tyzhu commited on
Commit
7264648
1 Parent(s): 0ef170d

Training in progress, epoch 48, checkpoint

Browse files
checkpoint-16392/README.md ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+ ### Framework versions
7
+
8
+
9
+ - PEFT 0.5.0
checkpoint-16392/adapter_config.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "Qwen/Qwen1.5-4B",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 16,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "k_proj",
19
+ "v_proj",
20
+ "down_proj",
21
+ "gate_proj",
22
+ "up_proj"
23
+ ],
24
+ "task_type": "CAUSAL_LM"
25
+ }
checkpoint-16392/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26dfa0ca347af297a5385927c98c7f5be29d4aaab4b6a7fc4781707c021966a8
3
+ size 112130416
checkpoint-16392/added_tokens.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "<|endoftext|>": 151643,
3
+ "<|im_end|>": 151645,
4
+ "<|im_start|>": 151644
5
+ }
checkpoint-16392/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-16392/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f6b062e63353441027410942451b97f888437c27ae191c64fc44b24722ddc13
3
+ size 224537202
checkpoint-16392/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5bce6f4b5ef2826dced57e893f03a6c9badbd0b40aba87c23eefdc24fcd4c7b
3
+ size 15024
checkpoint-16392/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5c9c28eaa5b303dedeeac19bc898e7a95828b1cfd1bd3d547961bb4d8b35be5e
3
+ size 15024
checkpoint-16392/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac1df24a28b177520161f2d895e4a53023444e8658f4b4fb7f82b7a8a52011b3
3
+ size 15024
checkpoint-16392/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ccb77e891bd8293030748eb101cfba8a5bcbbdbceb319731dfc30bb963ceeb0
3
+ size 15024
checkpoint-16392/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f5049af579892b92ece1454d006b1b86fdaedb4ce5c01cba36db1fc687aefda
3
+ size 1064
checkpoint-16392/special_tokens_map.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>"
5
+ ],
6
+ "eos_token": {
7
+ "content": "<|endoftext|>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false
12
+ },
13
+ "pad_token": "<|endoftext|>"
14
+ }
checkpoint-16392/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-16392/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "151643": {
5
+ "content": "<|endoftext|>",
6
+ "lstrip": false,
7
+ "normalized": false,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "151644": {
13
+ "content": "<|im_start|>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ },
20
+ "151645": {
21
+ "content": "<|im_end|>",
22
+ "lstrip": false,
23
+ "normalized": false,
24
+ "rstrip": false,
25
+ "single_word": false,
26
+ "special": true
27
+ }
28
+ },
29
+ "additional_special_tokens": [
30
+ "<|im_start|>",
31
+ "<|im_end|>"
32
+ ],
33
+ "bos_token": null,
34
+ "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
35
+ "clean_up_tokenization_spaces": false,
36
+ "eos_token": "<|endoftext|>",
37
+ "errors": "replace",
38
+ "model_max_length": 32768,
39
+ "pad_token": "<|endoftext|>",
40
+ "split_special_tokens": false,
41
+ "tokenizer_class": "Qwen2Tokenizer",
42
+ "unk_token": null
43
+ }
checkpoint-16392/trainer_state.json ADDED
@@ -0,0 +1,1942 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 48.0,
5
+ "eval_steps": 500,
6
+ "global_step": 16392,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.29282576866764276,
13
+ "grad_norm": 0.26115021109580994,
14
+ "learning_rate": 3e-05,
15
+ "loss": 2.0719,
16
+ "step": 100
17
+ },
18
+ {
19
+ "epoch": 0.5856515373352855,
20
+ "grad_norm": 0.2679530680179596,
21
+ "learning_rate": 3e-05,
22
+ "loss": 1.969,
23
+ "step": 200
24
+ },
25
+ {
26
+ "epoch": 0.8784773060029283,
27
+ "grad_norm": 0.2916224002838135,
28
+ "learning_rate": 3e-05,
29
+ "loss": 1.9686,
30
+ "step": 300
31
+ },
32
+ {
33
+ "epoch": 0.9985358711566618,
34
+ "eval_accuracy": 0.43743589743589745,
35
+ "eval_loss": 3.6918885707855225,
36
+ "eval_runtime": 5.2974,
37
+ "eval_samples_per_second": 94.385,
38
+ "eval_steps_per_second": 11.893,
39
+ "step": 341
40
+ },
41
+ {
42
+ "epoch": 0.9985358711566618,
43
+ "eval_bleu": 0.1527908371131025,
44
+ "eval_exact_match": 0.012,
45
+ "eval_prefix_exact_match": 0.026,
46
+ "step": 341
47
+ },
48
+ {
49
+ "epoch": 1.171303074670571,
50
+ "grad_norm": 0.3534347712993622,
51
+ "learning_rate": 3e-05,
52
+ "loss": 1.9603,
53
+ "step": 400
54
+ },
55
+ {
56
+ "epoch": 1.4641288433382138,
57
+ "grad_norm": 0.3338424563407898,
58
+ "learning_rate": 3e-05,
59
+ "loss": 1.9235,
60
+ "step": 500
61
+ },
62
+ {
63
+ "epoch": 1.7569546120058566,
64
+ "grad_norm": 0.4033390283584595,
65
+ "learning_rate": 3e-05,
66
+ "loss": 1.9337,
67
+ "step": 600
68
+ },
69
+ {
70
+ "epoch": 2.0,
71
+ "eval_accuracy": 0.44764102564102565,
72
+ "eval_loss": 3.7484843730926514,
73
+ "eval_runtime": 5.0649,
74
+ "eval_samples_per_second": 98.718,
75
+ "eval_steps_per_second": 12.438,
76
+ "step": 683
77
+ },
78
+ {
79
+ "epoch": 2.0,
80
+ "eval_bleu": 0.21311912714932216,
81
+ "eval_exact_match": 0.02,
82
+ "eval_prefix_exact_match": 0.032,
83
+ "step": 683
84
+ },
85
+ {
86
+ "epoch": 2.049780380673499,
87
+ "grad_norm": 0.38882532715797424,
88
+ "learning_rate": 3e-05,
89
+ "loss": 1.9267,
90
+ "step": 700
91
+ },
92
+ {
93
+ "epoch": 2.342606149341142,
94
+ "grad_norm": 0.44310709834098816,
95
+ "learning_rate": 3e-05,
96
+ "loss": 1.8823,
97
+ "step": 800
98
+ },
99
+ {
100
+ "epoch": 2.6354319180087846,
101
+ "grad_norm": 0.5196680426597595,
102
+ "learning_rate": 3e-05,
103
+ "loss": 1.901,
104
+ "step": 900
105
+ },
106
+ {
107
+ "epoch": 2.9282576866764276,
108
+ "grad_norm": 0.5293437242507935,
109
+ "learning_rate": 3e-05,
110
+ "loss": 1.9033,
111
+ "step": 1000
112
+ },
113
+ {
114
+ "epoch": 2.998535871156662,
115
+ "eval_accuracy": 0.4495897435897436,
116
+ "eval_loss": 3.882551670074463,
117
+ "eval_runtime": 5.1314,
118
+ "eval_samples_per_second": 97.439,
119
+ "eval_steps_per_second": 12.277,
120
+ "step": 1024
121
+ },
122
+ {
123
+ "epoch": 2.998535871156662,
124
+ "eval_bleu": 0.24307488023362148,
125
+ "eval_exact_match": 0.016,
126
+ "eval_prefix_exact_match": 0.032,
127
+ "step": 1024
128
+ },
129
+ {
130
+ "epoch": 3.22108345534407,
131
+ "grad_norm": 0.552943229675293,
132
+ "learning_rate": 3e-05,
133
+ "loss": 1.8529,
134
+ "step": 1100
135
+ },
136
+ {
137
+ "epoch": 3.513909224011713,
138
+ "grad_norm": 0.6421984434127808,
139
+ "learning_rate": 3e-05,
140
+ "loss": 1.8342,
141
+ "step": 1200
142
+ },
143
+ {
144
+ "epoch": 3.8067349926793557,
