tyzhu commited on
Commit
0a33c46
1 Parent(s): 124acd9

Training in progress, epoch 49, checkpoint

Browse files
checkpoint-9350/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-9350/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-9350/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b60282c12cd1d29fd292170e402f6c7dcb5d6e8a043db29f6bced36fafeb3ae5
3
+ size 112130416
checkpoint-9350/added_tokens.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "<|endoftext|>": 151643,
3
+ "<|im_end|>": 151645,
4
+ "<|im_start|>": 151644
5
+ }
checkpoint-9350/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-9350/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ed8cf18d8e47e5f7c658f1770ddda1288b58766713da340efd0b11b331ef123d
3
+ size 224537202
checkpoint-9350/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d22d3c262c59ff87f08ed4ee4987dbf5bf9b67ae8a5995438c9254fe6f1a112
3
+ size 15024
checkpoint-9350/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca85f43abb1f114f84f4895a450b9cc68a67525c5b3108a5f138d96e562a50b2
3
+ size 15024
checkpoint-9350/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5499bac628f0a74071c25e20dfafe98ffb58cebdc94f50004c559e8f71914f37
3
+ size 15024
checkpoint-9350/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:968500f62466e422b3fd4f6e35a95721fed7fd09ecd3a4cbc12da2c8fba470a6
3
+ size 15024
checkpoint-9350/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5cf59ef8413c54f1c215026ca3c787ea9250e212f2702ac62638175c1f835e9e
3
+ size 1064
checkpoint-9350/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-9350/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-9350/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-9350/trainer_state.json ADDED
@@ -0,0 +1,1634 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 49.86666666666667,
5
+ "eval_steps": 500,
6
+ "global_step": 9350,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.5333333333333333,
13
+ "grad_norm": 0.275871217250824,
14
+ "learning_rate": 3e-05,
15
+ "loss": 1.8545,
16
+ "step": 100
17
+ },
18
+ {
19
+ "epoch": 0.9973333333333333,
20
+ "eval_accuracy": 0.6037399103139014,
21
+ "eval_loss": 1.7123007774353027,
22
+ "eval_runtime": 6.2976,
23
+ "eval_samples_per_second": 79.395,
24
+ "eval_steps_per_second": 10.004,
25
+ "step": 187
26
+ },
27
+ {
28
+ "epoch": 0.9973333333333333,
29
+ "eval_exact_match": 14.4,
30
+ "eval_f1": 22.843015873015876,
31
+ "eval_qa_bleu": 9.471299126615676,
32
+ "eval_qa_exact_match": 0.112,
33
+ "eval_recite_bleu": 12.267251092702137,
34
+ "eval_recite_exact_match": 0.0,
35
+ "step": 187
36
+ },
37
+ {
38
+ "epoch": 1.0666666666666667,
39
+ "grad_norm": 0.32765188813209534,
40
+ "learning_rate": 3e-05,
41
+ "loss": 1.7408,
42
+ "step": 200
43
+ },
44
+ {
45
+ "epoch": 1.6,
46
+ "grad_norm": 0.3532409369945526,
47
+ "learning_rate": 3e-05,
48
+ "loss": 1.7063,
49
+ "step": 300
50
+ },
51
+ {
52
+ "epoch": 2.0,
53
+ "eval_accuracy": 0.6058565022421525,
54
+ "eval_loss": 1.69453763961792,
55
+ "eval_runtime": 6.5928,
56
+ "eval_samples_per_second": 75.841,
57
+ "eval_steps_per_second": 9.556,
58
+ "step": 375
59
+ },
60
+ {
61
+ "epoch": 2.0,
62
+ "eval_exact_match": 14.4,
63
+ "eval_f1": 23.29731601731603,
64
+ "eval_qa_bleu": 9.4408132575913,
65
+ "eval_qa_exact_match": 0.11,
66
+ "eval_recite_bleu": 13.027794116434837,
67
+ "eval_recite_exact_match": 0.0,
68
+ "step": 375
69
+ },
70
+ {
71
+ "epoch": 2.1333333333333333,
72
+ "grad_norm": 0.3777780532836914,
73
+ "learning_rate": 3e-05,
74
+ "loss": 1.7094,
75
+ "step": 400
76
+ },
77
+ {
78
+ "epoch": 2.6666666666666665,
79
+ "grad_norm": 0.3971036374568939,
80
+ "learning_rate": 3e-05,
81
+ "loss": 1.6702,
82
+ "step": 500
83
+ },
84
+ {
85
+ "epoch": 2.997333333333333,
86
+ "eval_accuracy": 0.6065919282511211,
87
+ "eval_loss": 1.6851236820220947,
88
+ "eval_runtime": 6.3104,
89
+ "eval_samples_per_second": 79.234,
90
+ "eval_steps_per_second": 9.984,
91
+ "step": 562
92
+ },
93
+ {
94
+ "epoch": 2.997333333333333,
95
+ "eval_exact_match": 14.0,
96
+ "eval_f1": 23.306825396825417,
97
+ "eval_qa_bleu": 8.683248465366908,
98
+ "eval_qa_exact_match": 0.11,
99
+ "eval_recite_bleu": 12.39336052958371,
100
+ "eval_recite_exact_match": 0.006,
101
+ "step": 562
102
+ },
103
+ {
104
+ "epoch": 3.2,
105
+ "grad_norm": 0.4782978594303131,
106
+ "learning_rate": 3e-05,
107
+ "loss": 1.6574,
108
+ "step": 600
109
+ },
110
+ {
111
+ "epoch": 3.7333333333333334,
112
+ "grad_norm": 0.52582186460495,
113
+ "learning_rate": 3e-05,
114
+ "loss": 1.6356,
115
+ "step": 700
116
+ },
117
+ {
118
+ "epoch": 4.0,
119
+ "eval_accuracy": 0.6074977578475337,
120
+ "eval_loss": 1.682567834854126,
121
+ "eval_runtime": 6.65,
122
+ "eval_samples_per_second": 75.188,
123
+ "eval_steps_per_second": 9.474,
124
+ "step": 750
125
+ },
126
+ {
127
+ "epoch": 4.0,
128
+ "eval_exact_match": 15.0,
129
+ "eval_f1": 23.366825396825416,
130
+ "eval_qa_bleu": 9.05994637010322,
131
+ "eval_qa_exact_match": 0.114,
132
+ "eval_recite_bleu": 13.060415058384661,
133
+ "eval_recite_exact_match": 0.006,
134
+ "step": 750
135
+ },
136
+ {
137
+ "epoch": 4.266666666666667,
138
+ "grad_norm": 0.6418241262435913,
139
+ "learning_rate": 3e-05,
140
+ "loss": 1.6005,
141
+ "step": 800
142
+ },
143
+ {
144
+ "epoch": 4.8,
145
+ "grad_norm": 0.6805307269096375,
146
+ "learning_rate": 3e-05,
147
+ "loss": 1.5775,
148
+ "step": 900
149
+ },
150
+ {
151
+ "epoch": 4.997333333333334,
152
+ "eval_accuracy": 0.6071031390134529,
153
+ "eval_loss": 1.69106125831604,
154
+ "eval_runtime": 5.6598,
155
+ "eval_samples_per_second": 88.343,
156
+ "eval_steps_per_second": 11.131,
157
+ "step": 937
158
+ },
159
+ {
160
