-05,
2866
+ "loss": 1.9603,
2867
+ "step": 2230
2868
+ },
2869
+ {
2870
+ "epoch": 1.96,
2871
+ "learning_rate": 2.866457366968958e-05,
2872
+ "loss": 1.96,
2873
+ "step": 2235
2874
+ },
2875
+ {
2876
+ "epoch": 1.96,
2877
+ "learning_rate": 2.8449767116024194e-05,
2878
+ "loss": 1.9828,
2879
+ "step": 2240
2880
+ },
2881
+ {
2882
+ "epoch": 1.97,
2883
+ "learning_rate": 2.8235447949150927e-05,
2884
+ "loss": 1.9684,
2885
+ "step": 2245
2886
+ },
2887
+ {
2888
+ "epoch": 1.97,
2889
+ "learning_rate": 2.802162101617899e-05,
2890
+ "loss": 1.9616,
2891
+ "step": 2250
2892
+ },
2893
+ {
2894
+ "epoch": 1.98,
2895
+ "learning_rate": 2.7808291153084988e-05,
2896
+ "loss": 1.9793,
2897
+ "step": 2255
2898
+ },
2899
+ {
2900
+ "epoch": 1.98,
2901
+ "learning_rate": 2.7595463184603705e-05,
2902
+ "loss": 1.9622,
2903
+ "step": 2260
2904
+ },
2905
+ {
2906
+ "epoch": 1.98,
2907
+ "learning_rate": 2.7383141924118865e-05,
2908
+ "loss": 1.9736,
2909
+ "step": 2265
2910
+ },
2911
+ {
2912
+ "epoch": 1.99,
2913
+ "learning_rate": 2.7171332173554394e-05,
2914
+ "loss": 1.9677,
2915
+ "step": 2270
2916
+ },
2917
+ {
2918
+ "epoch": 1.99,
2919
+ "learning_rate": 2.696003872326567e-05,
2920
+ "loss": 1.9506,
2921
+ "step": 2275
2922
+ },
2923
+ {
2924
+ "epoch": 2.0,
2925
+ "learning_rate": 2.674926635193134e-05,
2926
+ "loss": 1.9534,
2927
+ "step": 2280
2928
+ },
2929
+ {
2930
+ "epoch": 2.0,
2931
+ "learning_rate": 2.6539019826445117e-05,
2932
+ "loss": 1.8941,
2933
+ "step": 2285
2934
+ },
2935
+ {
2936
+ "epoch": 2.01,
2937
+ "learning_rate": 2.632930390180801e-05,
2938
+ "loss": 1.8062,
2939
+ "step": 2290
2940
+ },
2941
+ {
2942
+ "epoch": 2.01,
2943
+ "learning_rate": 2.612012332102085e-05,
2944
+ "loss": 1.796,
2945
+ "step": 2295
2946
+ },
2947
+ {
2948
+ "epoch": 2.01,
2949
+ "learning_rate": 2.5911482814976912e-05,
2950
+ "loss": 1.8079,
2951
+ "step": 2300
2952
+ },
2953
+ {
2954
+ "epoch": 2.01,
2955
+ "eval_loss": 2.136899471282959,
2956
+ "eval_runtime": 437.2683,
2957
+ "eval_samples_per_second": 36.97,
2958
+ "eval_steps_per_second": 1.157,
2959
+ "step": 2300
2960
+ },
2961
+ {
2962
+ "epoch": 2.02,
2963
+ "learning_rate": 2.5703387102355014e-05,
2964
+ "loss": 1.7992,
2965
+ "step": 2305
2966
+ },
2967
+ {
2968
+ "epoch": 2.02,
2969
+ "learning_rate": 2.5495840889512697e-05,
2970
+ "loss": 1.8124,
2971
+ "step": 2310
2972
+ },
2973
+ {
2974
+ "epoch": 2.03,
2975
+ "learning_rate": 2.5288848870379922e-05,
2976
+ "loss": 1.8099,
2977
+ "step": 2315
2978
+ },
2979
+ {
2980
+ "epoch": 2.03,
2981
+ "learning_rate": 2.5082415726352748e-05,
2982
+ "loss": 1.8173,
2983
+ "step": 2320
2984
+ },
2985
+ {
2986
+ "epoch": 2.04,
2987
+ "learning_rate": 2.487654612618762e-05,
2988
+ "loss": 1.8042,
2989
+ "step": 2325
2990
+ },
2991
+ {
2992
+ "epoch": 2.04,
2993
+ "learning_rate": 2.4671244725895637e-05,
2994
+ "loss": 1.8061,
2995
+ "step": 2330
2996
+ },
2997
+ {
2998
+ "epoch": 2.05,
2999
+ "learning_rate": 2.446651616863736e-05,
3000
+ "loss": 1.8031,
3001
+ "step": 2335
3002
+ },
3003
+ {
3004
+ "epoch": 2.05,
3005
+ "learning_rate": 2.42623650846177e-05,
3006
+ "loss": 1.8148,
3007
+ "step": 2340
3008
+ },
3009
+ {
3010
+ "epoch": 2.05,
3011
+ "learning_rate": 2.4058796090981313e-05,
3012
+ "loss": 1.8098,
3013
+ "step": 2345
3014
+ },
3015
+ {
3016
+ "epoch": 2.06,
3017
+ "learning_rate": 2.385581379170806e-05,
3018
+ "loss": 1.8066,
3019
+ "step": 2350
3020
+ },
3021
+ {
3022
+ "epoch": 2.06,
3023
+ "learning_rate": 2.3653422777508994e-05,
3024
+ "loss": 1.8125,
3025
+ "step": 2355
3026
+ },
3027
+ {
3028
+ "epoch": 2.07,
3029
+ "learning_rate": 2.345162762572244e-05,
3030
+ "loss": 1.8105,
3031
+ "step": 2360
3032
+ },
3033
+ {
3034
+ "epoch": 2.07,
3035
+ "learning_rate": 2.3250432900210507e-05,
3036
+ "loss": 1.8067,
3037
+ "step": 2365
3038
+ },
3039
+ {
3040
+ "epoch": 2.08,
3041
+ "learning_rate": 2.3049843151255936e-05,
3042
+ "loss": 1.8056,
3043
+ "step": 2370
3044
+ },
3045
+ {
3046
+ "epoch": 2.08,
3047
+ "learning_rate": 2.2849862915459036e-05,
3048
+ "loss": 1.7975,
3049
+ "step": 2375
3050
+ },
3051
+ {
3052
+ "epoch": 2.08,
3053
+ "learning_rate": 2.2650496715635283e-05,
3054
+ "loss": 1.8082,
3055
+ "step": 2380
3056
+ },
3057
+ {
3058
+ "epoch": 2.09,
3059
+ "learning_rate": 2.245174906071282e-05,
3060
