tyzhu commited on
Commit
39f6b72
1 Parent(s): b020d75

Training in progress, epoch 48, checkpoint

Browse files
checkpoint-12000/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-12000/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-12000/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aa82e3b7c22507e6dfae856c7f9f6ef39d2b90b9ded142d13e71438d5e0b3e0f
3
+ size 112130416
checkpoint-12000/added_tokens.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "<|endoftext|>": 151643,
3
+ "<|im_end|>": 151645,
4
+ "<|im_start|>": 151644
5
+ }
checkpoint-12000/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-12000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e8b1b8a870379ee09f8f8c13e0ba694655f2e40f2511b9de17b44ea3451a6a4
3
+ size 224537202
checkpoint-12000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1cc251df60fba6219ec1cb6fe076832703aebdc4cc10f4db952f0f0978fb23cb
3
+ size 15024
checkpoint-12000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9d727c0d20d5e463899459765ba6f87c8dac156377fdeb7916d38b8b8a3db76
3
+ size 15024
checkpoint-12000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e0d2be21d700ec49e20516a87166099e6b9c7ab829b8adfb6297c6f9a4233ea4
3
+ size 15024
checkpoint-12000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4e7968ff8df24d4335e7ad1201d6d5aef84b72ea8844961b80665284e0d891d
3
+ size 15024
checkpoint-12000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d509c18db3995f106cfe3bc371660913f64270b4caadefd9587ff0d8b1ea802c
3
+ size 1064
checkpoint-12000/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-12000/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-12000/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-12000/trainer_state.json ADDED
@@ -0,0 +1,1593 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 48.0,
5
+ "eval_steps": 500,
6
+ "global_step": 12000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.4,
13
+ "grad_norm": 0.6251392960548401,
14
+ "learning_rate": 5e-05,
15
+ "loss": 2.427,
16
+ "step": 100
17
+ },
18
+ {
19
+ "epoch": 0.8,
20
+ "grad_norm": 0.6590011715888977,
21
+ "learning_rate": 5e-05,
22
+ "loss": 2.2748,
23
+ "step": 200
24
+ },
25
+ {
26
+ "epoch": 1.0,
27
+ "eval_accuracy": 0.5142539682539683,
28
+ "eval_loss": 2.336690902709961,
29
+ "eval_runtime": 5.4733,
30
+ "eval_samples_per_second": 91.353,
31
+ "eval_steps_per_second": 11.51,
32
+ "step": 250
33
+ },
34
+ {
35
+ "epoch": 1.0,
36
+ "eval_exact_match": 16.8,
37
+ "eval_f1": 24.850246583481884,
38
+ "step": 250
39
+ },
40
+ {
41
+ "epoch": 1.2,
42
+ "grad_norm": 0.8194239139556885,
43
+ "learning_rate": 5e-05,
44
+ "loss": 2.2322,
45
+ "step": 300
46
+ },
47
+ {
48
+ "epoch": 1.6,
49
+ "grad_norm": 0.9867294430732727,
50
+ "learning_rate": 5e-05,
51
+ "loss": 2.1904,
52
+ "step": 400
53
+ },
54
+ {
55
+ "epoch": 2.0,
56
+ "grad_norm": 0.9660400152206421,
57
+ "learning_rate": 5e-05,
58
+ "loss": 2.1657,
59
+ "step": 500
60
+ },
61
+ {
62
+ "epoch": 2.0,
63
+ "eval_accuracy": 0.5168253968253969,
64
+ "eval_loss": 2.323688507080078,
65
+ "eval_runtime": 5.1683,
66
+ "eval_samples_per_second": 96.744,
67
+ "eval_steps_per_second": 12.19,
68
+ "step": 500
69
+ },
70
+ {
71
+ "epoch": 2.0,
72
+ "eval_exact_match": 17.2,
73
+ "eval_f1": 24.620036075036083,
74
+ "step": 500
75
+ },
76
+ {
77
+ "epoch": 2.4,
78
+ "grad_norm": 1.281128168106079,
79
+ "learning_rate": 5e-05,
80
+ "loss": 2.0251,
81
+ "step": 600
82
+ },
83
+ {
84
+ "epoch": 2.8,
85
+ "grad_norm": 1.3365116119384766,
86
+ "learning_rate": 5e-05,
87
+ "loss": 2.0158,
88
+ "step": 700
89
+ },
90
+ {
91
+ "epoch": 3.0,
92
+ "eval_accuracy": 0.5156507936507937,
93
+ "eval_loss": 2.3560824394226074,
94
+ "eval_runtime": 5.4675,
95
+ "eval_samples_per_second": 91.449,
96
+ "eval_steps_per_second": 11.523,
97
+ "step": 750
98
+ },
99
+ {
100
+ "epoch": 3.0,
101
+ "eval_exact_match": 19.0,
102
+ "eval_f1": 27.242790688843332,
103
+ "step": 750
104
+ },
105
+ {
106
+ "epoch": 3.2,
107
+ "grad_norm": 1.7840828895568848,
108
+ "learning_rate": 5e-05,
109
+ "loss": 1.9394,
110
+ "step": 800
111
+ },
112
+ {
113
+ "epoch": 3.6,
114
+ "grad_norm": 1.8030190467834473,
115
+ "learning_rate": 5e-05,
116
+ "loss": 1.8502,
117
+ "step": 900
118
+ },
119
+ {
120
+ "epoch": 4.0,
121
+ "grad_norm": 1.9682222604751587,
122
+ "learning_rate": 5e-05,
123
+ "loss": 1.8682,
124
+ "step": 1000
125
+ },
126
+ {
127
+ "epoch": 4.0,
128
+ "eval_accuracy": 0.5131746031746032,
129
+ "eval_loss": 2.4071946144104004,
130
+ "eval_runtime": 5.2825,
131
+ "eval_samples_per_second": 94.653,
132
+ "eval_steps_per_second": 11.926,
133
+ "step": 1000
134
+ },
135
+ {
136
+ "epoch": 4.0,
137
+ "eval_exact_match": 17.6,
138
+ "eval_f1": 26.010988455988468,
139
+ "step": 1000
140
+ },
141
+ {
142
+ "epoch": 4.4,
143
+ "grad_norm": 2.138505220413208,
144
+ "learning_rate": 5e-05,
145
+ "loss": 1.6795,