+ "epoch": 4.997333333333334,
161
+ "eval_exact_match": 15.2,
162
+ "eval_f1": 23.38539682539684,
163
+ "eval_qa_bleu": 11.854591678140883,
164
+ "eval_qa_exact_match": 0.112,
165
+ "eval_recite_bleu": 12.694062454678122,
166
+ "eval_recite_exact_match": 0.004,
167
+ "step": 937
168
+ },
169
+ {
170
+ "epoch": 5.333333333333333,
171
+ "grad_norm": 0.8362048864364624,
172
+ "learning_rate": 3e-05,
173
+ "loss": 1.5173,
174
+ "step": 1000
175
+ },
176
+ {
177
+ "epoch": 5.866666666666667,
178
+ "grad_norm": 0.8521270155906677,
179
+ "learning_rate": 3e-05,
180
+ "loss": 1.529,
181
+ "step": 1100
182
+ },
183
+ {
184
+ "epoch": 6.0,
185
+ "eval_accuracy": 0.6069417040358744,
186
+ "eval_loss": 1.7066757678985596,
187
+ "eval_runtime": 6.2922,
188
+ "eval_samples_per_second": 79.464,
189
+ "eval_steps_per_second": 10.012,
190
+ "step": 1125
191
+ },
192
+ {
193
+ "epoch": 6.0,
194
+ "eval_exact_match": 15.2,
195
+ "eval_f1": 24.04682539682541,
196
+ "eval_qa_bleu": 12.150293484705452,
197
+ "eval_qa_exact_match": 0.106,
198
+ "eval_recite_bleu": 12.732083161926754,
199
+ "eval_recite_exact_match": 0.002,
200
+ "step": 1125
201
+ },
202
+ {
203
+ "epoch": 6.4,
204
+ "grad_norm": 0.9372308254241943,
205
+ "learning_rate": 3e-05,
206
+ "loss": 1.4613,
207
+ "step": 1200
208
+ },
209
+ {
210
+ "epoch": 6.933333333333334,
211
+ "grad_norm": 1.0236693620681763,
212
+ "learning_rate": 3e-05,
213
+ "loss": 1.457,
214
+ "step": 1300
215
+ },
216
+ {
217
+ "epoch": 6.997333333333334,
218
+ "eval_accuracy": 0.6055515695067265,
219
+ "eval_loss": 1.72792649269104,
220
+ "eval_runtime": 6.3278,
221
+ "eval_samples_per_second": 79.016,
222
+ "eval_steps_per_second": 9.956,
223
+ "step": 1312
224
+ },
225
+ {
226
+ "epoch": 6.997333333333334,
227
+ "eval_exact_match": 15.0,
228
+ "eval_f1": 23.28492063492065,
229
+ "eval_qa_bleu": 9.849658422779212,
230
+ "eval_qa_exact_match": 0.114,
231
+ "eval_recite_bleu": 12.3997433702838,
232
+ "eval_recite_exact_match": 0.006,
233
+ "step": 1312
234
+ },
235
+ {
236
+ "epoch": 7.466666666666667,
237
+ "grad_norm": 1.0773049592971802,
238
+ "learning_rate": 3e-05,
239
+ "loss": 1.3996,
240
+ "step": 1400
241
+ },
242
+ {
243
+ "epoch": 8.0,
244
+ "grad_norm": 1.0705862045288086,
245
+ "learning_rate": 3e-05,
246
+ "loss": 1.3907,
247
+ "step": 1500
248
+ },
249
+ {
250
+ "epoch": 8.0,
251
+ "eval_accuracy": 0.6046188340807175,
252
+ "eval_loss": 1.7511730194091797,
253
+ "eval_runtime": 5.5895,
254
+ "eval_samples_per_second": 89.453,
255
+ "eval_steps_per_second": 11.271,
256
+ "step": 1500
257
+ },
258
+ {
259
+ "epoch": 8.0,
260
+ "eval_exact_match": 13.0,
261
+ "eval_f1": 19.93238095238096,
262
+ "eval_qa_bleu": 11.93942669425287,
263
+ "eval_qa_exact_match": 0.098,
264
+ "eval_recite_bleu": 11.949414001720045,
265
+ "eval_recite_exact_match": 0.006,
266
+ "step": 1500
267
+ },
268
+ {
269
+ "epoch": 8.533333333333333,
270
+ "grad_norm": 1.2424529790878296,
271
+ "learning_rate": 3e-05,
272
+ "loss": 1.3309,
273
+ "step": 1600
274
+ },
275
+ {
276
+ "epoch": 8.997333333333334,
277
+ "eval_accuracy": 0.6025022421524664,
278
+ "eval_loss": 1.7774019241333008,
279
+ "eval_runtime": 6.4013,
280
+ "eval_samples_per_second": 78.109,
281
+ "eval_steps_per_second": 9.842,
282
+ "step": 1687
283
+ },
284
+ {
285
+ "epoch": 8.997333333333334,
286
+ "eval_exact_match": 13.6,
287
+ "eval_f1": 21.089523809523815,
288
+ "eval_qa_bleu": 10.094627644241111,
289
+ "eval_qa_exact_match": 0.094,
290
+ "eval_recite_bleu": 12.197681191541987,
291
+ "eval_recite_exact_match": 0.006,
292
+ "step": 1687
293
+ },
294
+ {
295
+ "epoch": 9.066666666666666,
296
+ "grad_norm": 1.535229206085205,
297
+ "learning_rate": 3e-05,
298
+ "loss": 1.3336,
299
+ "step": 1700
300
+ },
301
+ {
302
+ "epoch": 9.6,
303
+ "grad_norm": 1.4936225414276123,
304
+ "learning_rate": 3e-05,
305
+ "loss": 1.2841,
306
+ "step": 1800
307
+ },
308
+ {
309
+ "epoch": 10.0,
310
+ "eval_accuracy": 0.6012645739910314,
311
+ "eval_loss": 1.804250717163086,
312
+ "eval_runtime": 6.385,
313
+ "eval_samples_per_second": 78.309,
314
+ "eval_steps_per_second": 9.867,
315
+ "step": 1875
316
+ },
317
+ {
318
+ "epoch": 10.0,
319
+ "eval_exact_match": 13.2,
320
+ "eval_f1": 21.172832722832737,
321
+ "eval_qa_bleu": 9.841075001407239,
322
+ "eval_qa_exact_match": 0.096,
323
+ "eval_recite_bleu": 11.691398764835105,
324
+ "eval_recite_exact_match": 0.004,
325
+ "step": 1875
326
+ },
327
+ {
328
+ "epoch": 10.133333333333333,
329
+ "grad_norm": 1.6687729358673096,
330
+ "learning_rate": 3e-05,
331
+ "loss": 1.2562,
332
+ "step": 1900
333
+ },
334
+ {
335
+ "epoch": 10.666666666666666,
336
+ "grad_norm": 1.437538504600525,
337
+ "learning_rate": 3e-05,
338
+ "loss": 1.2308,
339
+ "step": 2000
340
+ },
341
+ {
342
+ "epoch": 10.997333333333334,
343
+ "eval_accuracy": 0.6000717488789238,
344
+ "eval_loss": 1.8528188467025757,
345
+ "eval_runtime": 6.6644,
346
+ "eval_samples_per_second": 75.026,
347
+ "eval_steps_per_second": 9.453,
348
+ "step": 2062
349
+ },
350
+ {
351
+ "epoch": 10.997333333333334,
352
+ "eval_exact_match": 12.6,
353
+ "eval_f1": 20.865213675213692,
354
+ "eval_qa_bleu": 7.784433449171618,
355
+ "eval_qa_exact_match": 0.108,
356
+ "eval_recite_bleu": 12.305413068460787,
357
+ "eval_recite_exact_match": 0.008,
358
+ "step": 2062
359
+ },
360
+ {
361
+ "epoch": 11.2,
362
+ "grad_norm": 1.708260416984558,
363
+ "learning_rate": 3e-05,
364
+ "loss": 1.2011,
365
+ "step": 2100
366
+ },
367
+ {
368
+ "epoch": 11.733333333333333,
369
+ "grad_norm": 1.7026830911636353,
370
+ "learning_rate": 3e-05,
371
+ "loss": 1.1722,
372
+ "step": 2200
373
+ },
374
+ {
375
+ "epoch": 12.0,
376