145
+ "grad_norm": 0.7155198454856873,
146
+ "learning_rate": 3e-05,
147
+ "loss": 1.857,
148
+ "step": 1300
149
+ },
150
+ {
151
+ "epoch": 4.0,
152
+ "eval_accuracy": 0.4481025641025641,
153
+ "eval_loss": 3.970107316970825,
154
+ "eval_runtime": 5.1563,
155
+ "eval_samples_per_second": 96.969,
156
+ "eval_steps_per_second": 12.218,
157
+ "step": 1366
158
+ },
159
+ {
160
+ "epoch": 4.0,
161
+ "eval_bleu": 0.25309406645138827,
162
+ "eval_exact_match": 0.016,
163
+ "eval_prefix_exact_match": 0.044,
164
+ "step": 1366
165
+ },
166
+ {
167
+ "epoch": 4.099560761346998,
168
+ "grad_norm": 0.7837833166122437,
169
+ "learning_rate": 3e-05,
170
+ "loss": 1.8284,
171
+ "step": 1400
172
+ },
173
+ {
174
+ "epoch": 4.392386530014641,
175
+ "grad_norm": 0.784609854221344,
176
+ "learning_rate": 3e-05,
177
+ "loss": 1.7884,
178
+ "step": 1500
179
+ },
180
+ {
181
+ "epoch": 4.685212298682284,
182
+ "grad_norm": 0.834195613861084,
183
+ "learning_rate": 3e-05,
184
+ "loss": 1.7841,
185
+ "step": 1600
186
+ },
187
+ {
188
+ "epoch": 4.978038067349927,
189
+ "grad_norm": 0.8902062177658081,
190
+ "learning_rate": 3e-05,
191
+ "loss": 1.8042,
192
+ "step": 1700
193
+ },
194
+ {
195
+ "epoch": 4.998535871156662,
196
+ "eval_accuracy": 0.4472820512820513,
197
+ "eval_loss": 4.117065906524658,
198
+ "eval_runtime": 5.2376,
199
+ "eval_samples_per_second": 95.464,
200
+ "eval_steps_per_second": 12.028,
201
+ "step": 1707
202
+ },
203
+ {
204
+ "epoch": 4.998535871156662,
205
+ "eval_bleu": 0.2531049955279983,
206
+ "eval_exact_match": 0.014,
207
+ "eval_prefix_exact_match": 0.042,
208
+ "step": 1707
209
+ },
210
+ {
211
+ "epoch": 5.270863836017569,
212
+ "grad_norm": 0.9495383501052856,
213
+ "learning_rate": 3e-05,
214
+ "loss": 1.7493,
215
+ "step": 1800
216
+ },
217
+ {
218
+ "epoch": 5.563689604685212,
219
+ "grad_norm": 0.9647482633590698,
220
+ "learning_rate": 3e-05,
221
+ "loss": 1.7334,
222
+ "step": 1900
223
+ },
224
+ {
225
+ "epoch": 5.856515373352855,
226
+ "grad_norm": 0.9825767874717712,
227
+ "learning_rate": 3e-05,
228
+ "loss": 1.7443,
229
+ "step": 2000
230
+ },
231
+ {
232
+ "epoch": 6.0,
233
+ "eval_accuracy": 0.446974358974359,
234
+ "eval_loss": 4.183700084686279,
235
+ "eval_runtime": 5.2674,
236
+ "eval_samples_per_second": 94.923,
237
+ "eval_steps_per_second": 11.96,
238
+ "step": 2049
239
+ },
240
+ {
241
+ "epoch": 6.0,
242
+ "eval_bleu": 0.23587359955369594,
243
+ "eval_exact_match": 0.012,
244
+ "eval_prefix_exact_match": 0.04,
245
+ "step": 2049
246
+ },
247
+ {
248
+ "epoch": 6.149341142020498,
249
+ "grad_norm": 1.0430296659469604,
250
+ "learning_rate": 3e-05,
251
+ "loss": 1.6959,
252
+ "step": 2100
253
+ },
254
+ {
255
+ "epoch": 6.44216691068814,
256
+ "grad_norm": 1.0777732133865356,
257
+ "learning_rate": 3e-05,
258
+ "loss": 1.6795,
259
+ "step": 2200
260
+ },
261
+ {
262
+ "epoch": 6.734992679355783,
263
+ "grad_norm": 1.0959677696228027,
264
+ "learning_rate": 3e-05,
265
+ "loss": 1.7019,
266
+ "step": 2300
267
+ },
268
+ {
269
+ "epoch": 6.998535871156662,
270
+ "eval_accuracy": 0.4461538461538462,
271
+ "eval_loss": 4.260408401489258,
272
+ "eval_runtime": 5.178,
273
+ "eval_samples_per_second": 96.563,
274
+ "eval_steps_per_second": 12.167,
275
+ "step": 2390
276
+ },
277
+ {
278
+ "epoch": 6.998535871156662,
279
+ "eval_bleu": 0.2008158466316808,
280
+ "eval_exact_match": 0.006,
281
+ "eval_prefix_exact_match": 0.042,
282
+ "step": 2390
283
+ },
284
+ {
285
+ "epoch": 7.027818448023426,
286
+ "grad_norm": 1.0649549961090088,
287
+ "learning_rate": 3e-05,
288
+ "loss": 1.6812,
289
+ "step": 2400
290
+ },
291
+ {
292
+ "epoch": 7.320644216691068,
293
+ "grad_norm": 1.294122576713562,
294
+ "learning_rate": 3e-05,
295
+ "loss": 1.6226,
296
+ "step": 2500
297
+ },
298
+ {
299
+ "epoch": 7.613469985358711,
300
+ "grad_norm": 1.2657020092010498,
301
+ "learning_rate": 3e-05,
302
+ "loss": 1.6404,
303
+ "step": 2600
304
+ },
305
+ {
306
+ "epoch": 7.906295754026354,
307
+ "grad_norm": 1.1963133811950684,
308
+ "learning_rate": 3e-05,
309
+ "loss": 1.6305,
310
+ "step": 2700
311
+ },
312
+ {
313
+ "epoch": 8.0,
314
+ "eval_accuracy": 0.44148717948717947,
315
+ "eval_loss": 4.406544208526611,
316
+ "eval_runtime": 5.3355,
317
+ "eval_samples_per_second": 93.711,
318
+ "eval_steps_per_second": 11.808,
319
+ "step": 2732
320
+ },
321
+ {
322
+ "epoch": 8.0,
323
+ "eval_bleu": 0.24435500637461924,
324
+ "eval_exact_match": 0.008,
325
+ "eval_prefix_exact_match": 0.038,
326
+ "step": 2732
327
+ },
328
+ {
329
+ "epoch": 8.199121522693996,
330
+ "grad_norm": 1.3801360130310059,
331
+ "learning_rate": 3e-05,
332
+ "loss": 1.5734,
333
+ "step": 2800
334
+ },
335
+ {
336
+ "epoch": 8.49194729136164,
337
+ "grad_norm": 1.3306772708892822,
338
+ "learning_rate": 3e-05,
339
+ "loss": 1.5651,
340
+ "step": 2900
341
+ },
342
+ {
343
+ "epoch": 8.784773060029282,
344
+ "grad_norm": 1.5165746212005615,
345
+ "learning_rate": 3e-05,
346
+ "loss": 1.6056,
347
+ "step": 3000
348
+ },
349
+ {
350
+ "epoch": 8.998535871156662,
351
+ "eval_accuracy": 0.4397948717948718,
352
+ "eval_loss": 4.448728561401367,
353
+ "eval_runtime": 5.187,
354
+ "eval_samples_per_second": 96.395,
355
+ "eval_steps_per_second": 12.146,
356
+ "step": 3073
357
+ },
358
+ {
359
+ "epoch": 8.998535871156662,
360
+ "eval_bleu": 0.20120283162176952,
361
+ "eval_exact_match": 0.008,
362
+ "eval_prefix_exact_match": 0.036,
363
+ "step": 3073
364
+ },
365
+ {
366
+ "epoch": 9.077598828696924,
367
+ "grad_norm": 1.4077060222625732,
368
+ "learning_rate": 3e-05,
369
+ "loss": 1.5612,
370
+ "step": 3100
371
+ },
372
+ {
373
+ "epoch": 9.370424597364568,
374
+ "grad_norm": 1.475984811782837,
375
+ "learning_rate": 3e-05,
376
+ "loss": 1.5158,
377
+ "step": 3200
378
+ },
379
+ {
380
+ "epoch": 9.66325036603221,
381
+ "grad_norm": 1.586300015449524,
382
+ "learning_rate": 3e-05,
383
+ "loss": 1.5145,
384
+ "step": 3300
385
+ },
386
+ {
387
+ "epoch": 9.956076134699854,
388
+ "grad_norm": 1.510048270225525,
389
+ "learning_rate": 3e-05,
390
+ "loss": 1.5521,
391
+ "step": 3400
392
+ },
393
+ {
394
+ "epoch": 10.0,
395
+ "eval_accuracy": 0.4388717948717949,
396
+ "eval_loss": 4.5474138259887695,
397
+ "eval_runtime": 5.169,
398
+ "eval_samples_per_second": 96.731,
399
+ "eval_steps_per_second": 12.188,
400
+ "step": 3415
401
+ },
402
+ {
403
+ "epoch": 10.0,
404
+ "eval_bleu": 0.2258317707850195,
405
+ "eval_exact_match": 0.004,
406
+ "eval_prefix_exact_match": 0.032,
407
+ "step": 3415
408
+ },
409
+ {
410