146
+ "step": 1100
147
+ },
148
+ {
149
+ "epoch": 4.8,
150
+ "grad_norm": 2.3212926387786865,
151
+ "learning_rate": 5e-05,
152
+ "loss": 1.7036,
153
+ "step": 1200
154
+ },
155
+ {
156
+ "epoch": 5.0,
157
+ "eval_accuracy": 0.5099365079365079,
158
+ "eval_loss": 2.494770050048828,
159
+ "eval_runtime": 5.1799,
160
+ "eval_samples_per_second": 96.528,
161
+ "eval_steps_per_second": 12.162,
162
+ "step": 1250
163
+ },
164
+ {
165
+ "epoch": 5.0,
166
+ "eval_exact_match": 17.2,
167
+ "eval_f1": 25.84752525252526,
168
+ "step": 1250
169
+ },
170
+ {
171
+ "epoch": 5.2,
172
+ "grad_norm": 2.4878485202789307,
173
+ "learning_rate": 5e-05,
174
+ "loss": 1.6305,
175
+ "step": 1300
176
+ },
177
+ {
178
+ "epoch": 5.6,
179
+ "grad_norm": 2.599329710006714,
180
+ "learning_rate": 5e-05,
181
+ "loss": 1.5477,
182
+ "step": 1400
183
+ },
184
+ {
185
+ "epoch": 6.0,
186
+ "grad_norm": 2.5542759895324707,
187
+ "learning_rate": 5e-05,
188
+ "loss": 1.5708,
189
+ "step": 1500
190
+ },
191
+ {
192
+ "epoch": 6.0,
193
+ "eval_accuracy": 0.5065079365079365,
194
+ "eval_loss": 2.5797038078308105,
195
+ "eval_runtime": 5.6827,
196
+ "eval_samples_per_second": 87.986,
197
+ "eval_steps_per_second": 11.086,
198
+ "step": 1500
199
+ },
200
+ {
201
+ "epoch": 6.0,
202
+ "eval_exact_match": 17.0,
203
+ "eval_f1": 25.310858585858586,
204
+ "step": 1500
205
+ },
206
+ {
207
+ "epoch": 6.4,
208
+ "grad_norm": 3.1968846321105957,
209
+ "learning_rate": 5e-05,
210
+ "loss": 1.3896,
211
+ "step": 1600
212
+ },
213
+ {
214
+ "epoch": 6.8,
215
+ "grad_norm": 3.8637375831604004,
216
+ "learning_rate": 5e-05,
217
+ "loss": 1.421,
218
+ "step": 1700
219
+ },
220
+ {
221
+ "epoch": 7.0,
222
+ "eval_accuracy": 0.5038412698412699,
223
+ "eval_loss": 2.6977317333221436,
224
+ "eval_runtime": 5.3312,
225
+ "eval_samples_per_second": 93.787,
226
+ "eval_steps_per_second": 11.817,
227
+ "step": 1750
228
+ },
229
+ {
230
+ "epoch": 7.0,
231
+ "eval_exact_match": 16.0,
232
+ "eval_f1": 23.99441558441559,
233
+ "step": 1750
234
+ },
235
+ {
236
+ "epoch": 7.2,
237
+ "grad_norm": 3.468459367752075,
238
+ "learning_rate": 5e-05,
239
+ "loss": 1.3366,
240
+ "step": 1800
241
+ },
242
+ {
243
+ "epoch": 7.6,
244
+ "grad_norm": 3.5727667808532715,
245
+ "learning_rate": 5e-05,
246
+ "loss": 1.2585,
247
+ "step": 1900
248
+ },
249
+ {
250
+ "epoch": 8.0,
251
+ "grad_norm": 3.508258819580078,
252
+ "learning_rate": 5e-05,
253
+ "loss": 1.296,
254
+ "step": 2000
255
+ },
256
+ {
257
+ "epoch": 8.0,
258
+ "eval_accuracy": 0.5013333333333333,
259
+ "eval_loss": 2.786928653717041,
260
+ "eval_runtime": 5.8398,
261
+ "eval_samples_per_second": 85.62,
262
+ "eval_steps_per_second": 10.788,
263
+ "step": 2000
264
+ },
265
+ {
266
+ "epoch": 8.0,
267
+ "eval_exact_match": 15.4,
268
+ "eval_f1": 22.542871572871572,
269
+ "step": 2000
270
+ },
271
+ {
272
+ "epoch": 8.4,
273
+ "grad_norm": 4.042112350463867,
274
+ "learning_rate": 5e-05,
275
+ "loss": 1.1084,
276
+ "step": 2100
277
+ },
278
+ {
279
+ "epoch": 8.8,
280
+ "grad_norm": 3.7548155784606934,
281
+ "learning_rate": 5e-05,
282
+ "loss": 1.1457,
283
+ "step": 2200
284
+ },
285
+ {
286
+ "epoch": 9.0,
287
+ "eval_accuracy": 0.49692063492063493,
288
+ "eval_loss": 2.9215197563171387,
289
+ "eval_runtime": 5.1412,
290
+ "eval_samples_per_second": 97.254,
291
+ "eval_steps_per_second": 12.254,
292
+ "step": 2250
293
+ },
294
+ {
295
+ "epoch": 9.0,
296
+ "eval_exact_match": 15.0,
297
+ "eval_f1": 21.59087301587302,
298
+ "step": 2250
299
+ },
300
+ {
301
+ "epoch": 9.2,
302
+ "grad_norm": 4.0875773429870605,
303
+ "learning_rate": 5e-05,
304
+ "loss": 1.0677,
305
+ "step": 2300
306
+ },
307
+ {
308
+ "epoch": 9.6,
309
+ "grad_norm": 4.143901348114014,
310
+ "learning_rate": 5e-05,
311
+ "loss": 1.0028,
312
+ "step": 2400
313
+ },
314
+ {
315
+ "epoch": 10.0,
316
+ "grad_norm": 5.0514445304870605,
317
+ "learning_rate": 5e-05,
318
+ "loss": 1.0401,
319
+ "step": 2500
320
+ },
321
+ {
322
+ "epoch": 10.0,
323
+ "eval_accuracy": 0.4953015873015873,
324
+ "eval_loss": 3.054328203201294,
325
+ "eval_runtime": 5.4195,
326
+ "eval_samples_per_second": 92.259,
327
+ "eval_steps_per_second": 11.625,
328
+ "step": 2500
329
+ },
330
+ {
331
+ "epoch": 10.0,
332
+ "eval_exact_match": 15.4,
333
+ "eval_f1": 21.505119880119878,
334
+ "step": 2500
335
+ },
336
+ {
337
+ "epoch": 10.4,
338
+ "grad_norm": 3.8740394115448,
339
+ "learning_rate": 5e-05,
340
+ "loss": 0.8595,
341
+ "step": 2600
342
+ },
343
+ {
344
+ "epoch": 10.8,
345
+ "grad_norm": 4.170004844665527,
346
+ "learning_rate": 5e-05,
347
+ "loss": 0.9186,
348
+ "step": 2700
349
+ },
350
+ {
351
+ "epoch": 11.0,
352
+ "eval_accuracy": 0.49298412698412697,
353
+ "eval_loss": 3.177030324935913,
354
+ "eval_runtime": 5.9501,
355
+ "eval_samples_per_second": 84.033,
356
+ "eval_steps_per_second": 10.588,
357
+ "step": 2750
358