+ "eval_accuracy": 0.5987892376681614,
377
+ "eval_loss": 1.8851206302642822,
378
+ "eval_runtime": 5.5764,
379
+ "eval_samples_per_second": 89.663,
380
+ "eval_steps_per_second": 11.298,
381
+ "step": 2250
382
+ },
383
+ {
384
+ "epoch": 12.0,
385
+ "eval_exact_match": 13.0,
386
+ "eval_f1": 21.17809523809525,
387
+ "eval_qa_bleu": 11.19937598521782,
388
+ "eval_qa_exact_match": 0.096,
389
+ "eval_recite_bleu": 12.500534642121808,
390
+ "eval_recite_exact_match": 0.01,
391
+ "step": 2250
392
+ },
393
+ {
394
+ "epoch": 12.266666666666667,
395
+ "grad_norm": 1.6281794309616089,
396
+ "learning_rate": 3e-05,
397
+ "loss": 1.151,
398
+ "step": 2300
399
+ },
400
+ {
401
+ "epoch": 12.8,
402
+ "grad_norm": 1.686490535736084,
403
+ "learning_rate": 3e-05,
404
+ "loss": 1.1354,
405
+ "step": 2400
406
+ },
407
+ {
408
+ "epoch": 12.997333333333334,
409
+ "eval_accuracy": 0.5979910313901345,
410
+ "eval_loss": 1.9114351272583008,
411
+ "eval_runtime": 5.5718,
412
+ "eval_samples_per_second": 89.737,
413
+ "eval_steps_per_second": 11.307,
414
+ "step": 2437
415
+ },
416
+ {
417
+ "epoch": 12.997333333333334,
418
+ "eval_exact_match": 13.8,
419
+ "eval_f1": 21.856666666666673,
420
+ "eval_qa_bleu": 10.206087008658956,
421
+ "eval_qa_exact_match": 0.116,
422
+ "eval_recite_bleu": 12.669262328097592,
423
+ "eval_recite_exact_match": 0.008,
424
+ "step": 2437
425
+ },
426
+ {
427
+ "epoch": 13.333333333333334,
428
+ "grad_norm": 1.8666125535964966,
429
+ "learning_rate": 3e-05,
430
+ "loss": 1.097,
431
+ "step": 2500
432
+ },
433
+ {
434
+ "epoch": 13.866666666666667,
435
+ "grad_norm": 1.9091932773590088,
436
+ "learning_rate": 3e-05,
437
+ "loss": 1.0793,
438
+ "step": 2600
439
+ },
440
+ {
441
+ "epoch": 14.0,
442
+ "eval_accuracy": 0.5960986547085202,
443
+ "eval_loss": 1.9584920406341553,
444
+ "eval_runtime": 6.3056,
445
+ "eval_samples_per_second": 79.294,
446
+ "eval_steps_per_second": 9.991,
447
+ "step": 2625
448
+ },
449
+ {
450
+ "epoch": 14.0,
451
+ "eval_exact_match": 10.6,
452
+ "eval_f1": 18.697936507936518,
453
+ "eval_qa_bleu": 6.4581555766931436,
454
+ "eval_qa_exact_match": 0.082,
455
+ "eval_recite_bleu": 12.843526576123885,
456
+ "eval_recite_exact_match": 0.01,
457
+ "step": 2625
458
+ },
459
+ {
460
+ "epoch": 14.4,
461
+ "grad_norm": 2.042712688446045,
462
+ "learning_rate": 3e-05,
463
+ "loss": 1.0376,
464
+ "step": 2700
465
+ },
466
+ {
467
+ "epoch": 14.933333333333334,
468
+ "grad_norm": 1.8875784873962402,
469
+ "learning_rate": 3e-05,
470
+ "loss": 1.037,
471
+ "step": 2800
472
+ },
473
+ {
474
+ "epoch": 14.997333333333334,
475
+ "eval_accuracy": 0.5948161434977578,
476
+ "eval_loss": 1.9966908693313599,
477
+ "eval_runtime": 6.0322,
478
+ "eval_samples_per_second": 82.888,
479
+ "eval_steps_per_second": 10.444,
480
+ "step": 2812
481
+ },
482
+ {
483
+ "epoch": 14.997333333333334,
484
+ "eval_exact_match": 12.8,
485
+ "eval_f1": 20.438917748917753,
486
+ "eval_qa_bleu": 10.19364322169534,
487
+ "eval_qa_exact_match": 0.104,
488
+ "eval_recite_bleu": 12.69742056629334,
489
+ "eval_recite_exact_match": 0.012,
490
+ "step": 2812
491
+ },
492
+ {
493
+ "epoch": 15.466666666666667,
494
+ "grad_norm": 2.027426242828369,
495
+ "learning_rate": 3e-05,
496
+ "loss": 0.9844,
497
+ "step": 2900
498
+ },
499
+ {
500
+ "epoch": 16.0,
501
+ "grad_norm": 2.0708699226379395,
502
+ "learning_rate": 3e-05,
503
+ "loss": 0.9901,
504
+ "step": 3000
505
+ },
506
+ {
507
+ "epoch": 16.0,
508
+ "eval_accuracy": 0.5934080717488789,
509
+ "eval_loss": 2.0336477756500244,
510
+ "eval_runtime": 5.9811,
511
+ "eval_samples_per_second": 83.597,
512
+ "eval_steps_per_second": 10.533,
513
+ "step": 3000
514
+ },
515
+ {
516
+ "epoch": 16.0,
517
+ "eval_exact_match": 12.4,
518
+ "eval_f1": 20.427142857142865,
519
+ "eval_qa_bleu": 9.13217752730269,
520
+ "eval_qa_exact_match": 0.096,
521
+ "eval_recite_bleu": 12.688250071816965,
522
+ "eval_recite_exact_match": 0.014,
523
+ "step": 3000
524
+ },
525
+ {
526
+ "epoch": 16.533333333333335,
527
+ "grad_norm": 1.995875358581543,
528
+ "learning_rate": 3e-05,
529
+ "loss": 0.9316,
530
+ "step": 3100
531
+ },
532
+ {
533
+ "epoch": 16.997333333333334,
534
+ "eval_accuracy": 0.591372197309417,
535
+ "eval_loss": 2.088000535964966,
536
+ "eval_runtime": 5.9631,
537
+ "eval_samples_per_second": 83.849,
538
+ "eval_steps_per_second": 10.565,
539
+ "step": 3187
540
+ },
541
+ {
542
+ "epoch": 16.997333333333334,
543
+ "eval_exact_match": 11.4,
544
+ "eval_f1": 18.528571428571443,
545
+ "eval_qa_bleu": 7.115393148941145,
546
+ "eval_qa_exact_match": 0.084,
547
+ "eval_recite_bleu": 12.219544702526637,
548
+ "eval_recite_exact_match": 0.014,
549
+ "step": 3187
550
+ },
551
+ {
552
+ "epoch": 17.066666666666666,
553
+ "grad_norm": 2.623262643814087,
554
+ "learning_rate": 3e-05,
555
+ "loss": 0.9354,
556
+ "step": 3200
557
+ },
558
+ {
559
+ "epoch": 17.6,
560
+ "grad_norm": 2.4125397205352783,
561
+ "learning_rate": 3e-05,
562
+ "loss": 0.8802,
563
+ "step": 3300
564
+ },
565
+ {
566
+ "epoch": 18.0,
567
+ "eval_accuracy": 0.5900627802690583,
568
+ "eval_loss": 2.1439507007598877,
569
+ "eval_runtime": 5.8546,
570
+ "eval_samples_per_second": 85.404,
571
+ "eval_steps_per_second": 10.761,
572
+ "step": 3375
573
+ },
574
+ {
575
+ "epoch": 18.0,
576
+ "eval_exact_match": 10.8,
577
+ "eval_f1": 17.508095238095244,
578
+ "eval_qa_bleu": 7.769457934505562,
579
+ "eval_qa_exact_match": 0.09,
580
+ "eval_recite_bleu": 12.155079189562132,
581
+ "eval_recite_exact_match": 0.016,
582
+ "step": 3375
583
+ },
584
+ {
585
+ "epoch": 18.133333333333333,
586
+ "grad_norm": 2.3584065437316895,
587
+ "learning_rate": 3e-05,
588
+ "loss": 0.8913,