+ "epoch": 10.248901903367496,
411
+ "grad_norm": 1.6384536027908325,
412
+ "learning_rate": 3e-05,
413
+ "loss": 1.4631,
414
+ "step": 3500
415
+ },
416
+ {
417
+ "epoch": 10.541727672035138,
418
+ "grad_norm": 1.7626007795333862,
419
+ "learning_rate": 3e-05,
420
+ "loss": 1.4623,
421
+ "step": 3600
422
+ },
423
+ {
424
+ "epoch": 10.834553440702782,
425
+ "grad_norm": 1.6483694314956665,
426
+ "learning_rate": 3e-05,
427
+ "loss": 1.4934,
428
+ "step": 3700
429
+ },
430
+ {
431
+ "epoch": 10.998535871156662,
432
+ "eval_accuracy": 0.4367179487179487,
433
+ "eval_loss": 4.589761734008789,
434
+ "eval_runtime": 5.853,
435
+ "eval_samples_per_second": 85.426,
436
+ "eval_steps_per_second": 10.764,
437
+ "step": 3756
438
+ },
439
+ {
440
+ "epoch": 10.998535871156662,
441
+ "eval_bleu": 0.20792247275869374,
442
+ "eval_exact_match": 0.004,
443
+ "eval_prefix_exact_match": 0.026,
444
+ "step": 3756
445
+ },
446
+ {
447
+ "epoch": 11.127379209370424,
448
+ "grad_norm": 1.8367844820022583,
449
+ "learning_rate": 3e-05,
450
+ "loss": 1.4521,
451
+ "step": 3800
452
+ },
453
+ {
454
+ "epoch": 11.420204978038067,
455
+ "grad_norm": 1.8123779296875,
456
+ "learning_rate": 3e-05,
457
+ "loss": 1.4085,
458
+ "step": 3900
459
+ },
460
+ {
461
+ "epoch": 11.71303074670571,
462
+ "grad_norm": 1.9437752962112427,
463
+ "learning_rate": 3e-05,
464
+ "loss": 1.4287,
465
+ "step": 4000
466
+ },
467
+ {
468
+ "epoch": 12.0,
469
+ "eval_accuracy": 0.43553846153846154,
470
+ "eval_loss": 4.691086769104004,
471
+ "eval_runtime": 5.3738,
472
+ "eval_samples_per_second": 93.043,
473
+ "eval_steps_per_second": 11.723,
474
+ "step": 4098
475
+ },
476
+ {
477
+ "epoch": 12.0,
478
+ "eval_bleu": 0.21672573731424186,
479
+ "eval_exact_match": 0.006,
480
+ "eval_prefix_exact_match": 0.028,
481
+ "step": 4098
482
+ },
483
+ {
484
+ "epoch": 12.005856515373353,
485
+ "grad_norm": 1.5838289260864258,
486
+ "learning_rate": 3e-05,
487
+ "loss": 1.4361,
488
+ "step": 4100
489
+ },
490
+ {
491
+ "epoch": 12.298682284040996,
492
+ "grad_norm": 1.8575290441513062,
493
+ "learning_rate": 3e-05,
494
+ "loss": 1.3433,
495
+ "step": 4200
496
+ },
497
+ {
498
+ "epoch": 12.591508052708638,
499
+ "grad_norm": 1.9598041772842407,
500
+ "learning_rate": 3e-05,
501
+ "loss": 1.3661,
502
+ "step": 4300
503
+ },
504
+ {
505
+ "epoch": 12.88433382137628,
506
+ "grad_norm": 1.972550868988037,
507
+ "learning_rate": 3e-05,
508
+ "loss": 1.3846,
509
+ "step": 4400
510
+ },
511
+ {
512
+ "epoch": 12.998535871156662,
513
+ "eval_accuracy": 0.43553846153846154,
514
+ "eval_loss": 4.762892246246338,
515
+ "eval_runtime": 5.106,
516
+ "eval_samples_per_second": 97.925,
517
+ "eval_steps_per_second": 12.339,
518
+ "step": 4439
519
+ },
520
+ {
521
+ "epoch": 12.998535871156662,
522
+ "eval_bleu": 0.2416046504676966,
523
+ "eval_exact_match": 0.006,
524
+ "eval_prefix_exact_match": 0.028,
525
+ "step": 4439
526
+ },
527
+ {
528
+ "epoch": 13.177159590043924,
529
+ "grad_norm": 1.9469107389450073,
530
+ "learning_rate": 3e-05,
531
+ "loss": 1.3229,
532
+ "step": 4500
533
+ },
534
+ {
535
+ "epoch": 13.469985358711567,
536
+ "grad_norm": 2.1680469512939453,
537
+ "learning_rate": 3e-05,
538
+ "loss": 1.3172,
539
+ "step": 4600
540
+ },
541
+ {
542
+ "epoch": 13.762811127379209,
543
+ "grad_norm": 2.1960484981536865,
544
+ "learning_rate": 3e-05,
545
+ "loss": 1.3185,
546
+ "step": 4700
547
+ },
548
+ {
549
+ "epoch": 14.0,
550
+ "eval_accuracy": 0.4327692307692308,
551
+ "eval_loss": 4.758542537689209,
552
+ "eval_runtime": 5.778,
553
+ "eval_samples_per_second": 86.535,
554
+ "eval_steps_per_second": 10.903,
555
+ "step": 4781
556
+ },
557
+ {
558
+ "epoch": 14.0,
559
+ "eval_bleu": 0.19085834097154403,
560
+ "eval_exact_match": 0.004,
561
+ "eval_prefix_exact_match": 0.022,
562
+ "step": 4781
563
+ },
564
+ {
565
+ "epoch": 14.055636896046853,
566
+ "grad_norm": 2.4733147621154785,
567
+ "learning_rate": 3e-05,
568
+ "loss": 1.3027,
569
+ "step": 4800
570
+ },
571
+ {
572
+ "epoch": 14.348462664714495,
573
+ "grad_norm": 2.2826030254364014,
574
+ "learning_rate": 3e-05,
575
+ "loss": 1.2513,
576
+ "step": 4900
577
+ },
578
+ {
579
+ "epoch": 14.641288433382137,
580
+ "grad_norm": 2.041346788406372,
581
+ "learning_rate": 3e-05,
582
+ "loss": 1.2637,
583
+ "step": 5000
584
+ },
585
+ {
586
+ "epoch": 14.93411420204978,
587
+ "grad_norm": 2.3494467735290527,
588
+ "learning_rate": 3e-05,
589
+ "loss": 1.2667,
590
+ "step": 5100
591
+ },
592
+ {
593
+ "epoch": 14.998535871156662,
594
+ "eval_accuracy": 0.4308717948717949,
595
+ "eval_loss": 4.938868522644043,
596
+ "eval_runtime": 5.9024,
597
+ "eval_samples_per_second": 84.711,
598
+ "eval_steps_per_second": 10.674,
599
+ "step": 5122
600
+ },
601
+ {
602
+ "epoch": 14.998535871156662,
603
+ "eval_bleu": 0.18549760176612626,
604
+ "eval_exact_match": 0.002,
605
+ "eval_prefix_exact_match": 0.024,
606
+ "step": 5122
607
+ },
608
+ {
609
+ "epoch": 15.226939970717423,
610
+ "grad_norm": 2.411022663116455,
611
+ "learning_rate": 3e-05,
612
+ "loss": 1.2158,
613
+ "step": 5200
614
+ },
615
+ {
616
+ "epoch": 15.519765739385067,
617
+ "grad_norm": 2.3778491020202637,
618
+ "learning_rate": 3e-05,
619
+ "loss": 1.193,
620
+ "step": 5300
621
+ },
622
+ {
623
+ "epoch": 15.812591508052709,
624
+ "grad_norm": 2.2003743648529053,
625
+ "learning_rate": 3e-05,
626
+ "loss": 1.2144,
627
+ "step": 5400
628
+ },
629
+ {
630
+ "epoch": 16.0,
631
+ "eval_accuracy": 0.43025641025641026,
632
+ "eval_loss": 4.898688316345215,
633
+ "eval_runtime": 5.1834,
634
+ "eval_samples_per_second": 96.463,
635
+ "eval_steps_per_second": 12.154,
636
+ "step": 5464
637
+ },
638
+ {
639
+ "epoch": 16.0,
640
+ "eval_bleu": 0.20104450766148693,
641
+ "eval_exact_match": 0.004,
642
+ "eval_prefix_exact_match": 0.026,
643
+ "step": 5464
644
+ },
645
+ {
646
+ "epoch": 16.105417276720353,
647
+ "grad_norm": 2.1213016510009766,
648
+ "learning_rate": 3e-05,
649
+ "loss": 1.1842,
650
+ "step": 5500
651
+ },
652
+ {
653
+ "epoch": 16.398243045387993,
654
+ "grad_norm": 2.4457600116729736,
655
+ "learning_rate": 3e-05,
656
+ "loss": 1.1527,
657
+ "step": 5600
658
+ },
659
+ {
660
+ "epoch": 16.691068814055637,
661
+ "grad_norm": 2.5844104290008545,
662
+ "learning_rate": 3e-05,
663
+ "loss": 1.1536,
664
+ "step": 5700
665
+ },
666
+ {
667
+ "epoch": 16.98389458272328,
668
+ "grad_norm": 2.7657299041748047,
669
+ "learning_rate": 3e-05,
670
+ "loss": 1.1708,
671
+ "step": 5800
672
+ },
673
+ {
674
+ "epoch": 16.998535871156662,