+ "loss": 1.8125,
3061
+ "step": 2385
3062
+ },
3063
+ {
3064
+ "epoch": 2.09,
3065
+ "learning_rate": 2.2253624445630673e-05,
3066
+ "loss": 1.8078,
3067
+ "step": 2390
3068
+ },
3069
+ {
3070
+ "epoch": 2.1,
3071
+ "learning_rate": 2.2056127351236934e-05,
3072
+ "loss": 1.7967,
3073
+ "step": 2395
3074
+ },
3075
+ {
3076
+ "epoch": 2.1,
3077
+ "learning_rate": 2.1859262244187556e-05,
3078
+ "loss": 1.8132,
3079
+ "step": 2400
3080
+ },
3081
+ {
3082
+ "epoch": 2.1,
3083
+ "eval_loss": 2.1375112533569336,
3084
+ "eval_runtime": 437.2054,
3085
+ "eval_samples_per_second": 36.976,
3086
+ "eval_steps_per_second": 1.157,
3087
+ "step": 2400
3088
+ },
3089
+ {
3090
+ "epoch": 2.11,
3091
+ "learning_rate": 2.1663033576845205e-05,
3092
+ "loss": 1.8039,
3093
+ "step": 2405
3094
+ },
3095
+ {
3096
+ "epoch": 2.11,
3097
+ "learning_rate": 2.1467445787178686e-05,
3098
+ "loss": 1.7973,
3099
+ "step": 2410
3100
+ },
3101
+ {
3102
+ "epoch": 2.12,
3103
+ "learning_rate": 2.1272503298662477e-05,
3104
+ "loss": 1.8113,
3105
+ "step": 2415
3106
+ },
3107
+ {
3108
+ "epoch": 2.12,
3109
+ "learning_rate": 2.1078210520176717e-05,
3110
+ "loss": 1.8116,
3111
+ "step": 2420
3112
+ },
3113
+ {
3114
+ "epoch": 2.12,
3115
+ "learning_rate": 2.0884571845907546e-05,
3116
+ "loss": 1.809,
3117
+ "step": 2425
3118
+ },
3119
+ {
3120
+ "epoch": 2.13,
3121
+ "learning_rate": 2.069159165524762e-05,
3122
+ "loss": 1.8132,
3123
+ "step": 2430
3124
+ },
3125
+ {
3126
+ "epoch": 2.13,
3127
+ "learning_rate": 2.04992743126972e-05,
3128
+ "loss": 1.7959,
3129
+ "step": 2435
3130
+ },
3131
+ {
3132
+ "epoch": 2.14,
3133
+ "learning_rate": 2.0307624167765288e-05,
3134
+ "loss": 1.8103,
3135
+ "step": 2440
3136
+ },
3137
+ {
3138
+ "epoch": 2.14,
3139
+ "learning_rate": 2.011664555487136e-05,
3140
+ "loss": 1.8087,
3141
+ "step": 2445
3142
+ },
3143
+ {
3144
+ "epoch": 2.15,
3145
+ "learning_rate": 1.992634279324736e-05,
3146
+ "loss": 1.7934,
3147
+ "step": 2450
3148
+ },
3149
+ {
3150
+ "epoch": 2.15,
3151
+ "learning_rate": 1.973672018683991e-05,
3152
+ "loss": 1.8047,
3153
+ "step": 2455
3154
+ },
3155
+ {
3156
+ "epoch": 2.16,
3157
+ "learning_rate": 1.9547782024213047e-05,
3158
+ "loss": 1.7937,
3159
+ "step": 2460
3160
+ },
3161
+ {
3162
+ "epoch": 2.16,
3163
+ "learning_rate": 1.9359532578451245e-05,
3164
+ "loss": 1.8232,
3165
+ "step": 2465
3166
+ },
3167
+ {
3168
+ "epoch": 2.16,
3169
+ "learning_rate": 1.9171976107062684e-05,
3170
+ "loss": 1.8022,
3171
+ "step": 2470
3172
+ },
3173
+ {
3174
+ "epoch": 2.17,
3175
+ "learning_rate": 1.8985116851883088e-05,
3176
+ "loss": 1.8119,
3177
+ "step": 2475
3178
+ },
3179
+ {
3180
+ "epoch": 2.17,
3181
+ "learning_rate": 1.8798959038979652e-05,
3182
+ "loss": 1.7996,
3183
+ "step": 2480
3184
+ },
3185
+ {
3186
+ "epoch": 2.18,
3187
+ "learning_rate": 1.8613506878555597e-05,
3188
+ "loss": 1.8047,
3189
+ "step": 2485
3190
+ },
3191
+ {
3192
+ "epoch": 2.18,
3193
+ "learning_rate": 1.8428764564854828e-05,
3194
+ "loss": 1.7895,
3195
+ "step": 2490
3196
+ },
3197
+ {
3198
+ "epoch": 2.19,
3199
+ "learning_rate": 1.8244736276067198e-05,
3200
+ "loss": 1.8091,
3201
+ "step": 2495
3202
+ },
3203
+ {
3204
+ "epoch": 2.19,
3205
+ "learning_rate": 1.80614261742339e-05,
3206
+ "loss": 1.8043,
3207
+ "step": 2500
3208
+ },
3209
+ {
3210
+ "epoch": 2.19,
3211
+ "eval_loss": 2.136007308959961,
3212
+ "eval_runtime": 436.5151,
3213
+ "eval_samples_per_second": 37.034,
3214
+ "eval_steps_per_second": 1.159,
3215
+ "step": 2500
3216
+ },
3217
+ {
3218
+ "epoch": 2.19,
3219
+ "learning_rate": 1.7878838405153385e-05,
3220
+ "loss": 1.7965,
3221
+ "step": 2505
3222
+ },
3223
+ {
3224
+ "epoch": 2.2,
3225
+ "learning_rate": 1.7696977098287654e-05,
3226
+ "loss": 1.81,
3227
+ "step": 2510
3228
+ },
3229
+ {
3230
+ "epoch": 2.2,
3231
+ "learning_rate": 1.7515846366668737e-05,
3232
+ "loss": 1.8124,
3233
+ "step": 2515
3234
+ },
3235
+ {
3236
+ "epoch": 2.21,
3237
+ "learning_rate": 1.7335450306805827e-05,
3238
+ "loss": 1.7978,
3239
+ "step": 2520
3240
+ },
3241
+ {
3242
+ "epoch": 2.21,
3243
+ "learning_rate": 1.715579299859248e-05,
3244
+ "loss": 1.8076,
3245
+ "step": 2525
3246
+ },
3247
+ {
3248
+ "epoch": 2.22,
3249
+ "learning_rate": 1.697687850521447e-05,
3250
+ "loss": 1.8158,
3251
+ "step": 2530
3252
+ },
3253
+ {
3254