589
+ "step": 3400
590
+ },
591
+ {
592
+ "epoch": 18.666666666666668,
593
+ "grad_norm": 2.130357265472412,
594
+ "learning_rate": 3e-05,
595
+ "loss": 0.8382,
596
+ "step": 3500
597
+ },
598
+ {
599
+ "epoch": 18.997333333333334,
600
+ "eval_accuracy": 0.5892645739910314,
601
+ "eval_loss": 2.171532154083252,
602
+ "eval_runtime": 6.6257,
603
+ "eval_samples_per_second": 75.464,
604
+ "eval_steps_per_second": 9.508,
605
+ "step": 3562
606
+ },
607
+ {
608
+ "epoch": 18.997333333333334,
609
+ "eval_exact_match": 13.0,
610
+ "eval_f1": 18.838095238095242,
611
+ "eval_qa_bleu": 7.682885911703284,
612
+ "eval_qa_exact_match": 0.106,
613
+ "eval_recite_bleu": 12.350855746597011,
614
+ "eval_recite_exact_match": 0.014,
615
+ "step": 3562
616
+ },
617
+ {
618
+ "epoch": 19.2,
619
+ "grad_norm": 2.6397740840911865,
620
+ "learning_rate": 3e-05,
621
+ "loss": 0.8248,
622
+ "step": 3600
623
+ },
624
+ {
625
+ "epoch": 19.733333333333334,
626
+ "grad_norm": 3.049431085586548,
627
+ "learning_rate": 3e-05,
628
+ "loss": 0.7962,
629
+ "step": 3700
630
+ },
631
+ {
632
+ "epoch": 20.0,
633
+ "eval_accuracy": 0.5878834080717489,
634
+ "eval_loss": 2.2236711978912354,
635
+ "eval_runtime": 6.764,
636
+ "eval_samples_per_second": 73.921,
637
+ "eval_steps_per_second": 9.314,
638
+ "step": 3750
639
+ },
640
+ {
641
+ "epoch": 20.0,
642
+ "eval_exact_match": 12.2,
643
+ "eval_f1": 18.373809523809534,
644
+ "eval_qa_bleu": 8.448974659181921,
645
+ "eval_qa_exact_match": 0.096,
646
+ "eval_recite_bleu": 11.298504528713615,
647
+ "eval_recite_exact_match": 0.014,
648
+ "step": 3750
649
+ },
650
+ {
651
+ "epoch": 20.266666666666666,
652
+ "grad_norm": 2.647123098373413,
653
+ "learning_rate": 3e-05,
654
+ "loss": 0.7797,
655
+ "step": 3800
656
+ },
657
+ {
658
+ "epoch": 20.8,
659
+ "grad_norm": 2.798877239227295,
660
+ "learning_rate": 3e-05,
661
+ "loss": 0.7553,
662
+ "step": 3900
663
+ },
664
+ {
665
+ "epoch": 20.997333333333334,
666
+ "eval_accuracy": 0.5860896860986547,
667
+ "eval_loss": 2.295654535293579,
668
+ "eval_runtime": 5.5847,
669
+ "eval_samples_per_second": 89.531,
670
+ "eval_steps_per_second": 11.281,
671
+ "step": 3937
672
+ },
673
+ {
674
+ "epoch": 20.997333333333334,
675
+ "eval_exact_match": 10.2,
676
+ "eval_f1": 16.601428571428578,
677
+ "eval_qa_bleu": 7.90021052694059,
678
+ "eval_qa_exact_match": 0.08,
679
+ "eval_recite_bleu": 11.716166611151118,
680
+ "eval_recite_exact_match": 0.01,
681
+ "step": 3937
682
+ },
683
+ {
684
+ "epoch": 21.333333333333332,
685
+ "grad_norm": 2.7208638191223145,
686
+ "learning_rate": 3e-05,
687
+ "loss": 0.7306,
688
+ "step": 4000
689
+ },
690
+ {
691
+ "epoch": 21.866666666666667,
692
+ "grad_norm": 2.9887781143188477,
693
+ "learning_rate": 3e-05,
694
+ "loss": 0.7238,
695
+ "step": 4100
696
+ },
697
+ {
698
+ "epoch": 22.0,
699
+ "eval_accuracy": 0.5850941704035875,
700
+ "eval_loss": 2.331202507019043,
701
+ "eval_runtime": 6.6871,
702
+ "eval_samples_per_second": 74.771,
703
+ "eval_steps_per_second": 9.421,
704
+ "step": 4125
705
+ },
706
+ {
707
+ "epoch": 22.0,
708
+ "eval_exact_match": 10.8,
709
+ "eval_f1": 17.400692640692647,
710
+ "eval_qa_bleu": 4.77963863209427,
711
+ "eval_qa_exact_match": 0.084,
712
+ "eval_recite_bleu": 11.404518212072734,
713
+ "eval_recite_exact_match": 0.014,
714
+ "step": 4125
715
+ },
716
+ {
717
+ "epoch": 22.4,
718
+ "grad_norm": 2.706368923187256,
719
+ "learning_rate": 3e-05,
720
+ "loss": 0.6813,
721
+ "step": 4200
722
+ },
723
+ {
724
+ "epoch": 22.933333333333334,
725
+ "grad_norm": 2.850799798965454,
726
+ "learning_rate": 3e-05,
727
+ "loss": 0.676,
728
+ "step": 4300
729
+ },
730
+ {
731
+ "epoch": 22.997333333333334,
732
+ "eval_accuracy": 0.583237668161435,
733
+ "eval_loss": 2.4042670726776123,
734
+ "eval_runtime": 6.5747,
735
+ "eval_samples_per_second": 76.049,
736
+ "eval_steps_per_second": 9.582,
737
+ "step": 4312
738
+ },
739
+ {
740
+ "epoch": 22.997333333333334,
741
+ "eval_exact_match": 8.6,
742
+ "eval_f1": 16.314285714285724,
743
+ "eval_qa_bleu": 7.466573197185606,
744
+ "eval_qa_exact_match": 0.064,
745
+ "eval_recite_bleu": 11.568247151151132,
746
+ "eval_recite_exact_match": 0.014,
747
+ "step": 4312
748
+ },
749
+ {
750
+ "epoch": 23.466666666666665,
751
+ "grad_norm": 2.95532488822937,
752
+ "learning_rate": 3e-05,
753
+ "loss": 0.632,
754
+ "step": 4400
755
+ },
756
+ {
757
+ "epoch": 24.0,
758
+ "grad_norm": 3.36663556098938,
759
+ "learning_rate": 3e-05,
760
+ "loss": 0.644,
761
+ "step": 4500
762
+ },
763
+ {
764
+ "epoch": 24.0,
765
+ "eval_accuracy": 0.5824125560538117,
766
+ "eval_loss": 2.443950891494751,
767
+ "eval_runtime": 5.7069,
768
+ "eval_samples_per_second": 87.614,
769
+ "eval_steps_per_second": 11.039,
770
+ "step": 4500
771
+ },
772
+ {
773
+ "epoch": 24.0,
774
+ "eval_exact_match": 8.8,
775
+ "eval_f1": 15.491948051948057,
776
+ "eval_qa_bleu": 4.531789460915551,
777
+ "eval_qa_exact_match": 0.076,
778
+ "eval_recite_bleu": 10.65729884465588,
779
+ "eval_recite_exact_match": 0.014,
780
+ "step": 4500
781
+ },
782
+ {
783
+ "epoch": 24.533333333333335,
784
+ "grad_norm": 3.074622631072998,
785
+ "learning_rate": 3e-05,
786
+ "loss": 0.5939,
787
+ "step": 4600
788
+ },
789
+ {
790
+ "epoch": 24.997333333333334,
791
+ "eval_accuracy": 0.5818385650224215,
792
+ "eval_loss": 2.5127182006835938,
793
+ "eval_runtime": 5.6588,
794
+ "eval_samples_per_second": 88.357,
795
+ "eval_steps_per_second": 11.133,
796
+ "step": 4687
797
+ },
798
+ {
799
+ "epoch": 24.997333333333334,
800
+ "eval_exact_match": 9.0,
801
+ "eval_f1": 16.361587301587303,
802
+ "eval_qa_bleu": 7.691728089554307,
803
+ "eval_qa_exact_match": 0.068,