675
+ "eval_accuracy": 0.4297435897435897,
676
+ "eval_loss": 5.00166654586792,
677
+ "eval_runtime": 5.4251,
678
+ "eval_samples_per_second": 92.164,
679
+ "eval_steps_per_second": 11.613,
680
+ "step": 5805
681
+ },
682
+ {
683
+ "epoch": 16.998535871156662,
684
+ "eval_bleu": 0.2059810081312132,
685
+ "eval_exact_match": 0.004,
686
+ "eval_prefix_exact_match": 0.028,
687
+ "step": 5805
688
+ },
689
+ {
690
+ "epoch": 17.27672035139092,
691
+ "grad_norm": 2.4050705432891846,
692
+ "learning_rate": 3e-05,
693
+ "loss": 1.083,
694
+ "step": 5900
695
+ },
696
+ {
697
+ "epoch": 17.569546120058565,
698
+ "grad_norm": 2.584331750869751,
699
+ "learning_rate": 3e-05,
700
+ "loss": 1.0993,
701
+ "step": 6000
702
+ },
703
+ {
704
+ "epoch": 17.86237188872621,
705
+ "grad_norm": 3.257659912109375,
706
+ "learning_rate": 3e-05,
707
+ "loss": 1.1146,
708
+ "step": 6100
709
+ },
710
+ {
711
+ "epoch": 18.0,
712
+ "eval_accuracy": 0.43066666666666664,
713
+ "eval_loss": 4.977816581726074,
714
+ "eval_runtime": 5.2952,
715
+ "eval_samples_per_second": 94.425,
716
+ "eval_steps_per_second": 11.898,
717
+ "step": 6147
718
+ },
719
+ {
720
+ "epoch": 18.0,
721
+ "eval_bleu": 0.1948234970665932,
722
+ "eval_exact_match": 0.004,
723
+ "eval_prefix_exact_match": 0.032,
724
+ "step": 6147
725
+ },
726
+ {
727
+ "epoch": 18.15519765739385,
728
+ "grad_norm": 3.068880319595337,
729
+ "learning_rate": 3e-05,
730
+ "loss": 1.0673,
731
+ "step": 6200
732
+ },
733
+ {
734
+ "epoch": 18.448023426061493,
735
+ "grad_norm": 2.737912893295288,
736
+ "learning_rate": 3e-05,
737
+ "loss": 1.0381,
738
+ "step": 6300
739
+ },
740
+ {
741
+ "epoch": 18.740849194729137,
742
+ "grad_norm": 2.878462076187134,
743
+ "learning_rate": 3e-05,
744
+ "loss": 1.0531,
745
+ "step": 6400
746
+ },
747
+ {
748
+ "epoch": 18.998535871156662,
749
+ "eval_accuracy": 0.42866666666666664,
750
+ "eval_loss": 5.121572494506836,
751
+ "eval_runtime": 5.3996,
752
+ "eval_samples_per_second": 92.599,
753
+ "eval_steps_per_second": 11.667,
754
+ "step": 6488
755
+ },
756
+ {
757
+ "epoch": 18.998535871156662,
758
+ "eval_bleu": 0.19788734504730915,
759
+ "eval_exact_match": 0.0,
760
+ "eval_prefix_exact_match": 0.02,
761
+ "step": 6488
762
+ },
763
+ {
764
+ "epoch": 19.03367496339678,
765
+ "grad_norm": 2.6934640407562256,
766
+ "learning_rate": 3e-05,
767
+ "loss": 1.0564,
768
+ "step": 6500
769
+ },
770
+ {
771
+ "epoch": 19.32650073206442,
772
+ "grad_norm": 2.822169542312622,
773
+ "learning_rate": 3e-05,
774
+ "loss": 0.99,
775
+ "step": 6600
776
+ },
777
+ {
778
+ "epoch": 19.619326500732065,
779
+ "grad_norm": 2.520814895629883,
780
+ "learning_rate": 3e-05,
781
+ "loss": 0.991,
782
+ "step": 6700
783
+ },
784
+ {
785
+ "epoch": 19.91215226939971,
786
+ "grad_norm": 3.1566810607910156,
787
+ "learning_rate": 3e-05,
788
+ "loss": 1.0158,
789
+ "step": 6800
790
+ },
791
+ {
792
+ "epoch": 20.0,
793
+ "eval_accuracy": 0.42728205128205127,
794
+ "eval_loss": 5.120981216430664,
795
+ "eval_runtime": 5.394,
796
+ "eval_samples_per_second": 92.696,
797
+ "eval_steps_per_second": 11.68,
798
+ "step": 6830
799
+ },
800
+ {
801
+ "epoch": 20.0,
802
+ "eval_bleu": 0.15806516253752823,
803
+ "eval_exact_match": 0.006,
804
+ "eval_prefix_exact_match": 0.028,
805
+ "step": 6830
806
+ },
807
+ {
808
+ "epoch": 20.20497803806735,
809
+ "grad_norm": 2.937910556793213,
810
+ "learning_rate": 3e-05,
811
+ "loss": 0.9615,
812
+ "step": 6900
813
+ },
814
+ {
815
+ "epoch": 20.497803806734993,
816
+ "grad_norm": 3.2016685009002686,
817
+ "learning_rate": 3e-05,
818
+ "loss": 0.9404,
819
+ "step": 7000
820
+ },
821
+ {
822
+ "epoch": 20.790629575402637,
823
+ "grad_norm": 3.2996864318847656,
824
+ "learning_rate": 3e-05,
825
+ "loss": 0.9555,
826
+ "step": 7100
827
+ },
828
+ {
829
+ "epoch": 20.998535871156662,
830
+ "eval_accuracy": 0.42928205128205127,
831
+ "eval_loss": 5.198832988739014,
832
+ "eval_runtime": 5.8905,
833
+ "eval_samples_per_second": 84.882,
834
+ "eval_steps_per_second": 10.695,
835
+ "step": 7171
836
+ },
837
+ {
838
+ "epoch": 20.998535871156662,
839
+ "eval_bleu": 0.18303907416923465,
840
+ "eval_exact_match": 0.004,
841
+ "eval_prefix_exact_match": 0.02,
842
+ "step": 7171
843
+ },
844
+ {
845
+ "epoch": 21.083455344070277,
846
+ "grad_norm": 3.4803571701049805,
847
+ "learning_rate": 3e-05,
848
+ "loss": 0.9454,
849
+ "step": 7200
850
+ },
851
+ {
852
+ "epoch": 21.37628111273792,
853
+ "grad_norm": 2.6315698623657227,
854
+ "learning_rate": 3e-05,
855
+ "loss": 0.8835,
856
+ "step": 7300
857
+ },
858
+ {
859
+ "epoch": 21.669106881405565,
860
+ "grad_norm": 3.0926690101623535,
861
+ "learning_rate": 3e-05,
862
+ "loss": 0.9004,
863
+ "step": 7400
864
+ },
865
+ {
866
+ "epoch": 21.961932650073205,
867
+ "grad_norm": 2.9990713596343994,
868
+ "learning_rate": 3e-05,
869
+ "loss": 0.9205,
870
+ "step": 7500
871
+ },
872
+ {
873
+ "epoch": 22.0,
874
+ "eval_accuracy": 0.427025641025641,
875
+ "eval_loss": 5.2239508628845215,
876
+ "eval_runtime": 5.4025,
877
+ "eval_samples_per_second": 92.55,
878
+ "eval_steps_per_second": 11.661,
879
+ "step": 7513
880
+ },
881
+ {
882
+ "epoch": 22.0,
883
+ "eval_bleu": 0.1854279974336752,
884
+ "eval_exact_match": 0.004,
885
+ "eval_prefix_exact_match": 0.02,
886
+ "step": 7513
887
+ },
888
+ {
889
+ "epoch": 22.25475841874085,
890
+ "grad_norm": 3.2822601795196533,
891
+ "learning_rate": 3e-05,
892
+ "loss": 0.8461,
893
+ "step": 7600
894
+ },
895
+ {
896
+ "epoch": 22.547584187408493,
897
+ "grad_norm": 3.4992525577545166,
898
+ "learning_rate": 3e-05,
899
+ "loss": 0.8485,
900
+ "step": 7700
901
+ },
902
+ {
903
+ "epoch": 22.840409956076133,
904
+ "grad_norm": 3.7167255878448486,
905
+ "learning_rate": 3e-05,
906
+ "loss": 0.8711,
907
+ "step": 7800
908
+ },
909
+ {
910
+ "epoch": 22.998535871156662,
911
+ "eval_accuracy": 0.42512820512820515,
912
+ "eval_loss": 5.346656322479248,
913
+ "eval_runtime": 5.4264,
914
+ "eval_samples_per_second": 92.143,
915
+ "eval_steps_per_second": 11.61,
916
+ "step": 7854
917
+ },
918
+ {
919
+ "epoch": 22.998535871156662,
920
+ "eval_bleu": 0.18735744635859433,
921
+ "eval_exact_match": 0.002,
922
+ "eval_prefix_exact_match": 0.016,
923
+ "step": 7854
924
+ },
925
+ {
926
+ "epoch": 23.133235724743777,
927
+ "grad_norm": 3.3037803173065186,
928
+ "learning_rate": 3e-05,
929
+ "loss": 0.8378,
930
+ "step": 7900
931
+ },
932
+ {
933
+ "epoch": 23.42606149341142,
934
+ "grad_norm": 3.446135997772217,
935
+ "learning_rate": 3e-05,
936
+ "loss": 0.8102,