+ "epoch": 2.22,
3255
+ "learning_rate": 1.6798710873057787e-05,
3256
+ "loss": 1.7985,
3257
+ "step": 2535
3258
+ },
3259
+ {
3260
+ "epoch": 2.23,
3261
+ "learning_rate": 1.662129413161723e-05,
3262
+ "loss": 1.7988,
3263
+ "step": 2540
3264
+ },
3265
+ {
3266
+ "epoch": 2.23,
3267
+ "learning_rate": 1.644463229340516e-05,
3268
+ "loss": 1.7878,
3269
+ "step": 2545
3270
+ },
3271
+ {
3272
+ "epoch": 2.23,
3273
+ "learning_rate": 1.6268729353860867e-05,
3274
+ "loss": 1.8042,
3275
+ "step": 2550
3276
+ },
3277
+ {
3278
+ "epoch": 2.24,
3279
+ "learning_rate": 1.6093589291260103e-05,
3280
+ "loss": 1.8154,
3281
+ "step": 2555
3282
+ },
3283
+ {
3284
+ "epoch": 2.24,
3285
+ "learning_rate": 1.5919216066625208e-05,
3286
+ "loss": 1.7873,
3287
+ "step": 2560
3288
+ },
3289
+ {
3290
+ "epoch": 2.25,
3291
+ "learning_rate": 1.5745613623635413e-05,
3292
+ "loss": 1.8112,
3293
+ "step": 2565
3294
+ },
3295
+ {
3296
+ "epoch": 2.25,
3297
+ "learning_rate": 1.557278588853779e-05,
3298
+ "loss": 1.8059,
3299
+ "step": 2570
3300
+ },
3301
+ {
3302
+ "epoch": 2.26,
3303
+ "learning_rate": 1.5400736770058294e-05,
3304
+ "loss": 1.7951,
3305
+ "step": 2575
3306
+ },
3307
+ {
3308
+ "epoch": 2.26,
3309
+ "learning_rate": 1.522947015931348e-05,
3310
+ "loss": 1.807,
3311
+ "step": 2580
3312
+ },
3313
+ {
3314
+ "epoch": 2.26,
3315
+ "learning_rate": 1.5058989929722495e-05,
3316
+ "loss": 1.8121,
3317
+ "step": 2585
3318
+ },
3319
+ {
3320
+ "epoch": 2.27,
3321
+ "learning_rate": 1.4889299936919371e-05,
3322
+ "loss": 1.8003,
3323
+ "step": 2590
3324
+ },
3325
+ {
3326
+ "epoch": 2.27,
3327
+ "learning_rate": 1.4720404018665984e-05,
3328
+ "loss": 1.7861,
3329
+ "step": 2595
3330
+ },
3331
+ {
3332
+ "epoch": 2.28,
3333
+ "learning_rate": 1.4552305994765097e-05,
3334
+ "loss": 1.7927,
3335
+ "step": 2600
3336
+ },
3337
+ {
3338
+ "epoch": 2.28,
3339
+ "eval_loss": 2.1333673000335693,
3340
+ "eval_runtime": 436.639,
3341
+ "eval_samples_per_second": 37.024,
3342
+ "eval_steps_per_second": 1.159,
3343
+ "step": 2600
3344
+ },
3345
+ {
3346
+ "epoch": 2.28,
3347
+ "learning_rate": 1.4385009666974097e-05,
3348
+ "loss": 1.8122,
3349
+ "step": 2605
3350
+ },
3351
+ {
3352
+ "epoch": 2.29,
3353
+ "learning_rate": 1.4218518818918925e-05,
3354
+ "loss": 1.8045,
3355
+ "step": 2610
3356
+ },
3357
+ {
3358
+ "epoch": 2.29,
3359
+ "learning_rate": 1.4052837216008585e-05,
3360
+ "loss": 1.7979,
3361
+ "step": 2615
3362
+ },
3363
+ {
3364
+ "epoch": 2.3,
3365
+ "learning_rate": 1.3887968605349877e-05,
3366
+ "loss": 1.8117,
3367
+ "step": 2620
3368
+ },
3369
+ {
3370
+ "epoch": 2.3,
3371
+ "learning_rate": 1.3723916715662782e-05,
3372
+ "loss": 1.7819,
3373
+ "step": 2625
3374
+ },
3375
+ {
3376
+ "epoch": 2.3,
3377
+ "learning_rate": 1.356068525719601e-05,
3378
+ "loss": 1.807,
3379
+ "step": 2630
3380
+ },
3381
+ {
3382
+ "epoch": 2.31,
3383
+ "learning_rate": 1.3398277921643188e-05,
3384
+ "loss": 1.7996,
3385
+ "step": 2635
3386
+ },
3387
+ {
3388
+ "epoch": 2.31,
3389
+ "learning_rate": 1.3236698382059287e-05,
3390
+ "loss": 1.7984,
3391
+ "step": 2640
3392
+ },
3393
+ {
3394
+ "epoch": 2.32,
3395
+ "learning_rate": 1.3075950292777588e-05,
3396
+ "loss": 1.7975,
3397
+ "step": 2645
3398
+ },
3399
+ {
3400
+ "epoch": 2.32,
3401
+ "learning_rate": 1.2916037289327065e-05,
3402
+ "loss": 1.7987,
3403
+ "step": 2650
3404
+ },
3405
+ {
3406
+ "epoch": 2.33,
3407
+ "learning_rate": 1.2756962988350108e-05,
3408
+ "loss": 1.7958,
3409
+ "step": 2655
3410
+ },
3411
+ {
3412
+ "epoch": 2.33,
3413
+ "learning_rate": 1.2598730987520718e-05,
3414
+ "loss": 1.8064,
3415
+ "step": 2660
3416
+ },
3417
+ {
3418
+ "epoch": 2.33,
3419
+ "learning_rate": 1.2441344865463245e-05,
3420
+ "loss": 1.8101,
3421
+ "step": 2665
3422
+ },
3423
+ {
3424
+ "epoch": 2.34,
3425
+ "learning_rate": 1.2284808181671325e-05,
3426
+ "loss": 1.8195,
3427
+ "step": 2670
3428
+ },
3429
+ {
3430
+ "epoch": 2.34,
3431
+ "learning_rate": 1.212912447642744e-05,
3432
+ "loss": 1.7939,
3433
+ "step": 2675
3434
+ },
3435
+ {
3436
+ "epoch": 2.35,
3437
+ "learning_rate": 1.1974297270722867e-05,
3438
+ "loss": 1.8037,
3439
+ "step": 2680
3440
+ },
3441
+ {
3442
+ "epoch": 2.35,
3443
+ "learning_rate": 1.1820330066177998e-05,
3444
+ "loss": 1.7916,
3445
+ "step": 2685
3446
+ },
3447
+ {
3448