804
+ "eval_recite_bleu": 11.016738678559781,
805
+ "eval_recite_exact_match": 0.008,
806
+ "step": 4687
807
+ },
808
+ {
809
+ "epoch": 25.066666666666666,
810
+ "grad_norm": 2.702502727508545,
811
+ "learning_rate": 3e-05,
812
+ "loss": 0.5904,
813
+ "step": 4700
814
+ },
815
+ {
816
+ "epoch": 25.6,
817
+ "grad_norm": 3.233320951461792,
818
+ "learning_rate": 3e-05,
819
+ "loss": 0.5551,
820
+ "step": 4800
821
+ },
822
+ {
823
+ "epoch": 26.0,
824
+ "eval_accuracy": 0.5809596412556054,
825
+ "eval_loss": 2.538985252380371,
826
+ "eval_runtime": 6.6838,
827
+ "eval_samples_per_second": 74.808,
828
+ "eval_steps_per_second": 9.426,
829
+ "step": 4875
830
+ },
831
+ {
832
+ "epoch": 26.0,
833
+ "eval_exact_match": 9.2,
834
+ "eval_f1": 16.819884559884564,
835
+ "eval_qa_bleu": 4.658892987257927,
836
+ "eval_qa_exact_match": 0.078,
837
+ "eval_recite_bleu": 10.853797139069188,
838
+ "eval_recite_exact_match": 0.012,
839
+ "step": 4875
840
+ },
841
+ {
842
+ "epoch": 26.133333333333333,
843
+ "grad_norm": 3.280097246170044,
844
+ "learning_rate": 3e-05,
845
+ "loss": 0.5503,
846
+ "step": 4900
847
+ },
848
+ {
849
+ "epoch": 26.666666666666668,
850
+ "grad_norm": 3.1984827518463135,
851
+ "learning_rate": 3e-05,
852
+ "loss": 0.5163,
853
+ "step": 5000
854
+ },
855
+ {
856
+ "epoch": 26.997333333333334,
857
+ "eval_accuracy": 0.579829596412556,
858
+ "eval_loss": 2.5808944702148438,
859
+ "eval_runtime": 6.6339,
860
+ "eval_samples_per_second": 75.37,
861
+ "eval_steps_per_second": 9.497,
862
+ "step": 5062
863
+ },
864
+ {
865
+ "epoch": 26.997333333333334,
866
+ "eval_exact_match": 8.0,
867
+ "eval_f1": 16.641197691197693,
868
+ "eval_qa_bleu": 5.575351494624609,
869
+ "eval_qa_exact_match": 0.066,
870
+ "eval_recite_bleu": 11.099361911324253,
871
+ "eval_recite_exact_match": 0.01,
872
+ "step": 5062
873
+ },
874
+ {
875
+ "epoch": 27.2,
876
+ "grad_norm": 3.5599937438964844,
877
+ "learning_rate": 3e-05,
878
+ "loss": 0.52,
879
+ "step": 5100
880
+ },
881
+ {
882
+ "epoch": 27.733333333333334,
883
+ "grad_norm": 2.9382145404815674,
884
+ "learning_rate": 3e-05,
885
+ "loss": 0.4892,
886
+ "step": 5200
887
+ },
888
+ {
889
+ "epoch": 28.0,
890
+ "eval_accuracy": 0.5788609865470852,
891
+ "eval_loss": 2.6670191287994385,
892
+ "eval_runtime": 6.0585,
893
+ "eval_samples_per_second": 82.529,
894
+ "eval_steps_per_second": 10.399,
895
+ "step": 5250
896
+ },
897
+ {
898
+ "epoch": 28.0,
899
+ "eval_exact_match": 8.6,
900
+ "eval_f1": 15.144920634920638,
901
+ "eval_qa_bleu": 5.387088284739065,
902
+ "eval_qa_exact_match": 0.074,
903
+ "eval_recite_bleu": 11.457740165981269,
904
+ "eval_recite_exact_match": 0.012,
905
+ "step": 5250
906
+ },
907
+ {
908
+ "epoch": 28.266666666666666,
909
+ "grad_norm": 4.097602367401123,
910
+ "learning_rate": 3e-05,
911
+ "loss": 0.47,
912
+ "step": 5300
913
+ },
914
+ {
915
+ "epoch": 28.8,
916
+ "grad_norm": 3.9295685291290283,
917
+ "learning_rate": 3e-05,
918
+ "loss": 0.4669,
919
+ "step": 5400
920
+ },
921
+ {
922
+ "epoch": 28.997333333333334,
923
+ "eval_accuracy": 0.578609865470852,
924
+ "eval_loss": 2.669480085372925,
925
+ "eval_runtime": 6.6992,
926
+ "eval_samples_per_second": 74.636,
927
+ "eval_steps_per_second": 9.404,
928
+ "step": 5437
929
+ },
930
+ {
931
+ "epoch": 28.997333333333334,
932
+ "eval_exact_match": 9.2,
933
+ "eval_f1": 16.65718614718615,
934
+ "eval_qa_bleu": 4.637892076000507,
935
+ "eval_qa_exact_match": 0.074,
936
+ "eval_recite_bleu": 11.77754451628722,
937
+ "eval_recite_exact_match": 0.012,
938
+ "step": 5437
939
+ },
940
+ {
941
+ "epoch": 29.333333333333332,
942
+ "grad_norm": 3.4413976669311523,
943
+ "learning_rate": 3e-05,
944
+ "loss": 0.4398,
945
+ "step": 5500
946
+ },
947
+ {
948
+ "epoch": 29.866666666666667,
949
+ "grad_norm": 2.88114857673645,
950
+ "learning_rate": 3e-05,
951
+ "loss": 0.4353,
952
+ "step": 5600
953
+ },
954
+ {
955
+ "epoch": 30.0,
956
+ "eval_accuracy": 0.5786995515695067,
957
+ "eval_loss": 2.7645530700683594,
958
+ "eval_runtime": 6.3214,
959
+ "eval_samples_per_second": 79.096,
960
+ "eval_steps_per_second": 9.966,
961
+ "step": 5625
962
+ },
963
+ {
964
+ "epoch": 30.0,
965
+ "eval_exact_match": 8.0,
966
+ "eval_f1": 14.828455988455993,
967
+ "eval_qa_bleu": 7.042137393349148,
968
+ "eval_qa_exact_match": 0.068,
969
+ "eval_recite_bleu": 11.413791110087148,
970
+ "eval_recite_exact_match": 0.01,
971
+ "step": 5625
972
+ },
973
+ {
974
+ "epoch": 30.4,
975
+ "grad_norm": 4.096495628356934,
976
+ "learning_rate": 3e-05,
977
+ "loss": 0.4032,
978
+ "step": 5700
979
+ },
980
+ {
981
+ "epoch": 30.933333333333334,
982
+ "grad_norm": 3.567011833190918,
983
+ "learning_rate": 3e-05,
984
+ "loss": 0.4104,
985
+ "step": 5800
986
+ },
987
+ {
988
+ "epoch": 30.997333333333334,
989
+ "eval_accuracy": 0.5774887892376681,
990
+ "eval_loss": 2.829072952270508,
991
+ "eval_runtime": 6.655,
992
+ "eval_samples_per_second": 75.131,
993
+ "eval_steps_per_second": 9.467,
994
+ "step": 5812
995
+ },
996
+ {
997
+ "epoch": 30.997333333333334,
998
+ "eval_exact_match": 8.8,
999
+ "eval_f1": 14.71683982683983,
1000
+ "eval_qa_bleu": 7.548806097282496,
1001
+ "eval_qa_exact_match": 0.072,
1002
+ "eval_recite_bleu": 10.549616368516189,
1003
+ "eval_recite_exact_match": 0.01,
1004
+ "step": 5812
1005
+ },
1006
+ {
1007
+ "epoch": 31.466666666666665,
1008
+ "grad_norm": 3.1184980869293213,
1009
+ "learning_rate": 3e-05,
1010
+ "loss": 0.3695,
1011
+ "step": 5900
1012
+ },
1013
+ {
1014
+ "epoch": 32.0,
1015
+ "grad_norm": 3.125758647918701,
1016
+ "learning_rate": 3e-05,
1017
+ "loss": 0.3885,
1018
+ "step": 6000
1019
+ },
1020
+ {
1021
+ "epoch": 32.0,
1022