937
+ "step": 8000
938
+ },
939
+ {
940
+ "epoch": 23.718887262079065,
941
+ "grad_norm": 3.0248115062713623,
942
+ "learning_rate": 3e-05,
943
+ "loss": 0.8082,
944
+ "step": 8100
945
+ },
946
+ {
947
+ "epoch": 24.0,
948
+ "eval_accuracy": 0.42430769230769233,
949
+ "eval_loss": 5.35550594329834,
950
+ "eval_runtime": 5.1152,
951
+ "eval_samples_per_second": 97.747,
952
+ "eval_steps_per_second": 12.316,
953
+ "step": 8196
954
+ },
955
+ {
956
+ "epoch": 24.0,
957
+ "eval_bleu": 0.1404162407909808,
958
+ "eval_exact_match": 0.004,
959
+ "eval_prefix_exact_match": 0.016,
960
+ "step": 8196
961
+ },
962
+ {
963
+ "epoch": 24.011713030746705,
964
+ "grad_norm": 3.7103264331817627,
965
+ "learning_rate": 3e-05,
966
+ "loss": 0.8279,
967
+ "step": 8200
968
+ },
969
+ {
970
+ "epoch": 24.30453879941435,
971
+ "grad_norm": 4.033411026000977,
972
+ "learning_rate": 3e-05,
973
+ "loss": 0.7449,
974
+ "step": 8300
975
+ },
976
+ {
977
+ "epoch": 24.597364568081993,
978
+ "grad_norm": 3.5919322967529297,
979
+ "learning_rate": 3e-05,
980
+ "loss": 0.7654,
981
+ "step": 8400
982
+ },
983
+ {
984
+ "epoch": 24.890190336749633,
985
+ "grad_norm": 3.9135005474090576,
986
+ "learning_rate": 3e-05,
987
+ "loss": 0.7854,
988
+ "step": 8500
989
+ },
990
+ {
991
+ "epoch": 24.998535871156662,
992
+ "eval_accuracy": 0.4241025641025641,
993
+ "eval_loss": 5.4628777503967285,
994
+ "eval_runtime": 5.4188,
995
+ "eval_samples_per_second": 92.272,
996
+ "eval_steps_per_second": 11.626,
997
+ "step": 8537
998
+ },
999
+ {
1000
+ "epoch": 24.998535871156662,
1001
+ "eval_bleu": 0.16704717168781627,
1002
+ "eval_exact_match": 0.002,
1003
+ "eval_prefix_exact_match": 0.018,
1004
+ "step": 8537
1005
+ },
1006
+ {
1007
+ "epoch": 25.183016105417277,
1008
+ "grad_norm": 3.4372332096099854,
1009
+ "learning_rate": 3e-05,
1010
+ "loss": 0.7377,
1011
+ "step": 8600
1012
+ },
1013
+ {
1014
+ "epoch": 25.47584187408492,
1015
+ "grad_norm": 3.6895363330841064,
1016
+ "learning_rate": 3e-05,
1017
+ "loss": 0.7141,
1018
+ "step": 8700
1019
+ },
1020
+ {
1021
+ "epoch": 25.76866764275256,
1022
+ "grad_norm": 3.793881893157959,
1023
+ "learning_rate": 3e-05,
1024
+ "loss": 0.7359,
1025
+ "step": 8800
1026
+ },
1027
+ {
1028
+ "epoch": 26.0,
1029
+ "eval_accuracy": 0.4230769230769231,
1030
+ "eval_loss": 5.469931602478027,
1031
+ "eval_runtime": 5.4593,
1032
+ "eval_samples_per_second": 91.587,
1033
+ "eval_steps_per_second": 11.54,
1034
+ "step": 8879
1035
+ },
1036
+ {
1037
+ "epoch": 26.0,
1038
+ "eval_bleu": 0.20994054385069763,
1039
+ "eval_exact_match": 0.0,
1040
+ "eval_prefix_exact_match": 0.018,
1041
+ "step": 8879
1042
+ },
1043
+ {
1044
+ "epoch": 26.061493411420205,
1045
+ "grad_norm": 3.7599456310272217,
1046
+ "learning_rate": 3e-05,
1047
+ "loss": 0.7299,
1048
+ "step": 8900
1049
+ },
1050
+ {
1051
+ "epoch": 26.35431918008785,
1052
+ "grad_norm": 3.9128973484039307,
1053
+ "learning_rate": 3e-05,
1054
+ "loss": 0.6829,
1055
+ "step": 9000
1056
+ },
1057
+ {
1058
+ "epoch": 26.64714494875549,
1059
+ "grad_norm": 3.68682861328125,
1060
+ "learning_rate": 3e-05,
1061
+ "loss": 0.6835,
1062
+ "step": 9100
1063
+ },
1064
+ {
1065
+ "epoch": 26.939970717423133,
1066
+ "grad_norm": 3.563685894012451,
1067
+ "learning_rate": 3e-05,
1068
+ "loss": 0.7002,
1069
+ "step": 9200
1070
+ },
1071
+ {
1072
+ "epoch": 26.998535871156662,
1073
+ "eval_accuracy": 0.42107692307692307,
1074
+ "eval_loss": 5.505324363708496,
1075
+ "eval_runtime": 5.119,
1076
+ "eval_samples_per_second": 97.676,
1077
+ "eval_steps_per_second": 12.307,
1078
+ "step": 9220
1079
+ },
1080
+ {
1081
+ "epoch": 26.998535871156662,
1082
+ "eval_bleu": 0.18409646236051158,
1083
+ "eval_exact_match": 0.004,
1084
+ "eval_prefix_exact_match": 0.02,
1085
+ "step": 9220
1086
+ },
1087
+ {
1088
+ "epoch": 27.232796486090777,
1089
+ "grad_norm": 3.794928550720215,
1090
+ "learning_rate": 3e-05,
1091
+ "loss": 0.6346,
1092
+ "step": 9300
1093
+ },
1094
+ {
1095
+ "epoch": 27.525622254758417,
1096
+ "grad_norm": 3.7905471324920654,
1097
+ "learning_rate": 3e-05,
1098
+ "loss": 0.6439,
1099
+ "step": 9400
1100
+ },
1101
+ {
1102
+ "epoch": 27.81844802342606,
1103
+ "grad_norm": 4.2111711502075195,
1104
+ "learning_rate": 3e-05,
1105
+ "loss": 0.6684,
1106
+ "step": 9500
1107
+ },
1108
+ {
1109
+ "epoch": 28.0,
1110
+ "eval_accuracy": 0.42138461538461536,
1111
+ "eval_loss": 5.558403491973877,
1112
+ "eval_runtime": 5.3841,
1113
+ "eval_samples_per_second": 92.866,
1114
+ "eval_steps_per_second": 11.701,
1115
+ "step": 9562
1116
+ },
1117
+ {
1118
+ "epoch": 28.0,
1119
+ "eval_bleu": 0.15544965656402868,
1120
+ "eval_exact_match": 0.004,
1121
+ "eval_prefix_exact_match": 0.016,
1122
+ "step": 9562
1123
+ },
1124
+ {
1125
+ "epoch": 28.111273792093705,
1126
+ "grad_norm": 4.060698509216309,
1127
+ "learning_rate": 3e-05,
1128
+ "loss": 0.6482,
1129
+ "step": 9600
1130
+ },
1131
+ {
1132
+ "epoch": 28.404099560761345,
1133
+ "grad_norm": 3.7367396354675293,
1134
+ "learning_rate": 3e-05,
1135
+ "loss": 0.5946,
1136
+ "step": 9700
1137
+ },
1138
+ {
1139
+ "epoch": 28.69692532942899,
1140
+ "grad_norm": 3.5368525981903076,
1141
+ "learning_rate": 3e-05,
1142
+ "loss": 0.6243,
1143
+ "step": 9800
1144
+ },
1145
+ {
1146
+ "epoch": 28.989751098096633,
1147
+ "grad_norm": 3.896458864212036,
1148
+ "learning_rate": 3e-05,
1149
+ "loss": 0.6236,
1150
+ "step": 9900
1151
+ },
1152
+ {
1153
+ "epoch": 28.998535871156662,
1154
+ "eval_accuracy": 0.4196410256410256,
1155
+ "eval_loss": 5.61169958114624,
1156
+ "eval_runtime": 5.7867,
1157
+ "eval_samples_per_second": 86.405,
1158
+ "eval_steps_per_second": 10.887,
1159
+ "step": 9903
1160
+ },
1161
+ {
1162
+ "epoch": 28.998535871156662,
1163
+ "eval_bleu": 0.16226895197934726,
1164
+ "eval_exact_match": 0.004,
1165
+ "eval_prefix_exact_match": 0.02,
1166
+ "step": 9903
1167
+ },
1168
+ {
1169
+ "epoch": 29.282576866764277,
1170
+ "grad_norm": 3.8376243114471436,
1171
+ "learning_rate": 3e-05,
1172
+ "loss": 0.5619,
1173
+ "step": 10000
1174
+ },
1175
+ {
1176
+ "epoch": 29.575402635431917,
1177
+ "grad_norm": 5.112703800201416,
1178
+ "learning_rate": 3e-05,
1179
+ "loss": 0.5798,
1180
+ "step": 10100
1181
+ },
1182
+ {
1183
+ "epoch": 29.86822840409956,
1184
+ "grad_norm": 3.9134209156036377,
1185
+ "learning_rate": 3e-05,
1186
+ "loss": 0.5866,
1187
+ "step": 10200
1188
+ },
1189
+ {
1190
+ "epoch": 30.0,
1191
+ "eval_accuracy": 0.4196410256410256,
1192
+ "eval_loss": 5.569637775421143,
1193
+ "eval_runtime": 5.7621,