+ "epoch": 2.36,
3449
+ "learning_rate": 1.1667226344963194e-05,
3450
+ "loss": 1.7948,
3451
+ "step": 2690
3452
+ },
3453
+ {
3454
+ "epoch": 2.36,
3455
+ "learning_rate": 1.151498956971997e-05,
3456
+ "loss": 1.8054,
3457
+ "step": 2695
3458
+ },
3459
+ {
3460
+ "epoch": 2.37,
3461
+ "learning_rate": 1.1363623183482775e-05,
3462
+ "loss": 1.7935,
3463
+ "step": 2700
3464
+ },
3465
+ {
3466
+ "epoch": 2.37,
3467
+ "eval_loss": 2.133516550064087,
3468
+ "eval_runtime": 436.7537,
3469
+ "eval_samples_per_second": 37.014,
3470
+ "eval_steps_per_second": 1.159,
3471
+ "step": 2700
3472
+ },
3473
+ {
3474
+ "epoch": 2.37,
3475
+ "learning_rate": 1.1213130609601003e-05,
3476
+ "loss": 1.8049,
3477
+ "step": 2705
3478
+ },
3479
+ {
3480
+ "epoch": 2.37,
3481
+ "learning_rate": 1.10635152516617e-05,
3482
+ "loss": 1.8048,
3483
+ "step": 2710
3484
+ },
3485
+ {
3486
+ "epoch": 2.38,
3487
+ "learning_rate": 1.0914780493412468e-05,
3488
+ "loss": 1.812,
3489
+ "step": 2715
3490
+ },
3491
+ {
3492
+ "epoch": 2.38,
3493
+ "learning_rate": 1.0766929698685041e-05,
3494
+ "loss": 1.7978,
3495
+ "step": 2720
3496
+ },
3497
+ {
3498
+ "epoch": 2.39,
3499
+ "learning_rate": 1.0619966211319115e-05,
3500
+ "loss": 1.806,
3501
+ "step": 2725
3502
+ },
3503
+ {
3504
+ "epoch": 2.39,
3505
+ "learning_rate": 1.0473893355086773e-05,
3506
+ "loss": 1.803,
3507
+ "step": 2730
3508
+ },
3509
+ {
3510
+ "epoch": 2.4,
3511
+ "learning_rate": 1.0328714433617348e-05,
3512
+ "loss": 1.8001,
3513
+ "step": 2735
3514
+ },
3515
+ {
3516
+ "epoch": 2.4,
3517
+ "learning_rate": 1.0184432730322601e-05,
3518
+ "loss": 1.7951,
3519
+ "step": 2740
3520
+ },
3521
+ {
3522
+ "epoch": 2.4,
3523
+ "learning_rate": 1.004105150832259e-05,
3524
+ "loss": 1.8103,
3525
+ "step": 2745
3526
+ },
3527
+ {
3528
+ "epoch": 2.41,
3529
+ "learning_rate": 9.898574010371747e-06,
3530
+ "loss": 1.8028,
3531
+ "step": 2750
3532
+ },
3533
+ {
3534
+ "epoch": 2.41,
3535
+ "learning_rate": 9.757003458785663e-06,
3536
+ "loss": 1.8002,
3537
+ "step": 2755
3538
+ },
3539
+ {
3540
+ "epoch": 2.42,
3541
+ "learning_rate": 9.616343055368083e-06,
3542
+ "loss": 1.8008,
3543
+ "step": 2760
3544
+ },
3545
+ {
3546
+ "epoch": 2.42,
3547
+ "learning_rate": 9.47659598133861e-06,
3548
+ "loss": 1.8128,
3549
+ "step": 2765
3550
+ },
3551
+ {
3552
+ "epoch": 2.43,
3553
+ "learning_rate": 9.337765397260656e-06,
3554
+ "loss": 1.7967,
3555
+ "step": 2770
3556
+ },
3557
+ {
3558
+ "epoch": 2.43,
3559
+ "learning_rate": 9.199854442970068e-06,
3560
+ "loss": 1.8086,
3561
+ "step": 2775
3562
+ },
3563
+ {
3564
+ "epoch": 2.44,
3565
+ "learning_rate": 9.062866237503993e-06,
3566
+ "loss": 1.8022,
3567
+ "step": 2780
3568
+ },
3569
+ {
3570
+ "epoch": 2.44,
3571
+ "learning_rate": 8.926803879030466e-06,
3572
+ "loss": 1.8006,
3573
+ "step": 2785
3574
+ },
3575
+ {
3576
+ "epoch": 2.44,
3577
+ "learning_rate": 8.791670444778228e-06,
3578
+ "loss": 1.7894,
3579
+ "step": 2790
3580
+ },
3581
+ {
3582
+ "epoch": 2.45,
3583
+ "learning_rate": 8.6574689909672e-06,
3584
+ "loss": 1.798,
3585
+ "step": 2795
3586
+ },
3587
+ {
3588
+ "epoch": 2.45,
3589
+ "learning_rate": 8.524202552739357e-06,
3590
+ "loss": 1.7982,
3591
+ "step": 2800
3592
+ },
3593
+ {
3594
+ "epoch": 2.45,
3595
+ "eval_loss": 2.1320557594299316,
3596
+ "eval_runtime": 436.1068,
3597
+ "eval_samples_per_second": 37.069,
3598
+ "eval_steps_per_second": 1.16,
3599
+ "step": 2800
3600
+ },
3601
+ {
3602
+ "epoch": 2.46,
3603
+ "learning_rate": 8.391874144090029e-06,
3604
+ "loss": 1.794,
3605
+ "step": 2805
3606
+ },
3607
+ {
3608
+ "epoch": 2.46,
3609
+ "learning_rate": 8.260486757799834e-06,
3610
+ "loss": 1.8082,
3611
+ "step": 2810
3612
+ },
3613
+ {
3614
+ "epoch": 2.47,
3615
+ "learning_rate": 8.130043365366873e-06,
3616
+ "loss": 1.8078,
3617
+ "step": 2815
3618
+ },
3619
+ {
3620
+ "epoch": 2.47,
3621
+ "learning_rate": 8.000546916939644e-06,
3622
+ "loss": 1.8009,
3623
+ "step": 2820
3624
+ },
3625
+ {
3626
+ "epoch": 2.47,
3627
+ "learning_rate": 7.872000341250213e-06,
3628
+ "loss": 1.7924,
3629
+ "step": 2825
3630
+ },
3631
+ {
3632
+ "epoch": 2.48,
3633
+ "learning_rate": 7.744406545548071e-06,
3634
+ "loss": 1.8071,
3635
+ "step": 2830
3636
+ },
3637
+ {
3638
+ "epoch": 2.48,
3639
+ "learning_rate": 7.617768415534305e-06,
3640
+ "loss": 1.8094,