+ "eval_accuracy": 0.5763587443946189,
1023
+ "eval_loss": 2.893293857574463,
1024
+ "eval_runtime": 5.9693,
1025
+ "eval_samples_per_second": 83.762,
1026
+ "eval_steps_per_second": 10.554,
1027
+ "step": 6000
1028
+ },
1029
+ {
1030
+ "epoch": 32.0,
1031
+ "eval_exact_match": 8.6,
1032
+ "eval_f1": 15.2548051948052,
1033
+ "eval_qa_bleu": 5.915179328505316,
1034
+ "eval_qa_exact_match": 0.072,
1035
+ "eval_recite_bleu": 10.647049650020403,
1036
+ "eval_recite_exact_match": 0.008,
1037
+ "step": 6000
1038
+ },
1039
+ {
1040
+ "epoch": 32.53333333333333,
1041
+ "grad_norm": 2.6737849712371826,
1042
+ "learning_rate": 3e-05,
1043
+ "loss": 0.342,
1044
+ "step": 6100
1045
+ },
1046
+ {
1047
+ "epoch": 32.99733333333333,
1048
+ "eval_accuracy": 0.5755964125560538,
1049
+ "eval_loss": 2.9434149265289307,
1050
+ "eval_runtime": 6.0396,
1051
+ "eval_samples_per_second": 82.787,
1052
+ "eval_steps_per_second": 10.431,
1053
+ "step": 6187
1054
+ },
1055
+ {
1056
+ "epoch": 32.99733333333333,
1057
+ "eval_exact_match": 8.2,
1058
+ "eval_f1": 14.65269841269841,
1059
+ "eval_qa_bleu": 8.616100659809678,
1060
+ "eval_qa_exact_match": 0.068,
1061
+ "eval_recite_bleu": 11.341262248022645,
1062
+ "eval_recite_exact_match": 0.014,
1063
+ "step": 6187
1064
+ },
1065
+ {
1066
+ "epoch": 33.06666666666667,
1067
+ "grad_norm": 2.490464448928833,
1068
+ "learning_rate": 3e-05,
1069
+ "loss": 0.3528,
1070
+ "step": 6200
1071
+ },
1072
+ {
1073
+ "epoch": 33.6,
1074
+ "grad_norm": 3.0587589740753174,
1075
+ "learning_rate": 3e-05,
1076
+ "loss": 0.3213,
1077
+ "step": 6300
1078
+ },
1079
+ {
1080
+ "epoch": 34.0,
1081
+ "eval_accuracy": 0.5755874439461883,
1082
+ "eval_loss": 2.934598207473755,
1083
+ "eval_runtime": 5.9452,
1084
+ "eval_samples_per_second": 84.101,
1085
+ "eval_steps_per_second": 10.597,
1086
+ "step": 6375
1087
+ },
1088
+ {
1089
+ "epoch": 34.0,
1090
+ "eval_exact_match": 8.2,
1091
+ "eval_f1": 13.88111111111111,
1092
+ "eval_qa_bleu": 8.221633718322723,
1093
+ "eval_qa_exact_match": 0.066,
1094
+ "eval_recite_bleu": 11.218940899233823,
1095
+ "eval_recite_exact_match": 0.018,
1096
+ "step": 6375
1097
+ },
1098
+ {
1099
+ "epoch": 34.13333333333333,
1100
+ "grad_norm": 3.1941940784454346,
1101
+ "learning_rate": 3e-05,
1102
+ "loss": 0.328,
1103
+ "step": 6400
1104
+ },
1105
+ {
1106
+ "epoch": 34.666666666666664,
1107
+ "grad_norm": 3.2870864868164062,
1108
+ "learning_rate": 3e-05,
1109
+ "loss": 0.3065,
1110
+ "step": 6500
1111
+ },
1112
+ {
1113
+ "epoch": 34.99733333333333,
1114
+ "eval_accuracy": 0.5757757847533632,
1115
+ "eval_loss": 3.008178949356079,
1116
+ "eval_runtime": 6.5953,
1117
+ "eval_samples_per_second": 75.811,
1118
+ "eval_steps_per_second": 9.552,
1119
+ "step": 6562
1120
+ },
1121
+ {
1122
+ "epoch": 34.99733333333333,
1123
+ "eval_exact_match": 9.6,
1124
+ "eval_f1": 15.528095238095238,
1125
+ "eval_qa_bleu": 6.040085184981805,
1126
+ "eval_qa_exact_match": 0.066,
1127
+ "eval_recite_bleu": 11.118518942785343,
1128
+ "eval_recite_exact_match": 0.016,
1129
+ "step": 6562
1130
+ },
1131
+ {
1132
+ "epoch": 35.2,
1133
+ "grad_norm": 3.2354586124420166,
1134
+ "learning_rate": 3e-05,
1135
+ "loss": 0.3006,
1136
+ "step": 6600
1137
+ },
1138
+ {
1139
+ "epoch": 35.733333333333334,
1140
+ "grad_norm": 3.262101173400879,
1141
+ "learning_rate": 3e-05,
1142
+ "loss": 0.2842,
1143
+ "step": 6700
1144
+ },
1145
+ {
1146
+ "epoch": 36.0,
1147
+ "eval_accuracy": 0.5739013452914798,
1148
+ "eval_loss": 3.0947136878967285,
1149
+ "eval_runtime": 6.5534,
1150
+ "eval_samples_per_second": 76.297,
1151
+ "eval_steps_per_second": 9.613,
1152
+ "step": 6750
1153
+ },
1154
+ {
1155
+ "epoch": 36.0,
1156
+ "eval_exact_match": 10.0,
1157
+ "eval_f1": 17.241428571428575,
1158
+ "eval_qa_bleu": 8.55409762541613,
1159
+ "eval_qa_exact_match": 0.08,
1160
+ "eval_recite_bleu": 11.315935831255379,
1161
+ "eval_recite_exact_match": 0.014,
1162
+ "step": 6750
1163
+ },
1164
+ {
1165
+ "epoch": 36.266666666666666,
1166
+ "grad_norm": 2.951231002807617,
1167
+ "learning_rate": 3e-05,
1168
+ "loss": 0.2794,
1169
+ "step": 6800
1170
+ },
1171
+ {
1172
+ "epoch": 36.8,
1173
+ "grad_norm": 3.770897626876831,
1174
+ "learning_rate": 3e-05,
1175
+ "loss": 0.2695,
1176
+ "step": 6900
1177
+ },
1178
+ {
1179
+ "epoch": 36.99733333333333,
1180
+ "eval_accuracy": 0.5751569506726457,
1181
+ "eval_loss": 3.0905086994171143,
1182
+ "eval_runtime": 5.6522,
1183
+ "eval_samples_per_second": 88.46,
1184
+ "eval_steps_per_second": 11.146,
1185
+ "step": 6937
1186
+ },
1187
+ {
1188
+ "epoch": 36.99733333333333,
1189
+ "eval_exact_match": 7.0,
1190
+ "eval_f1": 14.115916305916304,
1191
+ "eval_qa_bleu": 7.552037183523391,
1192
+ "eval_qa_exact_match": 0.054,
1193
+ "eval_recite_bleu": 10.863808387375153,
1194
+ "eval_recite_exact_match": 0.01,
1195
+ "step": 6937
1196
+ },
1197
+ {
1198
+ "epoch": 37.333333333333336,
1199
+ "grad_norm": 3.664700984954834,
1200
+ "learning_rate": 3e-05,
1201
+ "loss": 0.2606,
1202
+ "step": 7000
1203
+ },
1204
+ {
1205
+ "epoch": 37.86666666666667,
1206
+ "grad_norm": 3.442922830581665,
1207
+ "learning_rate": 3e-05,
1208
+ "loss": 0.2541,
1209
+ "step": 7100
1210
+ },
1211
+ {
1212
+ "epoch": 38.0,
1213
+ "eval_accuracy": 0.5738385650224215,
1214
+ "eval_loss": 3.1831231117248535,
1215
+ "eval_runtime": 6.7239,
1216
+ "eval_samples_per_second": 74.361,
1217
+ "eval_steps_per_second": 9.37,
1218
+ "step": 7125
1219
+ },
1220
+ {
1221
+ "epoch": 38.0,
1222
+ "eval_exact_match": 8.8,
1223
+ "eval_f1": 14.986277056277055,
1224
+ "eval_qa_bleu": 6.307928648492083,
1225
+ "eval_qa_exact_match": 0.072,
1226
+ "eval_recite_bleu": 10.683207288869733,
1227