1194
+ "eval_samples_per_second": 86.773,
1195
+ "eval_steps_per_second": 10.933,
1196
+ "step": 10245
1197
+ },
1198
+ {
1199
+ "epoch": 30.0,
1200
+ "eval_bleu": 0.14900418852538308,
1201
+ "eval_exact_match": 0.002,
1202
+ "eval_prefix_exact_match": 0.022,
1203
+ "step": 10245
1204
+ },
1205
+ {
1206
+ "epoch": 30.161054172767205,
1207
+ "grad_norm": 3.566251754760742,
1208
+ "learning_rate": 3e-05,
1209
+ "loss": 0.5663,
1210
+ "step": 10300
1211
+ },
1212
+ {
1213
+ "epoch": 30.453879941434845,
1214
+ "grad_norm": 4.584939956665039,
1215
+ "learning_rate": 3e-05,
1216
+ "loss": 0.5335,
1217
+ "step": 10400
1218
+ },
1219
+ {
1220
+ "epoch": 30.74670571010249,
1221
+ "grad_norm": 3.595506429672241,
1222
+ "learning_rate": 3e-05,
1223
+ "loss": 0.5554,
1224
+ "step": 10500
1225
+ },
1226
+ {
1227
+ "epoch": 30.998535871156662,
1228
+ "eval_accuracy": 0.4184102564102564,
1229
+ "eval_loss": 5.657910346984863,
1230
+ "eval_runtime": 5.1292,
1231
+ "eval_samples_per_second": 97.48,
1232
+ "eval_steps_per_second": 12.283,
1233
+ "step": 10586
1234
+ },
1235
+ {
1236
+ "epoch": 30.998535871156662,
1237
+ "eval_bleu": 0.17143838636149805,
1238
+ "eval_exact_match": 0.004,
1239
+ "eval_prefix_exact_match": 0.018,
1240
+ "step": 10586
1241
+ },
1242
+ {
1243
+ "epoch": 31.039531478770133,
1244
+ "grad_norm": 3.704332113265991,
1245
+ "learning_rate": 3e-05,
1246
+ "loss": 0.549,
1247
+ "step": 10600
1248
+ },
1249
+ {
1250
+ "epoch": 31.332357247437773,
1251
+ "grad_norm": 3.8732120990753174,
1252
+ "learning_rate": 3e-05,
1253
+ "loss": 0.4969,
1254
+ "step": 10700
1255
+ },
1256
+ {
1257
+ "epoch": 31.625183016105417,
1258
+ "grad_norm": 3.426780939102173,
1259
+ "learning_rate": 3e-05,
1260
+ "loss": 0.5227,
1261
+ "step": 10800
1262
+ },
1263
+ {
1264
+ "epoch": 31.91800878477306,
1265
+ "grad_norm": 3.652219533920288,
1266
+ "learning_rate": 3e-05,
1267
+ "loss": 0.5291,
1268
+ "step": 10900
1269
+ },
1270
+ {
1271
+ "epoch": 32.0,
1272
+ "eval_accuracy": 0.41784615384615387,
1273
+ "eval_loss": 5.739648342132568,
1274
+ "eval_runtime": 5.9377,
1275
+ "eval_samples_per_second": 84.208,
1276
+ "eval_steps_per_second": 10.61,
1277
+ "step": 10928
1278
+ },
1279
+ {
1280
+ "epoch": 32.0,
1281
+ "eval_bleu": 0.1836146339343278,
1282
+ "eval_exact_match": 0.006,
1283
+ "eval_prefix_exact_match": 0.02,
1284
+ "step": 10928
1285
+ },
1286
+ {
1287
+ "epoch": 32.210834553440705,
1288
+ "grad_norm": 3.8332104682922363,
1289
+ "learning_rate": 3e-05,
1290
+ "loss": 0.4831,
1291
+ "step": 11000
1292
+ },
1293
+ {
1294
+ "epoch": 32.503660322108345,
1295
+ "grad_norm": 4.138996601104736,
1296
+ "learning_rate": 3e-05,
1297
+ "loss": 0.4795,
1298
+ "step": 11100
1299
+ },
1300
+ {
1301
+ "epoch": 32.796486090775986,
1302
+ "grad_norm": 3.8147737979888916,
1303
+ "learning_rate": 3e-05,
1304
+ "loss": 0.4968,
1305
+ "step": 11200
1306
+ },
1307
+ {
1308
+ "epoch": 32.99853587115666,
1309
+ "eval_accuracy": 0.41805128205128206,
1310
+ "eval_loss": 5.811046600341797,
1311
+ "eval_runtime": 5.4234,
1312
+ "eval_samples_per_second": 92.192,
1313
+ "eval_steps_per_second": 11.616,
1314
+ "step": 11269
1315
+ },
1316
+ {
1317
+ "epoch": 32.99853587115666,
1318
+ "eval_bleu": 0.14590511236244538,
1319
+ "eval_exact_match": 0.004,
1320
+ "eval_prefix_exact_match": 0.018,
1321
+ "step": 11269
1322
+ },
1323
+ {
1324
+ "epoch": 33.08931185944363,
1325
+ "grad_norm": 3.952998399734497,
1326
+ "learning_rate": 3e-05,
1327
+ "loss": 0.4835,
1328
+ "step": 11300
1329
+ },
1330
+ {
1331
+ "epoch": 33.38213762811127,
1332
+ "grad_norm": 3.589601516723633,
1333
+ "learning_rate": 3e-05,
1334
+ "loss": 0.4496,
1335
+ "step": 11400
1336
+ },
1337
+ {
1338
+ "epoch": 33.674963396778914,
1339
+ "grad_norm": 3.9857985973358154,
1340
+ "learning_rate": 3e-05,
1341
+ "loss": 0.4672,
1342
+ "step": 11500
1343
+ },
1344
+ {
1345
+ "epoch": 33.96778916544656,
1346
+ "grad_norm": 3.6186344623565674,
1347
+ "learning_rate": 3e-05,
1348
+ "loss": 0.4635,
1349
+ "step": 11600
1350
+ },
1351
+ {
1352
+ "epoch": 34.0,
1353
+ "eval_accuracy": 0.4167179487179487,
1354
+ "eval_loss": 5.871949195861816,
1355
+ "eval_runtime": 5.1434,
1356
+ "eval_samples_per_second": 97.213,
1357
+ "eval_steps_per_second": 12.249,
1358
+ "step": 11611
1359
+ },
1360
+ {
1361
+ "epoch": 34.0,
1362
+ "eval_bleu": 0.16219479229123285,
1363
+ "eval_exact_match": 0.004,
1364
+ "eval_prefix_exact_match": 0.018,
1365
+ "step": 11611
1366
+ },
1367
+ {
1368
+ "epoch": 34.2606149341142,
1369
+ "grad_norm": 3.6174490451812744,
1370
+ "learning_rate": 3e-05,
1371
+ "loss": 0.4212,
1372
+ "step": 11700
1373
+ },
1374
+ {
1375
+ "epoch": 34.55344070278184,
1376
+ "grad_norm": 3.5105185508728027,
1377
+ "learning_rate": 3e-05,
1378
+ "loss": 0.4261,
1379
+ "step": 11800
1380
+ },
1381
+ {
1382
+ "epoch": 34.84626647144949,
1383
+ "grad_norm": 3.934603452682495,
1384
+ "learning_rate": 3e-05,
1385
+ "loss": 0.4465,
1386
+ "step": 11900
1387
+ },
1388
+ {
1389
+ "epoch": 34.99853587115666,
1390
+ "eval_accuracy": 0.4162051282051282,
1391
+ "eval_loss": 5.8657732009887695,
1392
+ "eval_runtime": 5.2208,
1393
+ "eval_samples_per_second": 95.772,
1394
+ "eval_steps_per_second": 12.067,
1395
+ "step": 11952
1396
+ },
1397
+ {
1398
+ "epoch": 34.99853587115666,
1399
+ "eval_bleu": 0.1252956011670444,
1400
+ "eval_exact_match": 0.002,
1401
+ "eval_prefix_exact_match": 0.016,
1402
+ "step": 11952
1403
+ },
1404
+ {
1405
+ "epoch": 35.13909224011713,
1406
+ "grad_norm": 4.277783393859863,
1407
+ "learning_rate": 3e-05,
1408
+ "loss": 0.418,
1409
+ "step": 12000
1410
+ },
1411
+ {
1412
+ "epoch": 35.43191800878477,
1413
+ "grad_norm": 4.015263080596924,
1414
+ "learning_rate": 3e-05,
1415
+ "loss": 0.3969,
1416
+ "step": 12100
1417
+ },
1418
+ {
1419
+ "epoch": 35.72474377745242,
1420
+ "grad_norm": 3.9678900241851807,
1421
+ "learning_rate": 3e-05,
1422
+ "loss": 0.4184,
1423
+ "step": 12200
1424
+ },
1425
+ {
1426
+ "epoch": 36.0,
1427
+ "eval_accuracy": 0.4146666666666667,
1428
+ "eval_loss": 5.888679504394531,
1429
+ "eval_runtime": 5.1605,
1430
+ "eval_samples_per_second": 96.891,
1431
+ "eval_steps_per_second": 12.208,
1432
+ "step": 12294
1433
+ },
1434
+ {
1435
+ "epoch": 36.0,
1436
+ "eval_bleu": 0.1890401692938485,
1437
+ "eval_exact_match": 0.002,
1438
+ "eval_prefix_exact_match": 0.02,
1439
+ "step": 12294
1440
+ },
1441
+ {
1442
+ "epoch": 36.01756954612006,
1443
+ "grad_norm": 4.230247497558594,
1444
+ "learning_rate": 3e-05,
1445