+ },
359
+ {
360
+ "epoch": 11.0,
361
+ "eval_exact_match": 15.6,
362
+ "eval_f1": 20.912753357753356,
363
+ "step": 2750
364
+ },
365
+ {
366
+ "epoch": 11.2,
367
+ "grad_norm": 4.2533392906188965,
368
+ "learning_rate": 5e-05,
369
+ "loss": 0.8411,
370
+ "step": 2800
371
+ },
372
+ {
373
+ "epoch": 11.6,
374
+ "grad_norm": 4.165424346923828,
375
+ "learning_rate": 5e-05,
376
+ "loss": 0.7934,
377
+ "step": 2900
378
+ },
379
+ {
380
+ "epoch": 12.0,
381
+ "grad_norm": 4.517414093017578,
382
+ "learning_rate": 5e-05,
383
+ "loss": 0.8386,
384
+ "step": 3000
385
+ },
386
+ {
387
+ "epoch": 12.0,
388
+ "eval_accuracy": 0.4916190476190476,
389
+ "eval_loss": 3.2663722038269043,
390
+ "eval_runtime": 5.1518,
391
+ "eval_samples_per_second": 97.053,
392
+ "eval_steps_per_second": 12.229,
393
+ "step": 3000
394
+ },
395
+ {
396
+ "epoch": 12.0,
397
+ "eval_exact_match": 15.2,
398
+ "eval_f1": 20.465463425463433,
399
+ "step": 3000
400
+ },
401
+ {
402
+ "epoch": 12.4,
403
+ "grad_norm": 4.059298515319824,
404
+ "learning_rate": 5e-05,
405
+ "loss": 0.6908,
406
+ "step": 3100
407
+ },
408
+ {
409
+ "epoch": 12.8,
410
+ "grad_norm": 3.4847490787506104,
411
+ "learning_rate": 5e-05,
412
+ "loss": 0.7326,
413
+ "step": 3200
414
+ },
415
+ {
416
+ "epoch": 13.0,
417
+ "eval_accuracy": 0.49047619047619045,
418
+ "eval_loss": 3.38826584815979,
419
+ "eval_runtime": 5.2085,
420
+ "eval_samples_per_second": 95.997,
421
+ "eval_steps_per_second": 12.096,
422
+ "step": 3250
423
+ },
424
+ {
425
+ "epoch": 13.0,
426
+ "eval_exact_match": 14.0,
427
+ "eval_f1": 20.144243697478988,
428
+ "step": 3250
429
+ },
430
+ {
431
+ "epoch": 13.2,
432
+ "grad_norm": 4.07555627822876,
433
+ "learning_rate": 5e-05,
434
+ "loss": 0.6901,
435
+ "step": 3300
436
+ },
437
+ {
438
+ "epoch": 13.6,
439
+ "grad_norm": 3.918684482574463,
440
+ "learning_rate": 5e-05,
441
+ "loss": 0.6432,
442
+ "step": 3400
443
+ },
444
+ {
445
+ "epoch": 14.0,
446
+ "grad_norm": 4.3187384605407715,
447
+ "learning_rate": 5e-05,
448
+ "loss": 0.6762,
449
+ "step": 3500
450
+ },
451
+ {
452
+ "epoch": 14.0,
453
+ "eval_accuracy": 0.4887936507936508,
454
+ "eval_loss": 3.466782808303833,
455
+ "eval_runtime": 5.5309,
456
+ "eval_samples_per_second": 90.402,
457
+ "eval_steps_per_second": 11.391,
458
+ "step": 3500
459
+ },
460
+ {
461
+ "epoch": 14.0,
462
+ "eval_exact_match": 14.6,
463
+ "eval_f1": 20.548629148629157,
464
+ "step": 3500
465
+ },
466
+ {
467
+ "epoch": 14.4,
468
+ "grad_norm": 3.270817995071411,
469
+ "learning_rate": 5e-05,
470
+ "loss": 0.5785,
471
+ "step": 3600
472
+ },
473
+ {
474
+ "epoch": 14.8,
475
+ "grad_norm": 3.882906436920166,
476
+ "learning_rate": 5e-05,
477
+ "loss": 0.6117,
478
+ "step": 3700
479
+ },
480
+ {
481
+ "epoch": 15.0,
482
+ "eval_accuracy": 0.48904761904761906,
483
+ "eval_loss": 3.5602962970733643,
484
+ "eval_runtime": 5.4424,
485
+ "eval_samples_per_second": 91.871,
486
+ "eval_steps_per_second": 11.576,
487
+ "step": 3750
488
+ },
489
+ {
490
+ "epoch": 15.0,
491
+ "eval_exact_match": 14.8,
492
+ "eval_f1": 20.941592997645635,
493
+ "step": 3750
494
+ },
495
+ {
496
+ "epoch": 15.2,
497
+ "grad_norm": 4.002102851867676,
498
+ "learning_rate": 5e-05,
499
+ "loss": 0.5725,
500
+ "step": 3800
501
+ },
502
+ {
503
+ "epoch": 15.6,
504
+ "grad_norm": 2.8693177700042725,
505
+ "learning_rate": 5e-05,
506
+ "loss": 0.5581,
507
+ "step": 3900
508
+ },
509
+ {
510
+ "epoch": 16.0,
511
+ "grad_norm": 4.3405256271362305,
512
+ "learning_rate": 5e-05,
513
+ "loss": 0.5729,
514
+ "step": 4000
515
+ },
516
+ {
517
+ "epoch": 16.0,
518
+ "eval_accuracy": 0.48742857142857143,
519
+ "eval_loss": 3.6583478450775146,
520
+ "eval_runtime": 5.1711,
521
+ "eval_samples_per_second": 96.692,
522
+ "eval_steps_per_second": 12.183,
523
+ "step": 4000
524
+ },
525
+ {
526
+ "epoch": 16.0,
527
+ "eval_exact_match": 14.8,
528
+ "eval_f1": 20.59385852709383,
529
+ "step": 4000
530
+ },
531
+ {
532
+ "epoch": 16.4,
533
+ "grad_norm": 4.715936183929443,
534
+ "learning_rate": 5e-05,
535
+ "loss": 0.5028,
536
+ "step": 4100
537
+ },
538
+ {
539
+ "epoch": 16.8,
540
+ "grad_norm": 3.7355644702911377,
541
+ "learning_rate": 5e-05,
542
+ "loss": 0.5317,
543
+ "step": 4200
544
+ },
545
+ {
546
+ "epoch": 17.0,
547
+ "eval_accuracy": 0.4873968253968254,
548
+ "eval_loss": 3.6984667778015137,
549
+ "eval_runtime": 5.4593,
550
+ "eval_samples_per_second": 91.587,
551
+ "eval_steps_per_second": 11.54,
552
+ "step": 4250
553
+ },
554
+ {
555
+ "epoch": 17.0,
556
+ "eval_exact_match": 13.8,
557
+ "eval_f1": 20.517505228159344,
558
+ "step": 4250
559
+ },
560
+ {
561
+ "epoch": 17.2,
562
+ "grad_norm": 2.7771668434143066,
563
+ "learning_rate": 5e-05,
564
+ "loss": 0.505,
565
+ "step": 4300
566
+ },
567
+ {
568
+ "epoch": 17.6,
569
+ "grad_norm": 3.6721012592315674,
570
+ "learning_rate": 5e-05,
571
+ "loss": 0.496,