3641
+ "step": 2835
3642
+ },
3643
+ {
3644
+ "epoch": 2.49,
3645
+ "learning_rate": 7.4920888152964e-06,
3646
+ "loss": 1.8116,
3647
+ "step": 2840
3648
+ },
3649
+ {
3650
+ "epoch": 2.49,
3651
+ "learning_rate": 7.367370587243394e-06,
3652
+ "loss": 1.8,
3653
+ "step": 2845
3654
+ },
3655
+ {
3656
+ "epoch": 2.5,
3657
+ "learning_rate": 7.243616552041677e-06,
3658
+ "loss": 1.8054,
3659
+ "step": 2850
3660
+ },
3661
+ {
3662
+ "epoch": 2.5,
3663
+ "learning_rate": 7.120829508551119e-06,
3664
+ "loss": 1.8068,
3665
+ "step": 2855
3666
+ },
3667
+ {
3668
+ "epoch": 2.51,
3669
+ "learning_rate": 6.999012233761798e-06,
3670
+ "loss": 1.784,
3671
+ "step": 2860
3672
+ },
3673
+ {
3674
+ "epoch": 2.51,
3675
+ "learning_rate": 6.878167482731251e-06,
3676
+ "loss": 1.8162,
3677
+ "step": 2865
3678
+ },
3679
+ {
3680
+ "epoch": 2.51,
3681
+ "learning_rate": 6.758297988522055e-06,
3682
+ "loss": 1.799,
3683
+ "step": 2870
3684
+ },
3685
+ {
3686
+ "epoch": 2.52,
3687
+ "learning_rate": 6.639406462140096e-06,
3688
+ "loss": 1.7919,
3689
+ "step": 2875
3690
+ },
3691
+ {
3692
+ "epoch": 2.52,
3693
+ "learning_rate": 6.521495592473259e-06,
3694
+ "loss": 1.7979,
3695
+ "step": 2880
3696
+ },
3697
+ {
3698
+ "epoch": 2.53,
3699
+ "learning_rate": 6.404568046230541e-06,
3700
+ "loss": 1.8096,
3701
+ "step": 2885
3702
+ },
3703
+ {
3704
+ "epoch": 2.53,
3705
+ "learning_rate": 6.2886264678818475e-06,
3706
+ "loss": 1.812,
3707
+ "step": 2890
3708
+ },
3709
+ {
3710
+ "epoch": 2.54,
3711
+ "learning_rate": 6.17367347959808e-06,
3712
+ "loss": 1.8072,
3713
+ "step": 2895
3714
+ },
3715
+ {
3716
+ "epoch": 2.54,
3717
+ "learning_rate": 6.059711681191888e-06,
3718
+ "loss": 1.8029,
3719
+ "step": 2900
3720
+ },
3721
+ {
3722
+ "epoch": 2.54,
3723
+ "eval_loss": 2.1311283111572266,
3724
+ "eval_runtime": 435.6098,
3725
+ "eval_samples_per_second": 37.111,
3726
+ "eval_steps_per_second": 1.162,
3727
+ "step": 2900
3728
+ },
3729
+ {
3730
+ "epoch": 2.54,
3731
+ "learning_rate": 5.946743650058889e-06,
3732
+ "loss": 1.7884,
3733
+ "step": 2905
3734
+ },
3735
+ {
3736
+ "epoch": 2.55,
3737
+ "learning_rate": 5.83477194111931e-06,
3738
+ "loss": 1.796,
3739
+ "step": 2910
3740
+ },
3741
+ {
3742
+ "epoch": 2.55,
3743
+ "learning_rate": 5.723799086760279e-06,
3744
+ "loss": 1.8026,
3745
+ "step": 2915
3746
+ },
3747
+ {
3748
+ "epoch": 2.56,
3749
+ "learning_rate": 5.613827596778487e-06,
3750
+ "loss": 1.7973,
3751
+ "step": 2920
3752
+ },
3753
+ {
3754
+ "epoch": 2.56,
3755
+ "learning_rate": 5.504859958323488e-06,
3756
+ "loss": 1.8,
3757
+ "step": 2925
3758
+ },
3759
+ {
3760
+ "epoch": 2.57,
3761
+ "learning_rate": 5.39689863584138e-06,
3762
+ "loss": 1.7905,
3763
+ "step": 2930
3764
+ },
3765
+ {
3766
+ "epoch": 2.57,
3767
+ "learning_rate": 5.289946071019142e-06,
3768
+ "loss": 1.8092,
3769
+ "step": 2935
3770
+ },
3771
+ {
3772
+ "epoch": 2.58,
3773
+ "learning_rate": 5.184004682729348e-06,
3774
+ "loss": 1.7966,
3775
+ "step": 2940
3776
+ },
3777
+ {
3778
+ "epoch": 2.58,
3779
+ "learning_rate": 5.0790768669754905e-06,
3780
+ "loss": 1.7979,
3781
+ "step": 2945
3782
+ },
3783
+ {
3784
+ "epoch": 2.58,
3785
+ "learning_rate": 4.975164996837811e-06,
3786
+ "loss": 1.7885,
3787
+ "step": 2950
3788
+ },
3789
+ {
3790
+ "epoch": 2.59,
3791
+ "learning_rate": 4.872271422419572e-06,
3792
+ "loss": 1.7951,
3793
+ "step": 2955
3794
+ },
3795
+ {
3796
+ "epoch": 2.59,
3797
+ "learning_rate": 4.770398470793991e-06,
3798
+ "loss": 1.7922,
3799
+ "step": 2960
3800
+ },
3801
+ {
3802
+ "epoch": 2.6,
3803
+ "learning_rate": 4.6695484459515045e-06,
3804
+ "loss": 1.7961,
3805
+ "step": 2965
3806
+ },
3807
+ {
3808
+ "epoch": 2.6,
3809
+ "learning_rate": 4.5697236287477685e-06,
3810
+ "loss": 1.8079,
3811
+ "step": 2970
3812
+ },
3813
+ {
3814
+ "epoch": 2.61,
3815
+ "learning_rate": 4.470926276851983e-06,
3816
+ "loss": 1.7979,
3817
+ "step": 2975
3818
+ },
3819
+ {
3820
+ "epoch": 2.61,
3821
+ "learning_rate": 4.373158624695917e-06,
3822
+ "loss": 1.8001,
3823
+ "step": 2980
3824
+ },
3825
+ {
3826
+ "epoch": 2.61,
3827
+ "learning_rate": 4.27642288342327e-06,
3828
+ "loss": 1.8045,
3829
+ "step": 2985
3830
+ },
3831
+ {
3832
+ "epoch": 2.62,
3833
+ "learning_rate": 4.18072124083978e-06,
3834
+ "loss": 1.7914,
3835
+ "step": 2990
3836
+ },
3837