+ "eval_recite_exact_match": 0.014,
1228
+ "step": 7125
1229
+ },
1230
+ {
1231
+ "epoch": 38.4,
1232
+ "grad_norm": 2.9627795219421387,
1233
+ "learning_rate": 3e-05,
1234
+ "loss": 0.2391,
1235
+ "step": 7200
1236
+ },
1237
+ {
1238
+ "epoch": 38.93333333333333,
1239
+ "grad_norm": 3.517446517944336,
1240
+ "learning_rate": 3e-05,
1241
+ "loss": 0.2411,
1242
+ "step": 7300
1243
+ },
1244
+ {
1245
+ "epoch": 38.99733333333333,
1246
+ "eval_accuracy": 0.5740358744394619,
1247
+ "eval_loss": 3.213481903076172,
1248
+ "eval_runtime": 5.6174,
1249
+ "eval_samples_per_second": 89.009,
1250
+ "eval_steps_per_second": 11.215,
1251
+ "step": 7312
1252
+ },
1253
+ {
1254
+ "epoch": 38.99733333333333,
1255
+ "eval_exact_match": 9.4,
1256
+ "eval_f1": 15.155079365079372,
1257
+ "eval_qa_bleu": 5.289271493416481,
1258
+ "eval_qa_exact_match": 0.078,
1259
+ "eval_recite_bleu": 11.161407462076504,
1260
+ "eval_recite_exact_match": 0.018,
1261
+ "step": 7312
1262
+ },
1263
+ {
1264
+ "epoch": 39.46666666666667,
1265
+ "grad_norm": 2.9763762950897217,
1266
+ "learning_rate": 3e-05,
1267
+ "loss": 0.2232,
1268
+ "step": 7400
1269
+ },
1270
+ {
1271
+ "epoch": 40.0,
1272
+ "grad_norm": 2.818361520767212,
1273
+ "learning_rate": 3e-05,
1274
+ "loss": 0.228,
1275
+ "step": 7500
1276
+ },
1277
+ {
1278
+ "epoch": 40.0,
1279
+ "eval_accuracy": 0.5739192825112107,
1280
+ "eval_loss": 3.2505199909210205,
1281
+ "eval_runtime": 6.0109,
1282
+ "eval_samples_per_second": 83.182,
1283
+ "eval_steps_per_second": 10.481,
1284
+ "step": 7500
1285
+ },
1286
+ {
1287
+ "epoch": 40.0,
1288
+ "eval_exact_match": 8.6,
1289
+ "eval_f1": 14.30142857142857,
1290
+ "eval_qa_bleu": 6.210824473363826,
1291
+ "eval_qa_exact_match": 0.074,
1292
+ "eval_recite_bleu": 11.205695530168638,
1293
+ "eval_recite_exact_match": 0.014,
1294
+ "step": 7500
1295
+ },
1296
+ {
1297
+ "epoch": 40.53333333333333,
1298
+ "grad_norm": 3.813812494277954,
1299
+ "learning_rate": 3e-05,
1300
+ "loss": 0.2067,
1301
+ "step": 7600
1302
+ },
1303
+ {
1304
+ "epoch": 40.99733333333333,
1305
+ "eval_accuracy": 0.5743408071748879,
1306
+ "eval_loss": 3.286703586578369,
1307
+ "eval_runtime": 5.9771,
1308
+ "eval_samples_per_second": 83.652,
1309
+ "eval_steps_per_second": 10.54,
1310
+ "step": 7687
1311
+ },
1312
+ {
1313
+ "epoch": 40.99733333333333,
1314
+ "eval_exact_match": 7.8,
1315
+ "eval_f1": 13.974487734487733,
1316
+ "eval_qa_bleu": 9.0024711252003,
1317
+ "eval_qa_exact_match": 0.058,
1318
+ "eval_recite_bleu": 11.159844095103711,
1319
+ "eval_recite_exact_match": 0.016,
1320
+ "step": 7687
1321
+ },
1322
+ {
1323
+ "epoch": 41.06666666666667,
1324
+ "grad_norm": 2.2731709480285645,
1325
+ "learning_rate": 3e-05,
1326
+ "loss": 0.2137,
1327
+ "step": 7700
1328
+ },
1329
+ {
1330
+ "epoch": 41.6,
1331
+ "grad_norm": 2.825331449508667,
1332
+ "learning_rate": 3e-05,
1333
+ "loss": 0.1952,
1334
+ "step": 7800
1335
+ },
1336
+ {
1337
+ "epoch": 42.0,
1338
+ "eval_accuracy": 0.5750941704035875,
1339
+ "eval_loss": 3.3047103881835938,
1340
+ "eval_runtime": 5.951,
1341
+ "eval_samples_per_second": 84.02,
1342
+ "eval_steps_per_second": 10.586,
1343
+ "step": 7875
1344
+ },
1345
+ {
1346
+ "epoch": 42.0,
1347
+ "eval_exact_match": 10.6,
1348
+ "eval_f1": 15.946507936507942,
1349
+ "eval_qa_bleu": 6.53814009433881,
1350
+ "eval_qa_exact_match": 0.088,
1351
+ "eval_recite_bleu": 11.468470281750024,
1352
+ "eval_recite_exact_match": 0.02,
1353
+ "step": 7875
1354
+ },
1355
+ {
1356
+ "epoch": 42.13333333333333,
1357
+ "grad_norm": 2.672769546508789,
1358
+ "learning_rate": 3e-05,
1359
+ "loss": 0.2015,
1360
+ "step": 7900
1361
+ },
1362
+ {
1363
+ "epoch": 42.666666666666664,
1364
+ "grad_norm": 2.9053127765655518,
1365
+ "learning_rate": 3e-05,
1366
+ "loss": 0.1886,
1367
+ "step": 8000
1368
+ },
1369
+ {
1370
+ "epoch": 42.99733333333333,
1371
+ "eval_accuracy": 0.5742152466367713,
1372
+ "eval_loss": 3.352785587310791,
1373
+ "eval_runtime": 5.943,
1374
+ "eval_samples_per_second": 84.132,
1375
+ "eval_steps_per_second": 10.601,
1376
+ "step": 8062
1377
+ },
1378
+ {
1379
+ "epoch": 42.99733333333333,
1380
+ "eval_exact_match": 9.2,
1381
+ "eval_f1": 16.31507936507937,
1382
+ "eval_qa_bleu": 5.198241099912352,
1383
+ "eval_qa_exact_match": 0.078,
1384
+ "eval_recite_bleu": 12.049991185460714,
1385
+ "eval_recite_exact_match": 0.024,
1386
+ "step": 8062
1387
+ },
1388
+ {
1389
+ "epoch": 43.2,
1390
+ "grad_norm": 2.504420757293701,
1391
+ "learning_rate": 3e-05,
1392
+ "loss": 0.1872,
1393
+ "step": 8100
1394
+ },
1395
+ {
1396
+ "epoch": 43.733333333333334,
1397
+ "grad_norm": 2.765127420425415,
1398
+ "learning_rate": 3e-05,
1399
+ "loss": 0.1828,
1400
+ "step": 8200
1401
+ },
1402
+ {
1403
+ "epoch": 44.0,
1404
+ "eval_accuracy": 0.5730134529147982,
1405
+ "eval_loss": 3.4431397914886475,
1406
+ "eval_runtime": 6.6114,
1407
+ "eval_samples_per_second": 75.627,
1408
+ "eval_steps_per_second": 9.529,
1409
+ "step": 8250
1410
+ },
1411
+ {
1412
+ "epoch": 44.0,
1413
+ "eval_exact_match": 10.0,
1414
+ "eval_f1": 16.226709956709957,
1415
+ "eval_qa_bleu": 3.7525853527808333,
1416
+ "eval_qa_exact_match": 0.084,
1417
+ "eval_recite_bleu": 11.540062918449511,
1418
+ "eval_recite_exact_match": 0.016,
1419
+ "step": 8250
1420
+ },
1421
+ {
1422
+ "epoch": 44.266666666666666,
1423
+ "grad_norm": 2.8351898193359375,
1424
+ "learning_rate": 3e-05,
1425
+ "loss": 0.1753,
1426
+ "step": 8300
1427
+ },
1428
+ {
1429
+ "epoch": 44.8,
1430
+ "grad_norm": 2.966702461242676,
1431
+ "learning_rate": 3e-05,
1432
+ "loss": 0.1743,
1433
+ "step": 8400
1434
+ },
1435
+ {
1436
+ "epoch": 44.99733333333333,
1437
+ "eval_accuracy": 0.5726905829596413,