572
+ "step": 4400
573
+ },
574
+ {
575
+ "epoch": 18.0,
576
+ "grad_norm": 4.477765083312988,
577
+ "learning_rate": 5e-05,
578
+ "loss": 0.5109,
579
+ "step": 4500
580
+ },
581
+ {
582
+ "epoch": 18.0,
583
+ "eval_accuracy": 0.4871111111111111,
584
+ "eval_loss": 3.7672173976898193,
585
+ "eval_runtime": 5.1624,
586
+ "eval_samples_per_second": 96.855,
587
+ "eval_steps_per_second": 12.204,
588
+ "step": 4500
589
+ },
590
+ {
591
+ "epoch": 18.0,
592
+ "eval_exact_match": 13.2,
593
+ "eval_f1": 19.935008488243795,
594
+ "step": 4500
595
+ },
596
+ {
597
+ "epoch": 18.4,
598
+ "grad_norm": 2.88653302192688,
599
+ "learning_rate": 5e-05,
600
+ "loss": 0.4564,
601
+ "step": 4600
602
+ },
603
+ {
604
+ "epoch": 18.8,
605
+ "grad_norm": 2.8045480251312256,
606
+ "learning_rate": 5e-05,
607
+ "loss": 0.479,
608
+ "step": 4700
609
+ },
610
+ {
611
+ "epoch": 19.0,
612
+ "eval_accuracy": 0.4865714285714286,
613
+ "eval_loss": 3.847074270248413,
614
+ "eval_runtime": 5.4643,
615
+ "eval_samples_per_second": 91.503,
616
+ "eval_steps_per_second": 11.529,
617
+ "step": 4750
618
+ },
619
+ {
620
+ "epoch": 19.0,
621
+ "eval_exact_match": 14.2,
622
+ "eval_f1": 20.744217407979033,
623
+ "step": 4750
624
+ },
625
+ {
626
+ "epoch": 19.2,
627
+ "grad_norm": 2.717168092727661,
628
+ "learning_rate": 5e-05,
629
+ "loss": 0.4612,
630
+ "step": 4800
631
+ },
632
+ {
633
+ "epoch": 19.6,
634
+ "grad_norm": 3.0051259994506836,
635
+ "learning_rate": 5e-05,
636
+ "loss": 0.453,
637
+ "step": 4900
638
+ },
639
+ {
640
+ "epoch": 20.0,
641
+ "grad_norm": 2.753119468688965,
642
+ "learning_rate": 5e-05,
643
+ "loss": 0.4675,
644
+ "step": 5000
645
+ },
646
+ {
647
+ "epoch": 20.0,
648
+ "eval_accuracy": 0.4867936507936508,
649
+ "eval_loss": 3.914645195007324,
650
+ "eval_runtime": 5.9419,
651
+ "eval_samples_per_second": 84.148,
652
+ "eval_steps_per_second": 10.603,
653
+ "step": 5000
654
+ },
655
+ {
656
+ "epoch": 20.0,
657
+ "eval_exact_match": 13.2,
658
+ "eval_f1": 20.16965067612128,
659
+ "step": 5000
660
+ },
661
+ {
662
+ "epoch": 20.4,
663
+ "grad_norm": 2.736196517944336,
664
+ "learning_rate": 5e-05,
665
+ "loss": 0.425,
666
+ "step": 5100
667
+ },
668
+ {
669
+ "epoch": 20.8,
670
+ "grad_norm": 3.410940647125244,
671
+ "learning_rate": 5e-05,
672
+ "loss": 0.448,
673
+ "step": 5200
674
+ },
675
+ {
676
+ "epoch": 21.0,
677
+ "eval_accuracy": 0.485968253968254,
678
+ "eval_loss": 3.955155849456787,
679
+ "eval_runtime": 5.1881,
680
+ "eval_samples_per_second": 96.375,
681
+ "eval_steps_per_second": 12.143,
682
+ "step": 5250
683
+ },
684
+ {
685
+ "epoch": 21.0,
686
+ "eval_exact_match": 13.4,
687
+ "eval_f1": 19.219269782505087,
688
+ "step": 5250
689
+ },
690
+ {
691
+ "epoch": 21.2,
692
+ "grad_norm": 3.2622647285461426,
693
+ "learning_rate": 5e-05,
694
+ "loss": 0.4306,
695
+ "step": 5300
696
+ },
697
+ {
698
+ "epoch": 21.6,
699
+ "grad_norm": 2.9486541748046875,
700
+ "learning_rate": 5e-05,
701
+ "loss": 0.4291,
702
+ "step": 5400
703
+ },
704
+ {
705
+ "epoch": 22.0,
706
+ "grad_norm": 3.2296924591064453,
707
+ "learning_rate": 5e-05,
708
+ "loss": 0.4389,
709
+ "step": 5500
710
+ },
711
+ {
712
+ "epoch": 22.0,
713
+ "eval_accuracy": 0.4855238095238095,
714
+ "eval_loss": 3.967400312423706,
715
+ "eval_runtime": 5.4481,
716
+ "eval_samples_per_second": 91.776,
717
+ "eval_steps_per_second": 11.564,
718
+ "step": 5500
719
+ },
720
+ {
721
+ "epoch": 22.0,
722
+ "eval_exact_match": 14.6,
723
+ "eval_f1": 20.702058565191567,
724
+ "step": 5500
725
+ },
726
+ {
727
+ "epoch": 22.4,
728
+ "grad_norm": 2.210437059402466,
729
+ "learning_rate": 5e-05,
730
+ "loss": 0.4047,
731
+ "step": 5600
732
+ },
733
+ {
734
+ "epoch": 22.8,
735
+ "grad_norm": 2.7681708335876465,
736
+ "learning_rate": 5e-05,
737
+ "loss": 0.4227,
738
+ "step": 5700
739
+ },
740
+ {
741
+ "epoch": 23.0,
742
+ "eval_accuracy": 0.48682539682539683,
743
+ "eval_loss": 4.001584053039551,
744
+ "eval_runtime": 5.9051,
745
+ "eval_samples_per_second": 84.673,
746
+ "eval_steps_per_second": 10.669,
747
+ "step": 5750
748
+ },
749
+ {
750
+ "epoch": 23.0,
751
+ "eval_exact_match": 14.6,
752
+ "eval_f1": 22.022229437229445,
753
+ "step": 5750
754
+ },
755
+ {
756
+ "epoch": 23.2,
757
+ "grad_norm": 2.4847607612609863,
758
+ "learning_rate": 5e-05,
759
+ "loss": 0.4088,
760
+ "step": 5800
761
+ },
762
+ {
763
+ "epoch": 23.6,
764
+ "grad_norm": 2.022418260574341,
765
+ "learning_rate": 5e-05,
766
+ "loss": 0.4045,
767
+ "step": 5900
768
+ },
769
+ {
770
+ "epoch": 24.0,
771
+ "grad_norm": 1.8691329956054688,
772
+ "learning_rate": 5e-05,
773
+ "loss": 0.4234,
774
+ "step": 6000
775
+ },
776
+ {
777
+ "epoch": 24.0,
778
+ "eval_accuracy": 0.4853650793650794,
779
+ "eval_loss": 4.042004585266113,
780
+ "eval_runtime": 5.5095,
781
+ "eval_samples_per_second": 90.752,
782
+ "eval_steps_per_second": 11.435,
783