+ {
3838
+ "epoch": 2.62,
3839
+ "learning_rate": 4.086055861363647e-06,
3840
+ "loss": 1.7971,
3841
+ "step": 2995
3842
+ },
3843
+ {
3844
+ "epoch": 2.63,
3845
+ "learning_rate": 3.992428885976652e-06,
3846
+ "loss": 1.7919,
3847
+ "step": 3000
3848
+ },
3849
+ {
3850
+ "epoch": 2.63,
3851
+ "eval_loss": 2.129803419113159,
3852
+ "eval_runtime": 437.5268,
3853
+ "eval_samples_per_second": 36.949,
3854
+ "eval_steps_per_second": 1.157,
3855
+ "step": 3000
3856
+ },
3857
+ {
3858
+ "epoch": 2.63,
3859
+ "learning_rate": 3.899842432175671e-06,
3860
+ "loss": 1.8002,
3861
+ "step": 3005
3862
+ },
3863
+ {
3864
+ "epoch": 2.64,
3865
+ "learning_rate": 3.8082985939248417e-06,
3866
+ "loss": 1.7827,
3867
+ "step": 3010
3868
+ },
3869
+ {
3870
+ "epoch": 2.64,
3871
+ "learning_rate": 3.7177994416081885e-06,
3872
+ "loss": 1.7983,
3873
+ "step": 3015
3874
+ },
3875
+ {
3876
+ "epoch": 2.65,
3877
+ "learning_rate": 3.6283470219827643e-06,
3878
+ "loss": 1.7828,
3879
+ "step": 3020
3880
+ },
3881
+ {
3882
+ "epoch": 2.65,
3883
+ "learning_rate": 3.539943358132425e-06,
3884
+ "loss": 1.7992,
3885
+ "step": 3025
3886
+ },
3887
+ {
3888
+ "epoch": 2.65,
3889
+ "learning_rate": 3.4525904494220106e-06,
3890
+ "loss": 1.7907,
3891
+ "step": 3030
3892
+ },
3893
+ {
3894
+ "epoch": 2.66,
3895
+ "learning_rate": 3.3662902714521714e-06,
3896
+ "loss": 1.7855,
3897
+ "step": 3035
3898
+ },
3899
+ {
3900
+ "epoch": 2.66,
3901
+ "learning_rate": 3.281044776014652e-06,
3902
+ "loss": 1.7954,
3903
+ "step": 3040
3904
+ },
3905
+ {
3906
+ "epoch": 2.67,
3907
+ "learning_rate": 3.196855891048195e-06,
3908
+ "loss": 1.8015,
3909
+ "step": 3045
3910
+ },
3911
+ {
3912
+ "epoch": 2.67,
3913
+ "learning_rate": 3.113725520594879e-06,
3914
+ "loss": 1.7971,
3915
+ "step": 3050
3916
+ },
3917
+ {
3918
+ "epoch": 2.68,
3919
+ "learning_rate": 3.0316555447571282e-06,
3920
+ "loss": 1.8061,
3921
+ "step": 3055
3922
+ },
3923
+ {
3924
+ "epoch": 2.68,
3925
+ "learning_rate": 2.9506478196551055e-06,
3926
+ "loss": 1.7917,
3927
+ "step": 3060
3928
+ },
3929
+ {
3930
+ "epoch": 2.69,
3931
+ "learning_rate": 2.87070417738482e-06,
3932
+ "loss": 1.8035,
3933
+ "step": 3065
3934
+ },
3935
+ {
3936
+ "epoch": 2.69,
3937
+ "learning_rate": 2.7918264259766126e-06,
3938
+ "loss": 1.7921,
3939
+ "step": 3070
3940
+ },
3941
+ {
3942
+ "epoch": 2.69,
3943
+ "learning_rate": 2.714016349354348e-06,
3944
+ "loss": 1.7931,
3945
+ "step": 3075
3946
+ },
3947
+ {
3948
+ "epoch": 2.7,
3949
+ "learning_rate": 2.6372757072949896e-06,
3950
+ "loss": 1.7951,
3951
+ "step": 3080
3952
+ },
3953
+ {
3954
+ "epoch": 2.7,
3955
+ "learning_rate": 2.5616062353888415e-06,
3956
+ "loss": 1.8127,
3957
+ "step": 3085
3958
+ },
3959
+ {
3960
+ "epoch": 2.71,
3961
+ "learning_rate": 2.48700964500031e-06,
3962
+ "loss": 1.7914,
3963
+ "step": 3090
3964
+ },
3965
+ {
3966
+ "epoch": 2.71,
3967
+ "learning_rate": 2.413487623229149e-06,
3968
+ "loss": 1.797,
3969
+ "step": 3095
3970
+ },
3971
+ {
3972
+ "epoch": 2.72,
3973
+ "learning_rate": 2.341041832872348e-06,
3974
+ "loss": 1.7953,
3975
+ "step": 3100
3976
+ },
3977
+ {
3978
+ "epoch": 2.72,
3979
+ "eval_loss": 2.1286771297454834,
3980
+ "eval_runtime": 435.6868,
3981
+ "eval_samples_per_second": 37.105,
3982
+ "eval_steps_per_second": 1.161,
3983
+ "step": 3100
3984
+ },
3985
+ {
3986
+ "epoch": 2.72,
3987
+ "learning_rate": 2.2696739123865218e-06,
3988
+ "loss": 1.8038,
3989
+ "step": 3105
3990
+ },
3991
+ {
3992
+ "epoch": 2.72,
3993
+ "learning_rate": 2.199385475850824e-06,
3994
+ "loss": 1.7959,
3995
+ "step": 3110
3996
+ },
3997
+ {
3998
+ "epoch": 2.73,
3999
+ "learning_rate": 2.1301781129304587e-06,
4000
+ "loss": 1.8003,
4001
+ "step": 3115
4002
+ },
4003
+ {
4004
+ "epoch": 2.73,
4005
+ "learning_rate": 2.062053388840768e-06,
4006
+ "loss": 1.7858,
4007
+ "step": 3120
4008
+ },
4009
+ {
4010
+ "epoch": 2.74,
4011
+ "learning_rate": 1.99501284431175e-06,
4012
+ "loss": 1.7969,
4013
+ "step": 3125
4014
+ },
4015
+ {
4016
+ "epoch": 2.74,
4017
+ "learning_rate": 1.929057995553296e-06,
4018
+ "loss": 1.8083,
4019
+ "step": 3130
4020
+ },
4021
+ {
4022
+ "epoch": 2.75,
4023
+ "learning_rate": 1.8641903342208388e-06,
4024
+ "loss": 1.8103,
4025
+ "step": 3135
4026
+ },
4027
+ {
4028
+ "epoch": 2.75,
4029
+ "learning_rate": 1.800411327381657e-06,