1438
+ "eval_loss": 3.416621208190918,
1439
+ "eval_runtime": 6.3887,
1440
+ "eval_samples_per_second": 78.263,
1441
+ "eval_steps_per_second": 9.861,
1442
+ "step": 8437
1443
+ },
1444
+ {
1445
+ "epoch": 44.99733333333333,
1446
+ "eval_exact_match": 8.0,
1447
+ "eval_f1": 13.308095238095238,
1448
+ "eval_qa_bleu": 6.543422950603953,
1449
+ "eval_qa_exact_match": 0.07,
1450
+ "eval_recite_bleu": 12.294354515156972,
1451
+ "eval_recite_exact_match": 0.022,
1452
+ "step": 8437
1453
+ },
1454
+ {
1455
+ "epoch": 45.333333333333336,
1456
+ "grad_norm": 2.7301738262176514,
1457
+ "learning_rate": 3e-05,
1458
+ "loss": 0.1665,
1459
+ "step": 8500
1460
+ },
1461
+ {
1462
+ "epoch": 45.86666666666667,
1463
+ "grad_norm": 3.0317089557647705,
1464
+ "learning_rate": 3e-05,
1465
+ "loss": 0.1691,
1466
+ "step": 8600
1467
+ },
1468
+ {
1469
+ "epoch": 46.0,
1470
+ "eval_accuracy": 0.5738834080717489,
1471
+ "eval_loss": 3.432551622390747,
1472
+ "eval_runtime": 5.9757,
1473
+ "eval_samples_per_second": 83.672,
1474
+ "eval_steps_per_second": 10.543,
1475
+ "step": 8625
1476
+ },
1477
+ {
1478
+ "epoch": 46.0,
1479
+ "eval_exact_match": 8.2,
1480
+ "eval_f1": 14.94995115995116,
1481
+ "eval_qa_bleu": 7.112359364013774,
1482
+ "eval_qa_exact_match": 0.068,
1483
+ "eval_recite_bleu": 12.858920217646185,
1484
+ "eval_recite_exact_match": 0.028,
1485
+ "step": 8625
1486
+ },
1487
+ {
1488
+ "epoch": 46.4,
1489
+ "grad_norm": 2.5876522064208984,
1490
+ "learning_rate": 3e-05,
1491
+ "loss": 0.1579,
1492
+ "step": 8700
1493
+ },
1494
+ {
1495
+ "epoch": 46.93333333333333,
1496
+ "grad_norm": 2.891350746154785,
1497
+ "learning_rate": 3e-05,
1498
+ "loss": 0.1633,
1499
+ "step": 8800
1500
+ },
1501
+ {
1502
+ "epoch": 46.99733333333333,
1503
+ "eval_accuracy": 0.5728071748878923,
1504
+ "eval_loss": 3.455498218536377,
1505
+ "eval_runtime": 6.3565,
1506
+ "eval_samples_per_second": 78.66,
1507
+ "eval_steps_per_second": 9.911,
1508
+ "step": 8812
1509
+ },
1510
+ {
1511
+ "epoch": 46.99733333333333,
1512
+ "eval_exact_match": 8.2,
1513
+ "eval_f1": 14.909369519369525,
1514
+ "eval_qa_bleu": 4.953527598248164,
1515
+ "eval_qa_exact_match": 0.066,
1516
+ "eval_recite_bleu": 11.51764513447178,
1517
+ "eval_recite_exact_match": 0.022,
1518
+ "step": 8812
1519
+ },
1520
+ {
1521
+ "epoch": 47.46666666666667,
1522
+ "grad_norm": 2.360337734222412,
1523
+ "learning_rate": 3e-05,
1524
+ "loss": 0.1513,
1525
+ "step": 8900
1526
+ },
1527
+ {
1528
+ "epoch": 48.0,
1529
+ "grad_norm": 2.782274007797241,
1530
+ "learning_rate": 3e-05,
1531
+ "loss": 0.156,
1532
+ "step": 9000
1533
+ },
1534
+ {
1535
+ "epoch": 48.0,
1536
+ "eval_accuracy": 0.5729058295964126,
1537
+ "eval_loss": 3.4876203536987305,
1538
+ "eval_runtime": 5.9502,
1539
+ "eval_samples_per_second": 84.031,
1540
+ "eval_steps_per_second": 10.588,
1541
+ "step": 9000
1542
+ },
1543
+ {
1544
+ "epoch": 48.0,
1545
+ "eval_exact_match": 8.2,
1546
+ "eval_f1": 14.631904761904762,
1547
+ "eval_qa_bleu": 5.962023207850432,
1548
+ "eval_qa_exact_match": 0.066,
1549
+ "eval_recite_bleu": 11.814543905192453,
1550
+ "eval_recite_exact_match": 0.022,
1551
+ "step": 9000
1552
+ },
1553
+ {
1554
+ "epoch": 48.53333333333333,
1555
+ "grad_norm": 1.921025037765503,
1556
+ "learning_rate": 3e-05,
1557
+ "loss": 0.1441,
1558
+ "step": 9100
1559
+ },
1560
+ {
1561
+ "epoch": 48.99733333333333,
1562
+ "eval_accuracy": 0.5727085201793722,
1563
+ "eval_loss": 3.5367648601531982,
1564
+ "eval_runtime": 5.6172,
1565
+ "eval_samples_per_second": 89.013,
1566
+ "eval_steps_per_second": 11.216,
1567
+ "step": 9187
1568
+ },
1569
+ {
1570
+ "epoch": 48.99733333333333,
1571
+ "eval_exact_match": 10.4,
1572
+ "eval_f1": 16.249711399711405,
1573
+ "eval_qa_bleu": 7.160795674838566,
1574
+ "eval_qa_exact_match": 0.088,
1575
+ "eval_recite_bleu": 11.964709782448564,
1576
+ "eval_recite_exact_match": 0.024,
1577
+ "step": 9187
1578
+ },
1579
+ {
1580
+ "epoch": 49.06666666666667,
1581
+ "grad_norm": 1.8280539512634277,
1582
+ "learning_rate": 3e-05,
1583
+ "loss": 0.1496,
1584
+ "step": 9200
1585
+ },
1586
+ {
1587
+ "epoch": 49.6,
1588
+ "grad_norm": 2.6036791801452637,
1589
+ "learning_rate": 3e-05,
1590
+ "loss": 0.1401,
1591
+ "step": 9300
1592
+ },
1593
+ {
1594
+ "epoch": 49.86666666666667,
1595
+ "eval_accuracy": 0.5729865470852018,
1596
+ "eval_loss": 3.544297456741333,
1597
+ "eval_runtime": 6.1658,
1598
+ "eval_samples_per_second": 81.093,
1599
+ "eval_steps_per_second": 10.218,
1600
+ "step": 9350
1601
+ },
1602
+ {
1603
+ "epoch": 49.86666666666667,
1604
+ "eval_exact_match": 7.8,
1605
+ "eval_f1": 13.20940836940837,
1606
+ "eval_qa_bleu": 7.766714267098657,
1607
+ "eval_qa_exact_match": 0.062,
1608
+ "eval_recite_bleu": 11.735267575530193,
1609
+ "eval_recite_exact_match": 0.022,
1610
+ "step": 9350
1611
+ }
1612
+ ],
1613
+ "logging_steps": 100,
1614
+ "max_steps": 9350,
1615
+ "num_input_tokens_seen": 0,
1616
+ "num_train_epochs": 50,
1617
+ "save_steps": 500,
1618
+ "stateful_callbacks": {
1619
+ "TrainerControl": {
1620
+ "args": {
1621
+ "should_epoch_stop": false,
1622
+ "should_evaluate": false,
1623
+ "should_log": false,
1624
+ "should_save": true,
1625
+ "should_training_stop": true
1626
+ },
1627
+ "attributes": {}
1628
+ }
1629
+ },
1630
+ "total_flos": 1.0317599310547517e+18,
1631
+ "train_batch_size": 2,
1632
+ "trial_name": null,
1633
+ "trial_params": null
1634
+ }
checkpoint-9350/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e9c21c8367f69c99b21b20aa3dbf722814d1b222850e23dacbe471b235b2dd4
3
+ size 5304
checkpoint-9350/vocab.json ADDED
The diff for this file is too large to render. See raw diff