+ "step": 6000
784
+ },
785
+ {
786
+ "epoch": 24.0,
787
+ "eval_exact_match": 13.8,
788
+ "eval_f1": 21.028606654783133,
789
+ "step": 6000
790
+ },
791
+ {
792
+ "epoch": 24.4,
793
+ "grad_norm": 2.199568748474121,
794
+ "learning_rate": 5e-05,
795
+ "loss": 0.3915,
796
+ "step": 6100
797
+ },
798
+ {
799
+ "epoch": 24.8,
800
+ "grad_norm": 2.636310338973999,
801
+ "learning_rate": 5e-05,
802
+ "loss": 0.4051,
803
+ "step": 6200
804
+ },
805
+ {
806
+ "epoch": 25.0,
807
+ "eval_accuracy": 0.4865714285714286,
808
+ "eval_loss": 4.044601917266846,
809
+ "eval_runtime": 5.1578,
810
+ "eval_samples_per_second": 96.941,
811
+ "eval_steps_per_second": 12.215,
812
+ "step": 6250
813
+ },
814
+ {
815
+ "epoch": 25.0,
816
+ "eval_exact_match": 12.8,
817
+ "eval_f1": 20.273670088081865,
818
+ "step": 6250
819
+ },
820
+ {
821
+ "epoch": 25.2,
822
+ "grad_norm": 2.0600600242614746,
823
+ "learning_rate": 5e-05,
824
+ "loss": 0.3952,
825
+ "step": 6300
826
+ },
827
+ {
828
+ "epoch": 25.6,
829
+ "grad_norm": 2.2660067081451416,
830
+ "learning_rate": 5e-05,
831
+ "loss": 0.3939,
832
+ "step": 6400
833
+ },
834
+ {
835
+ "epoch": 26.0,
836
+ "grad_norm": 2.225553512573242,
837
+ "learning_rate": 5e-05,
838
+ "loss": 0.4043,
839
+ "step": 6500
840
+ },
841
+ {
842
+ "epoch": 26.0,
843
+ "eval_accuracy": 0.4855238095238095,
844
+ "eval_loss": 4.121946334838867,
845
+ "eval_runtime": 6.1465,
846
+ "eval_samples_per_second": 81.347,
847
+ "eval_steps_per_second": 10.25,
848
+ "step": 6500
849
+ },
850
+ {
851
+ "epoch": 26.0,
852
+ "eval_exact_match": 12.8,
853
+ "eval_f1": 18.98795730413378,
854
+ "step": 6500
855
+ },
856
+ {
857
+ "epoch": 26.4,
858
+ "grad_norm": 1.9317594766616821,
859
+ "learning_rate": 5e-05,
860
+ "loss": 0.3786,
861
+ "step": 6600
862
+ },
863
+ {
864
+ "epoch": 26.8,
865
+ "grad_norm": 2.19458270072937,
866
+ "learning_rate": 5e-05,
867
+ "loss": 0.396,
868
+ "step": 6700
869
+ },
870
+ {
871
+ "epoch": 27.0,
872
+ "eval_accuracy": 0.48542857142857143,
873
+ "eval_loss": 4.126931190490723,
874
+ "eval_runtime": 5.1829,
875
+ "eval_samples_per_second": 96.472,
876
+ "eval_steps_per_second": 12.155,
877
+ "step": 6750
878
+ },
879
+ {
880
+ "epoch": 27.0,
881
+ "eval_exact_match": 13.6,
882
+ "eval_f1": 20.55248196248197,
883
+ "step": 6750
884
+ },
885
+ {
886
+ "epoch": 27.2,
887
+ "grad_norm": 1.6904568672180176,
888
+ "learning_rate": 5e-05,
889
+ "loss": 0.3843,
890
+ "step": 6800
891
+ },
892
+ {
893
+ "epoch": 27.6,
894
+ "grad_norm": 2.3401496410369873,
895
+ "learning_rate": 5e-05,
896
+ "loss": 0.3841,
897
+ "step": 6900
898
+ },
899
+ {
900
+ "epoch": 28.0,
901
+ "grad_norm": 2.264298439025879,
902
+ "learning_rate": 5e-05,
903
+ "loss": 0.3947,
904
+ "step": 7000
905
+ },
906
+ {
907
+ "epoch": 28.0,
908
+ "eval_accuracy": 0.48533333333333334,
909
+ "eval_loss": 4.117328643798828,
910
+ "eval_runtime": 5.4257,
911
+ "eval_samples_per_second": 92.154,
912
+ "eval_steps_per_second": 11.611,
913
+ "step": 7000
914
+ },
915
+ {
916
+ "epoch": 28.0,
917
+ "eval_exact_match": 13.2,
918
+ "eval_f1": 19.55831890331891,
919
+ "step": 7000
920
+ },
921
+ {
922
+ "epoch": 28.4,
923
+ "grad_norm": 1.7389506101608276,
924
+ "learning_rate": 5e-05,
925
+ "loss": 0.3732,
926
+ "step": 7100
927
+ },
928
+ {
929
+ "epoch": 28.8,
930
+ "grad_norm": 2.5369668006896973,
931
+ "learning_rate": 5e-05,
932
+ "loss": 0.383,
933
+ "step": 7200
934
+ },
935
+ {
936
+ "epoch": 29.0,
937
+ "eval_accuracy": 0.4855555555555556,
938
+ "eval_loss": 4.185734272003174,
939
+ "eval_runtime": 5.5139,
940
+ "eval_samples_per_second": 90.679,
941
+ "eval_steps_per_second": 11.426,
942
+ "step": 7250
943
+ },
944
+ {
945
+ "epoch": 29.0,
946
+ "eval_exact_match": 14.6,
947
+ "eval_f1": 20.956385281385288,
948
+ "step": 7250
949
+ },
950
+ {
951
+ "epoch": 29.2,
952
+ "grad_norm": 1.586300253868103,
953
+ "learning_rate": 5e-05,
954
+ "loss": 0.3743,
955
+ "step": 7300
956
+ },
957
+ {
958
+ "epoch": 29.6,
959
+ "grad_norm": 1.5583587884902954,
960
+ "learning_rate": 5e-05,
961
+ "loss": 0.3777,
962
+ "step": 7400
963
+ },
964
+ {
965
+ "epoch": 30.0,
966
+ "grad_norm": 1.5072301626205444,
967
+ "learning_rate": 5e-05,
968
+ "loss": 0.3866,
969
+ "step": 7500
970
+ },
971
+ {
972
+ "epoch": 30.0,
973
+ "eval_accuracy": 0.48784126984126985,
974
+ "eval_loss": 4.153143405914307,
975
+ "eval_runtime": 5.6131,
976
+ "eval_samples_per_second": 89.078,
977
+ "eval_steps_per_second": 11.224,
978
+ "step": 7500
979
+ },
980
+ {
981
+ "epoch": 30.0,
982
+ "eval_exact_match": 13.0,
983
+ "eval_f1": 19.762439945675247,
984
+ "step": 7500
985
+ },
986
+ {
987
+ "epoch": 30.4,
988
+ "grad_norm": 2.850844144821167,
989
+ "learning_rate": 5e-05,
990
+ "loss": 0.3636,
991
+ "step": 7600
992
+ },
993
+ {
994
+ "epoch": 30.8,
995
+ "grad_norm": 3.9435479640960693,
996
+ "learning_rate": 5e-05,
997