+ "loss": 0.419,
1446
+ "step": 12300
1447
+ },
1448
+ {
1449
+ "epoch": 36.3103953147877,
1450
+ "grad_norm": 3.046856164932251,
1451
+ "learning_rate": 3e-05,
1452
+ "loss": 0.3716,
1453
+ "step": 12400
1454
+ },
1455
+ {
1456
+ "epoch": 36.603221083455345,
1457
+ "grad_norm": 4.008014678955078,
1458
+ "learning_rate": 3e-05,
1459
+ "loss": 0.3849,
1460
+ "step": 12500
1461
+ },
1462
+ {
1463
+ "epoch": 36.896046852122986,
1464
+ "grad_norm": 4.294738292694092,
1465
+ "learning_rate": 3e-05,
1466
+ "loss": 0.4002,
1467
+ "step": 12600
1468
+ },
1469
+ {
1470
+ "epoch": 36.99853587115666,
1471
+ "eval_accuracy": 0.41646153846153844,
1472
+ "eval_loss": 5.994997024536133,
1473
+ "eval_runtime": 5.8607,
1474
+ "eval_samples_per_second": 85.313,
1475
+ "eval_steps_per_second": 10.749,
1476
+ "step": 12635
1477
+ },
1478
+ {
1479
+ "epoch": 36.99853587115666,
1480
+ "eval_bleu": 0.13560559376907821,
1481
+ "eval_exact_match": 0.002,
1482
+ "eval_prefix_exact_match": 0.022,
1483
+ "step": 12635
1484
+ },
1485
+ {
1486
+ "epoch": 37.18887262079063,
1487
+ "grad_norm": 3.3634254932403564,
1488
+ "learning_rate": 3e-05,
1489
+ "loss": 0.36,
1490
+ "step": 12700
1491
+ },
1492
+ {
1493
+ "epoch": 37.48169838945827,
1494
+ "grad_norm": 3.8797719478607178,
1495
+ "learning_rate": 3e-05,
1496
+ "loss": 0.3593,
1497
+ "step": 12800
1498
+ },
1499
+ {
1500
+ "epoch": 37.774524158125914,
1501
+ "grad_norm": 3.8350632190704346,
1502
+ "learning_rate": 3e-05,
1503
+ "loss": 0.3716,
1504
+ "step": 12900
1505
+ },
1506
+ {
1507
+ "epoch": 38.0,
1508
+ "eval_accuracy": 0.4155384615384615,
1509
+ "eval_loss": 5.999116897583008,
1510
+ "eval_runtime": 5.1523,
1511
+ "eval_samples_per_second": 97.044,
1512
+ "eval_steps_per_second": 12.228,
1513
+ "step": 12977
1514
+ },
1515
+ {
1516
+ "epoch": 38.0,
1517
+ "eval_bleu": 0.1596199306892504,
1518
+ "eval_exact_match": 0.002,
1519
+ "eval_prefix_exact_match": 0.016,
1520
+ "step": 12977
1521
+ },
1522
+ {
1523
+ "epoch": 38.06734992679356,
1524
+ "grad_norm": 3.2861812114715576,
1525
+ "learning_rate": 3e-05,
1526
+ "loss": 0.368,
1527
+ "step": 13000
1528
+ },
1529
+ {
1530
+ "epoch": 38.3601756954612,
1531
+ "grad_norm": 4.50984001159668,
1532
+ "learning_rate": 3e-05,
1533
+ "loss": 0.3347,
1534
+ "step": 13100
1535
+ },
1536
+ {
1537
+ "epoch": 38.65300146412884,
1538
+ "grad_norm": 4.76189661026001,
1539
+ "learning_rate": 3e-05,
1540
+ "loss": 0.3473,
1541
+ "step": 13200
1542
+ },
1543
+ {
1544
+ "epoch": 38.94582723279649,
1545
+ "grad_norm": 4.357806205749512,
1546
+ "learning_rate": 3e-05,
1547
+ "loss": 0.3537,
1548
+ "step": 13300
1549
+ },
1550
+ {
1551
+ "epoch": 38.99853587115666,
1552
+ "eval_accuracy": 0.41512820512820514,
1553
+ "eval_loss": 6.072280406951904,
1554
+ "eval_runtime": 5.1624,
1555
+ "eval_samples_per_second": 96.855,
1556
+ "eval_steps_per_second": 12.204,
1557
+ "step": 13318
1558
+ },
1559
+ {
1560
+ "epoch": 38.99853587115666,
1561
+ "eval_bleu": 0.16955034123413643,
1562
+ "eval_exact_match": 0.002,
1563
+ "eval_prefix_exact_match": 0.018,
1564
+ "step": 13318
1565
+ },
1566
+ {
1567
+ "epoch": 39.23865300146413,
1568
+ "grad_norm": 3.087754726409912,
1569
+ "learning_rate": 3e-05,
1570
+ "loss": 0.3191,
1571
+ "step": 13400
1572
+ },
1573
+ {
1574
+ "epoch": 39.53147877013177,
1575
+ "grad_norm": 3.584862232208252,
1576
+ "learning_rate": 3e-05,
1577
+ "loss": 0.3182,
1578
+ "step": 13500
1579
+ },
1580
+ {
1581
+ "epoch": 39.82430453879942,
1582
+ "grad_norm": 3.957606792449951,
1583
+ "learning_rate": 3e-05,
1584
+ "loss": 0.3361,
1585
+ "step": 13600
1586
+ },
1587
+ {
1588
+ "epoch": 40.0,
1589
+ "eval_accuracy": 0.4127179487179487,
1590
+ "eval_loss": 6.077693462371826,
1591
+ "eval_runtime": 5.4005,
1592
+ "eval_samples_per_second": 92.584,
1593
+ "eval_steps_per_second": 11.666,
1594
+ "step": 13660
1595
+ },
1596
+ {
1597
+ "epoch": 40.0,
1598
+ "eval_bleu": 0.15255072429100688,
1599
+ "eval_exact_match": 0.002,
1600
+ "eval_prefix_exact_match": 0.018,
1601
+ "step": 13660
1602
+ },
1603
+ {
1604
+ "epoch": 40.11713030746706,
1605
+ "grad_norm": 2.7203288078308105,
1606
+ "learning_rate": 3e-05,
1607
+ "loss": 0.3213,
1608
+ "step": 13700
1609
+ },
1610
+ {
1611
+ "epoch": 40.4099560761347,
1612
+ "grad_norm": 3.8791556358337402,
1613
+ "learning_rate": 3e-05,
1614
+ "loss": 0.3001,
1615
+ "step": 13800
1616
+ },
1617
+ {
1618
+ "epoch": 40.702781844802345,
1619
+ "grad_norm": 3.7646310329437256,
1620
+ "learning_rate": 3e-05,
1621
+ "loss": 0.3145,
1622
+ "step": 13900
1623
+ },
1624
+ {
1625
+ "epoch": 40.995607613469986,
1626
+ "grad_norm": 3.44431471824646,
1627
+ "learning_rate": 3e-05,
1628
+ "loss": 0.3199,
1629
+ "step": 14000
1630
+ },
1631
+ {
1632
+ "epoch": 40.99853587115666,
1633
+ "eval_accuracy": 0.41497435897435897,
1634
+ "eval_loss": 6.118078231811523,
1635
+ "eval_runtime": 5.8762,
1636
+ "eval_samples_per_second": 85.089,
1637
+ "eval_steps_per_second": 10.721,
1638
+ "step": 14001
1639
+ },
1640
+ {
1641
+ "epoch": 40.99853587115666,
1642
+ "eval_bleu": 0.1668915012587408,
1643
+ "eval_exact_match": 0.002,
1644
+ "eval_prefix_exact_match": 0.02,
1645
+ "step": 14001
1646
+ },
1647
+ {
1648
+ "epoch": 41.288433382137626,
1649
+ "grad_norm": 2.8688769340515137,
1650
+ "learning_rate": 3e-05,
1651
+ "loss": 0.2825,
1652
+ "step": 14100
1653
+ },
1654
+ {
1655
+ "epoch": 41.58125915080527,
1656
+ "grad_norm": 3.6143219470977783,
1657
+ "learning_rate": 3e-05,
1658
+ "loss": 0.2929,
1659
+ "step": 14200
1660
+ },
1661
+ {
1662
+ "epoch": 41.874084919472914,
1663
+ "grad_norm": 3.6292898654937744,
1664
+ "learning_rate": 3e-05,
1665
+ "loss": 0.303,
1666
+ "step": 14300
1667
+ },
1668
+ {
1669
+ "epoch": 42.0,
1670
+ "eval_accuracy": 0.4134358974358974,
1671
+ "eval_loss": 6.091122150421143,
1672
+ "eval_runtime": 5.2917,
1673
+ "eval_samples_per_second": 94.488,
1674
+ "eval_steps_per_second": 11.905,
1675
+ "step": 14343
1676
+ },
1677
+ {
1678
+ "epoch": 42.0,
1679
+ "eval_bleu": 0.12870012549393656,
1680
+ "eval_exact_match": 0.002,
1681
+ "eval_prefix_exact_match": 0.018,
1682
+ "step": 14343
1683
+ },
1684
+ {
1685
+ "epoch": 42.166910688140554,
1686
+ "grad_norm": 3.775843858718872,
1687
+ "learning_rate": 3e-05,
1688
+ "loss": 0.2812,
1689
+ "step": 14400
1690
+ },
1691
+ {
1692
+ "epoch": 42.4597364568082,
1693
+ "grad_norm": 5.015231132507324,
1694
+ "learning_rate": 3e-05,
1695
+ "loss": 0.277,
1696
+ "step": 14500
1697
+ },
1698
+ {
1699
+ "epoch": 42.75256222547584,
1700
