tyzhu commited on
Commit
c684f29
1 Parent(s): 37f08c3

Training in progress, epoch 42, checkpoint

Browse files
checkpoint-8041/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-8041/adapter_config.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
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-8041/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0b7ef800a9a65bc9d039f1db35c96ad59705bf76e042453eaf1e35436e1d29b9
3
+ size 143269386
checkpoint-8041/added_tokens.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "</s>": 2,
3
+ "<s>": 1,
4
+ "<unk>": 0
5
+ }
checkpoint-8041/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e196274eab48762ad0ed0a69d7aa1eea240f627c6ad5dd5b39a75a6fb8576cc
3
+ size 286590610
checkpoint-8041/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dd7e5afa67308628615b6c51fa64f476e38c6b7ef02f3da208bd32b32dfa5f90
3
+ size 15024
checkpoint-8041/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:148ab48d11be80ef1b696954300805d3ede75353a003ebc3435e92f365195c2c
3
+ size 15024
checkpoint-8041/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13ad95dcea59cc73f3b6417fbebef728ec347207c77a3177c20eb0886721e7e6
3
+ size 15024
checkpoint-8041/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:efe77ed374d7632a79d75f6c21faebdbe30edb1ee0fbff7e69bd1716fbf894e9
3
+ size 15024
checkpoint-8041/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42e43a05b1b1908c959309dea3945111bd5caba68725f3b1f002cabae993a7a2
3
+ size 1064
checkpoint-8041/special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": "</s>",
5
+ "unk_token": "<unk>"
6
+ }
checkpoint-8041/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-8041/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-8041/tokenizer_config.json ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<unk>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<s>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "</s>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ }
27
+ },
28
+ "additional_special_tokens": [],
29
+ "bos_token": "<s>",
30
+ "clean_up_tokenization_spaces": false,
31
+ "eos_token": "</s>",
32
+ "legacy": false,
33
+ "model_max_length": 1000000000000000019884624838656,
34
+ "pad_token": "</s>",
35
+ "padding_side": "left",
36
+ "sp_model_kwargs": {},
37
+ "tokenizer_class": "LlamaTokenizer",
38
+ "unk_token": "<unk>",
39
+ "use_default_system_prompt": true
40
+ }
checkpoint-8041/trainer_state.json ADDED
@@ -0,0 +1,1316 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 42.99733333333333,
5
+ "eval_steps": 500,
6
+ "global_step": 8041,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.53,
13
+ "learning_rate": 3e-05,
14
+ "loss": 1.4597,
15
+ "step": 100
16
+ },
17
+ {
18
+ "epoch": 1.0,
19
+ "eval_accuracy": 0.6618980392156862,
20
+ "eval_loss": 1.2718806266784668,
21
+ "eval_runtime": 7.6194,
22
+ "eval_samples_per_second": 65.622,
23
+ "eval_steps_per_second": 8.268,
24
+ "step": 187
25
+ },
26
+ {
27
+ "epoch": 1.0,
28
+ "eval_exact_match": 22.4,
29
+ "eval_f1": 30.941064425770318,
30
+ "eval_qa_bleu": 7.097403515221558,
31
+ "eval_qa_exact_match": 0.196,
32
+ "eval_recite_bleu": 13.258373740102225,
33
+ "eval_recite_exact_match": 0.0,
34
+ "step": 187
35
+ },
36
+ {
37
+ "epoch": 1.07,
38
+ "learning_rate": 3e-05,
39
+ "loss": 1.2905,
40
+ "step": 200
41
+ },
42
+ {
43
+ "epoch": 1.6,
44
+ "learning_rate": 3e-05,
45
+ "loss": 1.2157,
46
+ "step": 300
47
+ },
48
+ {
49
+ "epoch": 2.0,
50
+ "eval_accuracy": 0.6670039215686274,
51
+ "eval_loss": 1.206207036972046,
52
+ "eval_runtime": 7.6793,