+ "loss": 0.3779,
998
+ "step": 7700
999
+ },
1000
+ {
1001
+ "epoch": 31.0,
1002
+ "eval_accuracy": 0.48568253968253966,
1003
+ "eval_loss": 4.214567184448242,
1004
+ "eval_runtime": 5.4963,
1005
+ "eval_samples_per_second": 90.971,
1006
+ "eval_steps_per_second": 11.462,
1007
+ "step": 7750
1008
+ },
1009
+ {
1010
+ "epoch": 31.0,
1011
+ "eval_exact_match": 13.6,
1012
+ "eval_f1": 20.160829725829732,
1013
+ "step": 7750
1014
+ },
1015
+ {
1016
+ "epoch": 31.2,
1017
+ "grad_norm": 1.4926341772079468,
1018
+ "learning_rate": 5e-05,
1019
+ "loss": 0.3721,
1020
+ "step": 7800
1021
+ },
1022
+ {
1023
+ "epoch": 31.6,
1024
+ "grad_norm": 1.510084629058838,
1025
+ "learning_rate": 5e-05,
1026
+ "loss": 0.3731,
1027
+ "step": 7900
1028
+ },
1029
+ {
1030
+ "epoch": 32.0,
1031
+ "grad_norm": 3.654505729675293,
1032
+ "learning_rate": 5e-05,
1033
+ "loss": 0.3785,
1034
+ "step": 8000
1035
+ },
1036
+ {
1037
+ "epoch": 32.0,
1038
+ "eval_accuracy": 0.48565079365079367,
1039
+ "eval_loss": 4.271883964538574,
1040
+ "eval_runtime": 5.5273,
1041
+ "eval_samples_per_second": 90.46,
1042
+ "eval_steps_per_second": 11.398,
1043
+ "step": 8000
1044
+ },
1045
+ {
1046
+ "epoch": 32.0,
1047
+ "eval_exact_match": 13.4,
1048
+ "eval_f1": 19.94316450216451,
1049
+ "step": 8000
1050
+ },
1051
+ {
1052
+ "epoch": 32.4,
1053
+ "grad_norm": 1.457706332206726,
1054
+ "learning_rate": 5e-05,
1055
+ "loss": 0.3574,
1056
+ "step": 8100
1057
+ },
1058
+ {
1059
+ "epoch": 32.8,
1060
+ "grad_norm": 2.00523042678833,
1061
+ "learning_rate": 5e-05,
1062
+ "loss": 0.3764,
1063
+ "step": 8200
1064
+ },
1065
+ {
1066
+ "epoch": 33.0,
1067
+ "eval_accuracy": 0.48765079365079367,
1068
+ "eval_loss": 4.250087738037109,
1069
+ "eval_runtime": 5.3196,
1070
+ "eval_samples_per_second": 93.992,
1071
+ "eval_steps_per_second": 11.843,
1072
+ "step": 8250
1073
+ },
1074
+ {
1075
+ "epoch": 33.0,
1076
+ "eval_exact_match": 14.4,
1077
+ "eval_f1": 20.652601842601847,
1078
+ "step": 8250
1079
+ },
1080
+ {
1081
+ "epoch": 33.2,
1082
+ "grad_norm": 2.976921319961548,
1083
+ "learning_rate": 5e-05,
1084
+ "loss": 0.3656,
1085
+ "step": 8300
1086
+ },
1087
+ {
1088
+ "epoch": 33.6,
1089
+ "grad_norm": 2.1466617584228516,
1090
+ "learning_rate": 5e-05,
1091
+ "loss": 0.3646,
1092
+ "step": 8400
1093
+ },
1094
+ {
1095
+ "epoch": 34.0,
1096
+ "grad_norm": 1.3529678583145142,
1097
+ "learning_rate": 5e-05,
1098
+ "loss": 0.3756,
1099
+ "step": 8500
1100
+ },
1101
+ {
1102
+ "epoch": 34.0,
1103
+ "eval_accuracy": 0.4862222222222222,
1104
+ "eval_loss": 4.241517543792725,
1105
+ "eval_runtime": 5.3044,
1106
+ "eval_samples_per_second": 94.262,
1107
+ "eval_steps_per_second": 11.877,
1108
+ "step": 8500
1109
+ },
1110
+ {
1111
+ "epoch": 34.0,
1112
+ "eval_exact_match": 14.0,
1113
+ "eval_f1": 20.943053613053618,
1114
+ "step": 8500
1115
+ },
1116
+ {
1117
+ "epoch": 34.4,
1118
+ "grad_norm": 1.9960662126541138,
1119
+ "learning_rate": 5e-05,
1120
+ "loss": 0.3545,
1121
+ "step": 8600
1122
+ },
1123
+ {
1124
+ "epoch": 34.8,
1125
+ "grad_norm": 1.7610321044921875,
1126
+ "learning_rate": 5e-05,
1127
+ "loss": 0.3687,
1128
+ "step": 8700
1129
+ },
1130
+ {
1131
+ "epoch": 35.0,
1132
+ "eval_accuracy": 0.4872063492063492,
1133
+ "eval_loss": 4.275779724121094,
1134
+ "eval_runtime": 5.1886,
1135
+ "eval_samples_per_second": 96.364,
1136
+ "eval_steps_per_second": 12.142,
1137
+ "step": 8750
1138
+ },
1139
+ {
1140
+ "epoch": 35.0,
1141
+ "eval_exact_match": 13.4,
1142
+ "eval_f1": 19.964098287333588,
1143
+ "step": 8750
1144
+ },
1145
+ {
1146
+ "epoch": 35.2,
1147
+ "grad_norm": 1.8763376474380493,
1148
+ "learning_rate": 5e-05,
1149
+ "loss": 0.3597,
1150
+ "step": 8800
1151
+ },
1152
+ {
1153
+ "epoch": 35.6,
1154
+ "grad_norm": 3.3443212509155273,
1155
+ "learning_rate": 5e-05,
1156
+ "loss": 0.3611,
1157
+ "step": 8900
1158
+ },
1159
+ {
1160
+ "epoch": 36.0,
1161
+ "grad_norm": 1.9623003005981445,
1162
+ "learning_rate": 5e-05,
1163
+ "loss": 0.3707,
1164
+ "step": 9000
1165
+ },
1166
+ {
1167
+ "epoch": 36.0,
1168
+ "eval_accuracy": 0.48558730158730157,
1169
+ "eval_loss": 4.269599437713623,
1170
+ "eval_runtime": 5.3187,
1171
+ "eval_samples_per_second": 94.007,
1172
+ "eval_steps_per_second": 11.845,
1173
+ "step": 9000
1174
+ },
1175
+ {
1176
+ "epoch": 36.0,
1177
+ "eval_exact_match": 14.0,
1178
+ "eval_f1": 20.398275613275622,
1179
+ "step": 9000
1180
+ },
1181
+ {
1182
+ "epoch": 36.4,
1183
+ "grad_norm": 1.66078782081604,
1184
+ "learning_rate": 5e-05,
1185
+ "loss": 0.3523,
1186
+ "step": 9100
1187
+ },
1188
+ {
1189
+ "epoch": 36.8,
1190
+ "grad_norm": 1.3792146444320679,
1191
+ "learning_rate": 5e-05,
1192
+ "loss": 0.3657,
1193
+ "step": 9200
1194
+ },
1195
+ {
1196
+ "epoch": 37.0,
1197
+ "eval_accuracy": 0.4855238095238095,
1198
+ "eval_loss": 4.312060356140137,
1199
+ "eval_runtime": 5.8808,
1200