4030
+ "loss": 1.7886,
4031
+ "step": 3140
4032
+ },
4033
+ {
4034
+ "epoch": 2.76,
4035
+ "learning_rate": 1.7377224174816687e-06,
4036
+ "loss": 1.7859,
4037
+ "step": 3145
4038
+ },
4039
+ {
4040
+ "epoch": 2.76,
4041
+ "learning_rate": 1.6761250223128299e-06,
4042
+ "loss": 1.8073,
4043
+ "step": 3150
4044
+ },
4045
+ {
4046
+ "epoch": 2.76,
4047
+ "learning_rate": 1.6156205349810494e-06,
4048
+ "loss": 1.8025,
4049
+ "step": 3155
4050
+ },
4051
+ {
4052
+ "epoch": 2.77,
4053
+ "learning_rate": 1.556210323874685e-06,
4054
+ "loss": 1.7855,
4055
+ "step": 3160
4056
+ },
4057
+ {
4058
+ "epoch": 2.77,
4059
+ "learning_rate": 1.4978957326336252e-06,
4060
+ "loss": 1.7863,
4061
+ "step": 3165
4062
+ },
4063
+ {
4064
+ "epoch": 2.78,
4065
+ "learning_rate": 1.4406780801188514e-06,
4066
+ "loss": 1.7958,
4067
+ "step": 3170
4068
+ },
4069
+ {
4070
+ "epoch": 2.78,
4071
+ "learning_rate": 1.3845586603826677e-06,
4072
+ "loss": 1.7935,
4073
+ "step": 3175
4074
+ },
4075
+ {
4076
+ "epoch": 2.79,
4077
+ "learning_rate": 1.329538742639358e-06,
4078
+ "loss": 1.7953,
4079
+ "step": 3180
4080
+ },
4081
+ {
4082
+ "epoch": 2.79,
4083
+ "learning_rate": 1.2756195712365804e-06,
4084
+ "loss": 1.791,
4085
+ "step": 3185
4086
+ },
4087
+ {
4088
+ "epoch": 2.79,
4089
+ "learning_rate": 1.2228023656271182e-06,
4090
+ "loss": 1.8091,
4091
+ "step": 3190
4092
+ },
4093
+ {
4094
+ "epoch": 2.8,
4095
+ "learning_rate": 1.1710883203413904e-06,
4096
+ "loss": 1.8014,
4097
+ "step": 3195
4098
+ },
4099
+ {
4100
+ "epoch": 2.8,
4101
+ "learning_rate": 1.1204786049603789e-06,
4102
+ "loss": 1.798,
4103
+ "step": 3200
4104
+ },
4105
+ {
4106
+ "epoch": 2.8,
4107
+ "eval_loss": 2.127976179122925,
4108
+ "eval_runtime": 437.1616,
4109
+ "eval_samples_per_second": 36.979,
4110
+ "eval_steps_per_second": 1.157,
4111
+ "step": 3200
4112
+ },
4113
+ {
4114
+ "epoch": 2.81,
4115
+ "learning_rate": 1.0709743640892e-06,
4116
+ "loss": 1.7822,
4117
+ "step": 3205
4118
+ },
4119
+ {
4120
+ "epoch": 2.81,
4121
+ "learning_rate": 1.022576717331214e-06,
4122
+ "loss": 1.7974,
4123
+ "step": 3210
4124
+ },
4125
+ {
4126
+ "epoch": 2.82,
4127
+ "learning_rate": 9.752867592627112e-07,
4128
+ "loss": 1.797,
4129
+ "step": 3215
4130
+ },
4131
+ {
4132
+ "epoch": 2.82,
4133
+ "learning_rate": 9.291055594081278e-07,
4134
+ "loss": 1.8001,
4135
+ "step": 3220
4136
+ },
4137
+ {
4138
+ "epoch": 2.83,
4139
+ "learning_rate": 8.840341622159076e-07,
4140
+ "loss": 1.7971,
4141
+ "step": 3225
4142
+ },
4143
+ {
4144
+ "epoch": 2.83,
4145
+ "learning_rate": 8.400735870348231e-07,
4146
+ "loss": 1.8057,
4147
+ "step": 3230
4148
+ },
4149
+ {
4150
+ "epoch": 2.83,
4151
+ "learning_rate": 7.972248280909589e-07,
4152
+ "loss": 1.7854,
4153
+ "step": 3235
4154
+ },
4155
+ {
4156
+ "epoch": 2.84,
4157
+ "learning_rate": 7.554888544652305e-07,
4158
+ "loss": 1.8117,
4159
+ "step": 3240
4160
+ },
4161
+ {
4162
+ "epoch": 2.84,
4163
+ "learning_rate": 7.148666100714352e-07,
4164
+ "loss": 1.779,
4165
+ "step": 3245
4166
+ },
4167
+ {
4168
+ "epoch": 2.85,
4169
+ "learning_rate": 6.753590136349298e-07,
4170
+ "loss": 1.8081,
4171
+ "step": 3250
4172
+ },
4173
+ {
4174
+ "epoch": 2.85,
4175
+ "learning_rate": 6.369669586718419e-07,
4176
+ "loss": 1.7938,
4177
+ "step": 3255
4178
+ },
4179
+ {
4180
+ "epoch": 2.86,
4181
+ "learning_rate": 5.996913134688753e-07,
4182
+ "loss": 1.8021,
4183
+ "step": 3260
4184
+ },
4185
+ {
4186
+ "epoch": 2.86,
4187
+ "learning_rate": 5.635329210636531e-07,
4188
+ "loss": 1.7888,
4189
+ "step": 3265
4190
+ },
4191
+ {
4192
+ "epoch": 2.86,
4193
+ "learning_rate": 5.284925992256551e-07,
4194
+ "loss": 1.7845,
4195
+ "step": 3270
4196
+ },
4197
+ {
4198
+ "epoch": 2.87,
4199
+ "learning_rate": 4.945711404377385e-07,
4200
+ "loss": 1.8015,
4201
+ "step": 3275
4202
+ },
4203
+ {
4204
+ "epoch": 2.87,
4205
+ "learning_rate": 4.617693118782074e-07,
4206
+ "loss": 1.7979,
4207
+ "step": 3280
4208
+ },
4209
+ {
4210
+ "epoch": 2.88,
4211
+ "learning_rate": 4.3008785540344376e-07,
4212
+ "loss": 1.7888,
4213
+ "step": 3285
4214
+ },
4215
+ {
4216
+ "epoch": 2.88,
4217
+ "learning_rate": 3.9952748753115945e-07,
4218
+ "loss": 1.7901,
4219
+ "step": 3290
4220
+ },
4221
+ {
4222
+ "epoch": 2.89,
4223
+ "learning_rate": 3.7008889942417025e-07,
4224