+ "grad_norm": 3.9435009956359863,
1701
+ "learning_rate": 3e-05,
1702
+ "loss": 0.2797,
1703
+ "step": 14600
1704
+ },
1705
+ {
1706
+ "epoch": 42.99853587115666,
1707
+ "eval_accuracy": 0.41446153846153844,
1708
+ "eval_loss": 6.160712718963623,
1709
+ "eval_runtime": 5.4041,
1710
+ "eval_samples_per_second": 92.522,
1711
+ "eval_steps_per_second": 11.658,
1712
+ "step": 14684
1713
+ },
1714
+ {
1715
+ "epoch": 42.99853587115666,
1716
+ "eval_bleu": 0.11309097139944661,
1717
+ "eval_exact_match": 0.002,
1718
+ "eval_prefix_exact_match": 0.014,
1719
+ "step": 14684
1720
+ },
1721
+ {
1722
+ "epoch": 43.04538799414348,
1723
+ "grad_norm": 3.2064316272735596,
1724
+ "learning_rate": 3e-05,
1725
+ "loss": 0.2848,
1726
+ "step": 14700
1727
+ },
1728
+ {
1729
+ "epoch": 43.33821376281113,
1730
+ "grad_norm": 3.3895232677459717,
1731
+ "learning_rate": 3e-05,
1732
+ "loss": 0.2572,
1733
+ "step": 14800
1734
+ },
1735
+ {
1736
+ "epoch": 43.63103953147877,
1737
+ "grad_norm": 2.8636820316314697,
1738
+ "learning_rate": 3e-05,
1739
+ "loss": 0.2647,
1740
+ "step": 14900
1741
+ },
1742
+ {
1743
+ "epoch": 43.92386530014641,
1744
+ "grad_norm": 3.237957715988159,
1745
+ "learning_rate": 3e-05,
1746
+ "loss": 0.2762,
1747
+ "step": 15000
1748
+ },
1749
+ {
1750
+ "epoch": 44.0,
1751
+ "eval_accuracy": 0.4126153846153846,
1752
+ "eval_loss": 6.112813472747803,
1753
+ "eval_runtime": 5.1493,
1754
+ "eval_samples_per_second": 97.101,
1755
+ "eval_steps_per_second": 12.235,
1756
+ "step": 15026
1757
+ },
1758
+ {
1759
+ "epoch": 44.0,
1760
+ "eval_bleu": 0.15184742628269826,
1761
+ "eval_exact_match": 0.002,
1762
+ "eval_prefix_exact_match": 0.016,
1763
+ "step": 15026
1764
+ },
1765
+ {
1766
+ "epoch": 44.21669106881406,
1767
+ "grad_norm": 4.982185363769531,
1768
+ "learning_rate": 3e-05,
1769
+ "loss": 0.2529,
1770
+ "step": 15100
1771
+ },
1772
+ {
1773
+ "epoch": 44.5095168374817,
1774
+ "grad_norm": 3.329613447189331,
1775
+ "learning_rate": 3e-05,
1776
+ "loss": 0.2484,
1777
+ "step": 15200
1778
+ },
1779
+ {
1780
+ "epoch": 44.80234260614934,
1781
+ "grad_norm": 3.69146728515625,
1782
+ "learning_rate": 3e-05,
1783
+ "loss": 0.2633,
1784
+ "step": 15300
1785
+ },
1786
+ {
1787
+ "epoch": 44.99853587115666,
1788
+ "eval_accuracy": 0.4126666666666667,
1789
+ "eval_loss": 6.144558429718018,
1790
+ "eval_runtime": 5.0796,
1791
+ "eval_samples_per_second": 98.433,
1792
+ "eval_steps_per_second": 12.403,
1793
+ "step": 15367
1794
+ },
1795
+ {
1796
+ "epoch": 44.99853587115666,
1797
+ "eval_bleu": 0.12121489324074226,
1798
+ "eval_exact_match": 0.002,
1799
+ "eval_prefix_exact_match": 0.022,
1800
+ "step": 15367
1801
+ },
1802
+ {
1803
+ "epoch": 45.095168374816986,
1804
+ "grad_norm": 2.742215394973755,
1805
+ "learning_rate": 3e-05,
1806
+ "loss": 0.2526,
1807
+ "step": 15400
1808
+ },
1809
+ {
1810
+ "epoch": 45.387994143484626,
1811
+ "grad_norm": 3.184886932373047,
1812
+ "learning_rate": 3e-05,
1813
+ "loss": 0.2367,
1814
+ "step": 15500
1815
+ },
1816
+ {
1817
+ "epoch": 45.680819912152266,
1818
+ "grad_norm": 3.4894936084747314,
1819
+ "learning_rate": 3e-05,
1820
+ "loss": 0.2467,
1821
+ "step": 15600
1822
+ },
1823
+ {
1824
+ "epoch": 45.973645680819914,
1825
+ "grad_norm": 3.247821569442749,
1826
+ "learning_rate": 3e-05,
1827
+ "loss": 0.2508,
1828
+ "step": 15700
1829
+ },
1830
+ {
1831
+ "epoch": 46.0,
1832
+ "eval_accuracy": 0.4133846153846154,
1833
+ "eval_loss": 6.233008861541748,
1834
+ "eval_runtime": 5.4421,
1835
+ "eval_samples_per_second": 91.876,
1836
+ "eval_steps_per_second": 11.576,
1837
+ "step": 15709
1838
+ },
1839
+ {
1840
+ "epoch": 46.0,
1841
+ "eval_bleu": 0.15833529804586344,
1842
+ "eval_exact_match": 0.002,
1843
+ "eval_prefix_exact_match": 0.02,
1844
+ "step": 15709
1845
+ },
1846
+ {
1847
+ "epoch": 46.266471449487554,
1848
+ "grad_norm": 3.061795234680176,
1849
+ "learning_rate": 3e-05,
1850
+ "loss": 0.2257,
1851
+ "step": 15800
1852
+ },
1853
+ {
1854
+ "epoch": 46.559297218155194,
1855
+ "grad_norm": 3.158151149749756,
1856
+ "learning_rate": 3e-05,
1857
+ "loss": 0.231,
1858
+ "step": 15900
1859
+ },
1860
+ {
1861
+ "epoch": 46.85212298682284,
1862
+ "grad_norm": 3.3166606426239014,
1863
+ "learning_rate": 3e-05,
1864
+ "loss": 0.2397,
1865
+ "step": 16000
1866
+ },
1867
+ {
1868
+ "epoch": 46.99853587115666,
1869
+ "eval_accuracy": 0.41246153846153844,
1870
+ "eval_loss": 6.236908435821533,
1871
+ "eval_runtime": 5.1631,
1872
+ "eval_samples_per_second": 96.841,
1873
+ "eval_steps_per_second": 12.202,
1874
+ "step": 16050
1875
+ },
1876
+ {
1877
+ "epoch": 46.99853587115666,
1878
+ "eval_bleu": 0.15053700005805185,
1879
+ "eval_exact_match": 0.0,
1880
+ "eval_prefix_exact_match": 0.016,
1881
+ "step": 16050
1882
+ },
1883
+ {
1884
+ "epoch": 47.14494875549048,
1885
+ "grad_norm": 3.8332810401916504,
1886
+ "learning_rate": 3e-05,
1887
+ "loss": 0.2285,
1888
+ "step": 16100
1889
+ },
1890
+ {
1891
+ "epoch": 47.43777452415813,
1892
+ "grad_norm": 3.71746563911438,
1893
+ "learning_rate": 3e-05,
1894
+ "loss": 0.2187,
1895
+ "step": 16200
1896
+ },
1897
+ {
1898
+ "epoch": 47.73060029282577,
1899
+ "grad_norm": 3.2944347858428955,
1900
+ "learning_rate": 3e-05,
1901
+ "loss": 0.2259,
1902
+ "step": 16300
1903
+ },
1904
+ {
1905
+ "epoch": 48.0,
1906
+ "eval_accuracy": 0.4142051282051282,
1907
+ "eval_loss": 6.277541637420654,
1908
+ "eval_runtime": 5.5532,
1909
+ "eval_samples_per_second": 90.038,
1910
+ "eval_steps_per_second": 11.345,
1911
+ "step": 16392
1912
+ },
1913
+ {
1914
+ "epoch": 48.0,
1915
+ "eval_bleu": 0.13924800202556614,
1916
+ "eval_exact_match": 0.0,
1917
+ "eval_prefix_exact_match": 0.016,
1918
+ "step": 16392
1919
+ }
1920
+ ],
1921
+ "logging_steps": 100,
1922
+ "max_steps": 17050,
1923
+ "num_input_tokens_seen": 0,
1924
+ "num_train_epochs": 50,
1925
+ "save_steps": 500,
1926
+ "stateful_callbacks": {
1927
+ "TrainerControl": {
1928
+ "args": {
1929
+ "should_epoch_stop": false,
1930
+ "should_evaluate": false,
1931
+ "should_log": false,
1932
+ "should_save": true,
1933
+ "should_training_stop": false
1934
+ },
1935
+ "attributes": {}
1936
+ }
1937
+ },
1938
+ "total_flos": 1.4076073438598922e+18,
1939
+ "train_batch_size": 1,
1940
+ "trial_name": null,
1941
+ "trial_params": null
1942
+ }
checkpoint-16392/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9cef08c7f10bf396c269dfc26104fd5a94756f5431d6448312805ef8e298380
3
+ size 5304
checkpoint-16392/vocab.json ADDED
The diff for this file is too large to render. See raw diff