+ "eval_samples_per_second": 85.023,
1201
+ "eval_steps_per_second": 10.713,
1202
+ "step": 9250
1203
+ },
1204
+ {
1205
+ "epoch": 37.0,
1206
+ "eval_exact_match": 14.0,
1207
+ "eval_f1": 20.292691197691205,
1208
+ "step": 9250
1209
+ },
1210
+ {
1211
+ "epoch": 37.2,
1212
+ "grad_norm": 1.2422215938568115,
1213
+ "learning_rate": 5e-05,
1214
+ "loss": 0.3569,
1215
+ "step": 9300
1216
+ },
1217
+ {
1218
+ "epoch": 37.6,
1219
+ "grad_norm": 2.2910890579223633,
1220
+ "learning_rate": 5e-05,
1221
+ "loss": 0.3581,
1222
+ "step": 9400
1223
+ },
1224
+ {
1225
+ "epoch": 38.0,
1226
+ "grad_norm": 1.8451393842697144,
1227
+ "learning_rate": 5e-05,
1228
+ "loss": 0.3682,
1229
+ "step": 9500
1230
+ },
1231
+ {
1232
+ "epoch": 38.0,
1233
+ "eval_accuracy": 0.48593650793650794,
1234
+ "eval_loss": 4.3264689445495605,
1235
+ "eval_runtime": 5.866,
1236
+ "eval_samples_per_second": 85.237,
1237
+ "eval_steps_per_second": 10.74,
1238
+ "step": 9500
1239
+ },
1240
+ {
1241
+ "epoch": 38.0,
1242
+ "eval_exact_match": 13.8,
1243
+ "eval_f1": 19.786422629657935,
1244
+ "step": 9500
1245
+ },
1246
+ {
1247
+ "epoch": 38.4,
1248
+ "grad_norm": 1.5858269929885864,
1249
+ "learning_rate": 5e-05,
1250
+ "loss": 0.3487,
1251
+ "step": 9600
1252
+ },
1253
+ {
1254
+ "epoch": 38.8,
1255
+ "grad_norm": 1.2035001516342163,
1256
+ "learning_rate": 5e-05,
1257
+ "loss": 0.3596,
1258
+ "step": 9700
1259
+ },
1260
+ {
1261
+ "epoch": 39.0,
1262
+ "eval_accuracy": 0.4847619047619048,
1263
+ "eval_loss": 4.394227027893066,
1264
+ "eval_runtime": 5.543,
1265
+ "eval_samples_per_second": 90.204,
1266
+ "eval_steps_per_second": 11.366,
1267
+ "step": 9750
1268
+ },
1269
+ {
1270
+ "epoch": 39.0,
1271
+ "eval_exact_match": 14.2,
1272
+ "eval_f1": 20.00341991341992,
1273
+ "step": 9750
1274
+ },
1275
+ {
1276
+ "epoch": 39.2,
1277
+ "grad_norm": 2.2688498497009277,
1278
+ "learning_rate": 5e-05,
1279
+ "loss": 0.3564,
1280
+ "step": 9800
1281
+ },
1282
+ {
1283
+ "epoch": 39.6,
1284
+ "grad_norm": 1.2643145322799683,
1285
+ "learning_rate": 5e-05,
1286
+ "loss": 0.3545,
1287
+ "step": 9900
1288
+ },
1289
+ {
1290
+ "epoch": 40.0,
1291
+ "grad_norm": 2.0884294509887695,
1292
+ "learning_rate": 5e-05,
1293
+ "loss": 0.3629,
1294
+ "step": 10000
1295
+ },
1296
+ {
1297
+ "epoch": 40.0,
1298
+ "eval_accuracy": 0.4852063492063492,
1299
+ "eval_loss": 4.303957462310791,
1300
+ "eval_runtime": 5.1423,
1301
+ "eval_samples_per_second": 97.233,
1302
+ "eval_steps_per_second": 12.251,
1303
+ "step": 10000
1304
+ },
1305
+ {
1306
+ "epoch": 40.0,
1307
+ "eval_exact_match": 13.0,
1308
+ "eval_f1": 18.57473686444276,
1309
+ "step": 10000
1310
+ },
1311
+ {
1312
+ "epoch": 40.4,
1313
+ "grad_norm": 1.535133719444275,
1314
+ "learning_rate": 5e-05,
1315
+ "loss": 0.3451,
1316
+ "step": 10100
1317
+ },
1318
+ {
1319
+ "epoch": 40.8,
1320
+ "grad_norm": 1.328994870185852,
1321
+ "learning_rate": 5e-05,
1322
+ "loss": 0.3596,
1323
+ "step": 10200
1324
+ },
1325
+ {
1326
+ "epoch": 41.0,
1327
+ "eval_accuracy": 0.48707936507936506,
1328
+ "eval_loss": 4.302526473999023,
1329
+ "eval_runtime": 5.2959,
1330
+ "eval_samples_per_second": 94.412,
1331
+ "eval_steps_per_second": 11.896,
1332
+ "step": 10250
1333
+ },
1334
+ {
1335
+ "epoch": 41.0,
1336
+ "eval_exact_match": 12.4,
1337
+ "eval_f1": 18.663619158619166,
1338
+ "step": 10250
1339
+ },
1340
+ {
1341
+ "epoch": 41.2,
1342
+ "grad_norm": 1.0413349866867065,
1343
+ "learning_rate": 5e-05,
1344
+ "loss": 0.3504,
1345
+ "step": 10300
1346
+ },
1347
+ {
1348
+ "epoch": 41.6,
1349
+ "grad_norm": 1.6248804330825806,
1350
+ "learning_rate": 5e-05,
1351
+ "loss": 0.3522,
1352
+ "step": 10400
1353
+ },
1354
+ {
1355
+ "epoch": 42.0,
1356
+ "grad_norm": 1.2970424890518188,
1357
+ "learning_rate": 5e-05,
1358
+ "loss": 0.3611,
1359
+ "step": 10500
1360
+ },
1361
+ {
1362
+ "epoch": 42.0,
1363
+ "eval_accuracy": 0.48568253968253966,
1364
+ "eval_loss": 4.312370300292969,
1365
+ "eval_runtime": 5.3427,
1366
+ "eval_samples_per_second": 93.585,
1367
+ "eval_steps_per_second": 11.792,
1368
+ "step": 10500
1369
+ },
1370
+ {
1371
+ "epoch": 42.0,
1372
+ "eval_exact_match": 13.6,
1373
+ "eval_f1": 19.535006390006394,
1374
+ "step": 10500
1375
+ },
1376
+ {
1377
+ "epoch": 42.4,
1378
+ "grad_norm": 1.0228374004364014,
1379
+ "learning_rate": 5e-05,
1380
+ "loss": 0.3424,
1381
+ "step": 10600
1382
+ },
1383
+ {
1384
+ "epoch": 42.8,
1385
+ "grad_norm": 1.2492530345916748,
1386
+ "learning_rate": 5e-05,
1387
+ "loss": 0.3552,
1388
+ "step": 10700
1389
+ },
1390
+ {
1391
+ "epoch": 43.0,
1392
+ "eval_accuracy": 0.4858095238095238,
1393
+ "eval_loss": 4.322521209716797,
1394
+ "eval_runtime": 5.4428,
1395
+ "eval_samples_per_second": 91.865,
1396
+ "eval_steps_per_second": 11.575,
1397
+ "step": 10750
1398
+ },
1399
+ {
1400
+ "epoch": 43.0,
1401
+ "eval_exact_match": 13.2,
1402