+ "loss": 1.7981,
4225
+ "step": 3295
4226
+ },
4227
+ {
4228
+ "epoch": 2.89,
4229
+ "learning_rate": 3.4177275687476974e-07,
4230
+ "loss": 1.7947,
4231
+ "step": 3300
4232
+ },
4233
+ {
4234
+ "epoch": 2.89,
4235
+ "eval_loss": 2.1282496452331543,
4236
+ "eval_runtime": 437.2091,
4237
+ "eval_samples_per_second": 36.975,
4238
+ "eval_steps_per_second": 1.157,
4239
+ "step": 3300
4240
+ },
4241
+ {
4242
+ "epoch": 2.9,
4243
+ "learning_rate": 3.1457970028968e-07,
4244
+ "loss": 1.7772,
4245
+ "step": 3305
4246
+ },
4247
+ {
4248
+ "epoch": 2.9,
4249
+ "learning_rate": 2.885103446755466e-07,
4250
+ "loss": 1.7805,
4251
+ "step": 3310
4252
+ },
4253
+ {
4254
+ "epoch": 2.9,
4255
+ "learning_rate": 2.635652796250554e-07,
4256
+ "loss": 1.7927,
4257
+ "step": 3315
4258
+ },
4259
+ {
4260
+ "epoch": 2.91,
4261
+ "learning_rate": 2.3974506930358743e-07,
4262
+ "loss": 1.7941,
4263
+ "step": 3320
4264
+ },
4265
+ {
4266
+ "epoch": 2.91,
4267
+ "learning_rate": 2.1705025243644594e-07,
4268
+ "loss": 1.7991,
4269
+ "step": 3325
4270
+ },
4271
+ {
4272
+ "epoch": 2.92,
4273
+ "learning_rate": 1.9548134229669947e-07,
4274
+ "loss": 1.7973,
4275
+ "step": 3330
4276
+ },
4277
+ {
4278
+ "epoch": 2.92,
4279
+ "learning_rate": 1.7503882669354098e-07,
4280
+ "loss": 1.802,
4281
+ "step": 3335
4282
+ },
4283
+ {
4284
+ "epoch": 2.93,
4285
+ "learning_rate": 1.5572316796128583e-07,
4286
+ "loss": 1.7945,
4287
+ "step": 3340
4288
+ },
4289
+ {
4290
+ "epoch": 2.93,
4291
+ "learning_rate": 1.3753480294890763e-07,
4292
+ "loss": 1.7975,
4293
+ "step": 3345
4294
+ },
4295
+ {
4296
+ "epoch": 2.93,
4297
+ "learning_rate": 1.2047414301013527e-07,
4298
+ "loss": 1.7829,
4299
+ "step": 3350
4300
+ },
4301
+ {
4302
+ "epoch": 2.94,
4303
+ "learning_rate": 1.045415739941824e-07,
4304
+ "loss": 1.8042,
4305
+ "step": 3355
4306
+ },
4307
+ {
4308
+ "epoch": 2.94,
4309
+ "learning_rate": 8.973745623699903e-08,
4310
+ "loss": 1.8032,
4311
+ "step": 3360
4312
+ },
4313
+ {
4314
+ "epoch": 2.95,
4315
+ "learning_rate": 7.6062124553139e-08,
4316
+ "loss": 1.8013,
4317
+ "step": 3365
4318
+ },
4319
+ {
4320
+ "epoch": 2.95,
4321
+ "learning_rate": 6.351588822817167e-08,
4322
+ "loss": 1.7983,
4323
+ "step": 3370
4324
+ },
4325
+ {
4326
+ "epoch": 2.96,
4327
+ "learning_rate": 5.209903101168201e-08,
4328
+ "loss": 1.7967,
4329
+ "step": 3375
4330
+ },
4331
+ {
4332
+ "epoch": 2.96,
4333
+ "learning_rate": 4.1811811110886724e-08,
4334
+ "loss": 1.7997,
4335
+ "step": 3380
4336
+ },
4337
+ {
4338
+ "epoch": 2.97,
4339
+ "learning_rate": 3.265446118475568e-08,
4340
+ "loss": 1.7907,
4341
+ "step": 3385
4342
+ },
4343
+ {
4344
+ "epoch": 2.97,
4345
+ "learning_rate": 2.4627188338777196e-08,
4346
+ "loss": 1.7963,
4347
+ "step": 3390
4348
+ },
4349
+ {
4350
+ "epoch": 2.97,
4351
+ "learning_rate": 1.7730174120250687e-08,
4352
+ "loss": 1.7791,
4353
+ "step": 3395
4354
+ },
4355
+ {
4356
+ "epoch": 2.98,
4357
+ "learning_rate": 1.196357451421215e-08,
4358
+ "loss": 1.8015,
4359
+ "step": 3400
4360
+ },
4361
+ {
4362
+ "epoch": 2.98,
4363
+ "eval_loss": 2.1283345222473145,
4364
+ "eval_runtime": 436.7591,
4365
+ "eval_samples_per_second": 37.014,
4366
+ "eval_steps_per_second": 1.159,
4367
+ "step": 3400
4368
+ },
4369
+ {
4370
+ "epoch": 2.98,
4371
+ "learning_rate": 7.327519939870353e-09,
4372
+ "loss": 1.7954,
4373
+ "step": 3405
4374
+ },
4375
+ {
4376
+ "epoch": 2.99,
4377
+ "learning_rate": 3.822115247686941e-09,
4378
+ "loss": 1.79,
4379
+ "step": 3410
4380
+ },
4381
+ {
4382
+ "epoch": 2.99,
4383
+ "learning_rate": 1.4474397169894626e-09,
4384
+ "loss": 1.79,
4385
+ "step": 3415
4386
+ },
4387
+ {
4388
+ "epoch": 3.0,
4389
+ "learning_rate": 2.0354705417280351e-10,
4390
+ "loss": 1.7978,
4391
+ "step": 3420
4392
+ },
4393
+ {
4394
+ "epoch": 3.0,
4395
+ "step": 3423,
4396
+ "total_flos": 5.633765648992567e+18,
4397
+ "train_loss": 1.9906484538130407,
4398
+ "train_runtime": 48039.6785,
4399
+ "train_samples_per_second": 9.121,
4400
+ "train_steps_per_second": 0.071
4401
+ }
4402
+ ],
4403
+ "logging_steps": 5,
4404
+ "max_steps": 3423,
4405
+ "num_input_tokens_seen": 0,
4406
+ "num_train_epochs": 3,
4407
+ "save_steps": 100,
4408
+ "total_flos": 5.633765648992567e+18,
4409
+ "train_batch_size": 16,
4410
+ "trial_name": null,
4411
+ "trial_params": null
4412
+ }