+ "eval_f1": 19.179316934316944,
1403
+ "step": 10750
1404
+ },
1405
+ {
1406
+ "epoch": 43.2,
1407
+ "grad_norm": 1.3973324298858643,
1408
+ "learning_rate": 5e-05,
1409
+ "loss": 0.3531,
1410
+ "step": 10800
1411
+ },
1412
+ {
1413
+ "epoch": 43.6,
1414
+ "grad_norm": 1.3194059133529663,
1415
+ "learning_rate": 5e-05,
1416
+ "loss": 0.3489,
1417
+ "step": 10900
1418
+ },
1419
+ {
1420
+ "epoch": 44.0,
1421
+ "grad_norm": 2.390622615814209,
1422
+ "learning_rate": 5e-05,
1423
+ "loss": 0.3593,
1424
+ "step": 11000
1425
+ },
1426
+ {
1427
+ "epoch": 44.0,
1428
+ "eval_accuracy": 0.48438095238095236,
1429
+ "eval_loss": 4.3125104904174805,
1430
+ "eval_runtime": 5.8599,
1431
+ "eval_samples_per_second": 85.326,
1432
+ "eval_steps_per_second": 10.751,
1433
+ "step": 11000
1434
+ },
1435
+ {
1436
+ "epoch": 44.0,
1437
+ "eval_exact_match": 14.2,
1438
+ "eval_f1": 20.37479271479273,
1439
+ "step": 11000
1440
+ },
1441
+ {
1442
+ "epoch": 44.4,
1443
+ "grad_norm": 2.320546865463257,
1444
+ "learning_rate": 5e-05,
1445
+ "loss": 0.3426,
1446
+ "step": 11100
1447
+ },
1448
+ {
1449
+ "epoch": 44.8,
1450
+ "grad_norm": 1.38444185256958,
1451
+ "learning_rate": 5e-05,
1452
+ "loss": 0.3551,
1453
+ "step": 11200
1454
+ },
1455
+ {
1456
+ "epoch": 45.0,
1457
+ "eval_accuracy": 0.4860952380952381,
1458
+ "eval_loss": 4.3344855308532715,
1459
+ "eval_runtime": 5.152,
1460
+ "eval_samples_per_second": 97.05,
1461
+ "eval_steps_per_second": 12.228,
1462
+ "step": 11250
1463
+ },
1464
+ {
1465
+ "epoch": 45.0,
1466
+ "eval_exact_match": 14.4,
1467
+ "eval_f1": 20.7534892884893,
1468
+ "step": 11250
1469
+ },
1470
+ {
1471
+ "epoch": 45.2,
1472
+ "grad_norm": 2.1086132526397705,
1473
+ "learning_rate": 5e-05,
1474
+ "loss": 0.3451,
1475
+ "step": 11300
1476
+ },
1477
+ {
1478
+ "epoch": 45.6,
1479
+ "grad_norm": 5.763082027435303,
1480
+ "learning_rate": 5e-05,
1481
+ "loss": 0.3512,
1482
+ "step": 11400
1483
+ },
1484
+ {
1485
+ "epoch": 46.0,
1486
+ "grad_norm": 1.1683377027511597,
1487
+ "learning_rate": 5e-05,
1488
+ "loss": 0.3538,
1489
+ "step": 11500
1490
+ },
1491
+ {
1492
+ "epoch": 46.0,
1493
+ "eval_accuracy": 0.48634920634920636,
1494
+ "eval_loss": 4.335656642913818,
1495
+ "eval_runtime": 5.1561,
1496
+ "eval_samples_per_second": 96.973,
1497
+ "eval_steps_per_second": 12.219,
1498
+ "step": 11500
1499
+ },
1500
+ {
1501
+ "epoch": 46.0,
1502
+ "eval_exact_match": 12.6,
1503
+ "eval_f1": 19.21051393051394,
1504
+ "step": 11500
1505
+ },
1506
+ {
1507
+ "epoch": 46.4,
1508
+ "grad_norm": 1.80911123752594,
1509
+ "learning_rate": 5e-05,
1510
+ "loss": 0.3426,
1511
+ "step": 11600
1512
+ },
1513
+ {
1514
+ "epoch": 46.8,
1515
+ "grad_norm": 1.8256689310073853,
1516
+ "learning_rate": 5e-05,
1517
+ "loss": 0.3497,
1518
+ "step": 11700
1519
+ },
1520
+ {
1521
+ "epoch": 47.0,
1522
+ "eval_accuracy": 0.4864126984126984,
1523
+ "eval_loss": 4.379979610443115,
1524
+ "eval_runtime": 5.3692,
1525
+ "eval_samples_per_second": 93.123,
1526
+ "eval_steps_per_second": 11.733,
1527
+ "step": 11750
1528
+ },
1529
+ {
1530
+ "epoch": 47.0,
1531
+ "eval_exact_match": 13.8,
1532
+ "eval_f1": 20.139399652634953,
1533
+ "step": 11750
1534
+ },
1535
+ {
1536
+ "epoch": 47.2,
1537
+ "grad_norm": 0.8401724696159363,
1538
+ "learning_rate": 5e-05,
1539
+ "loss": 0.3448,
1540
+ "step": 11800
1541
+ },
1542
+ {
1543
+ "epoch": 47.6,
1544
+ "grad_norm": 1.1846015453338623,
1545
+ "learning_rate": 5e-05,
1546
+ "loss": 0.3485,
1547
+ "step": 11900
1548
+ },
1549
+ {
1550
+ "epoch": 48.0,
1551
+ "grad_norm": 1.4425411224365234,
1552
+ "learning_rate": 5e-05,
1553
+ "loss": 0.3535,
1554
+ "step": 12000
1555
+ },
1556
+ {
1557
+ "epoch": 48.0,
1558
+ "eval_accuracy": 0.4866984126984127,
1559
+ "eval_loss": 4.32601261138916,
1560
+ "eval_runtime": 5.1934,
1561
+ "eval_samples_per_second": 96.276,
1562
+ "eval_steps_per_second": 12.131,
1563
+ "step": 12000
1564
+ },
1565
+ {
1566
+ "epoch": 48.0,
1567
+ "eval_exact_match": 12.8,
1568
+ "eval_f1": 20.51256981580512,
1569
+ "step": 12000
1570
+ }
1571
+ ],
1572
+ "logging_steps": 100,
1573
+ "max_steps": 12500,
1574
+ "num_input_tokens_seen": 0,
1575
+ "num_train_epochs": 50,
1576
+ "save_steps": 500,
1577
+ "stateful_callbacks": {
1578
+ "TrainerControl": {
1579
+ "args": {
1580
+ "should_epoch_stop": false,
1581
+ "should_evaluate": false,
1582
+ "should_log": false,
1583
+ "should_save": true,
1584
+ "should_training_stop": false
1585
+ },
1586
+ "attributes": {}
1587
+ }
1588
+ },
1589
+ "total_flos": 3.104492632366121e+17,
1590
+ "train_batch_size": 1,
1591
+ "trial_name": null,
1592
+ "trial_params": null
1593
+ }
checkpoint-12000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ede3565a5658e9b8743d6efdd527fd23709623a130469c81f1191d07c33fbcb1
3
+ size 5304
checkpoint-12000/vocab.json ADDED
The diff for this file is too large to render. See raw diff