53
+ "eval_samples_per_second": 65.11,
54
+ "eval_steps_per_second": 8.204,
55
+ "step": 375
56
+ },
57
+ {
58
+ "epoch": 2.0,
59
+ "eval_exact_match": 3.2,
60
+ "eval_f1": 3.96976236704394,
61
+ "eval_qa_bleu": 0.20484366294597936,
62
+ "eval_qa_exact_match": 0.026,
63
+ "eval_recite_bleu": 2.5778714769111824e-05,
64
+ "eval_recite_exact_match": 0.0,
65
+ "step": 375
66
+ },
67
+ {
68
+ "epoch": 2.13,
69
+ "learning_rate": 3e-05,
70
+ "loss": 1.1852,
71
+ "step": 400
72
+ },
73
+ {
74
+ "epoch": 2.67,
75
+ "learning_rate": 3e-05,
76
+ "loss": 1.1861,
77
+ "step": 500
78
+ },
79
+ {
80
+ "epoch": 3.0,
81
+ "eval_accuracy": 0.667513725490196,
82
+ "eval_loss": 1.19828462600708,
83
+ "eval_runtime": 7.1425,
84
+ "eval_samples_per_second": 70.004,
85
+ "eval_steps_per_second": 8.82,
86
+ "step": 562
87
+ },
88
+ {
89
+ "epoch": 3.0,
90
+ "eval_exact_match": 4.4,
91
+ "eval_f1": 6.333660342551587,
92
+ "eval_qa_bleu": 0.522391008398391,
93
+ "eval_qa_exact_match": 0.028,
94
+ "eval_recite_bleu": 0.027569975851374308,
95
+ "eval_recite_exact_match": 0.0,
96
+ "step": 562
97
+ },
98
+ {
99
+ "epoch": 3.2,
100
+ "learning_rate": 3e-05,
101
+ "loss": 1.1722,
102
+ "step": 600
103
+ },
104
+ {
105
+ "epoch": 3.73,
106
+ "learning_rate": 3e-05,
107
+ "loss": 1.1445,
108
+ "step": 700
109
+ },
110
+ {
111
+ "epoch": 4.0,
112
+ "eval_accuracy": 0.6672,
113
+ "eval_loss": 1.197709560394287,
114
+ "eval_runtime": 7.7427,
115
+ "eval_samples_per_second": 64.577,
116
+ "eval_steps_per_second": 8.137,
117
+ "step": 750
118
+ },
119
+ {
120
+ "epoch": 4.0,
121
+ "eval_exact_match": 7.4,
122
+ "eval_f1": 10.950016966795278,
123
+ "eval_qa_bleu": 0.6700083846206139,
124
+ "eval_qa_exact_match": 0.052,
125
+ "eval_recite_bleu": 1.259776140465531,
126
+ "eval_recite_exact_match": 0.0,
127
+ "step": 750
128
+ },
129
+ {
130
+ "epoch": 4.27,
131
+ "learning_rate": 3e-05,
132
+ "loss": 1.1018,
133
+ "step": 800
134
+ },
135
+ {
136
+ "epoch": 4.8,
137
+ "learning_rate": 3e-05,
138
+ "loss": 1.1,
139
+ "step": 900
140
+ },
141
+ {
142
+ "epoch": 5.0,
143
+ "eval_accuracy": 0.6664941176470588,
144
+ "eval_loss": 1.2078713178634644,
145
+ "eval_runtime": 7.7721,
146
+ "eval_samples_per_second": 64.333,
147
+ "eval_steps_per_second": 8.106,
148
+ "step": 937
149
+ },
150
+ {
151
+ "epoch": 5.0,
152
+ "eval_exact_match": 6.6,
153
+ "eval_f1": 10.12048028049545,
154
+ "eval_qa_bleu": 0.8314625442591895,
155
+ "eval_qa_exact_match": 0.05,
156
+ "eval_recite_bleu": 0.8030246920772521,
157
+ "eval_recite_exact_match": 0.0,
158
+ "step": 937
159
+ },
160
+ {
161
+ "epoch": 5.33,
162
+ "learning_rate": 3e-05,
163
+ "loss": 1.0666,
164
+ "step": 1000
165
+ },
166
+ {
167
+ "epoch": 5.87,
168
+ "learning_rate": 3e-05,
169
+ "loss": 1.0439,
170
+ "step": 1100
171
+ },
172
+ {
173
+ "epoch": 6.0,
174
+ "eval_accuracy": 0.6649725490196079,
175
+ "eval_loss": 1.22383713722229,
176
+ "eval_runtime": 7.6639,
177
+ "eval_samples_per_second": 65.241,
178
+ "eval_steps_per_second": 8.22,
179
+ "step": 1125
180
+ },
181
+ {
182
+ "epoch": 6.0,
183
+ "eval_exact_match": 9.0,
184
+ "eval_f1": 14.19791549030679,
185
+ "eval_qa_bleu": 2.7960558519683327,
186
+ "eval_qa_exact_match": 0.07,
187
+ "eval_recite_bleu": 2.082318255561628,
188
+ "eval_recite_exact_match": 0.0,
189
+ "step": 1125
190
+ },
191
+ {
192
+ "epoch": 6.4,
193
+ "learning_rate": 3e-05,
194
+ "loss": 1.0029,
195
+ "step": 1200
196
+ },
197
+ {
198
+ "epoch": 6.93,
199
+ "learning_rate": 3e-05,
200
+ "loss": 0.9888,
201
+ "step": 1300
202
+ },
203
+ {
204
+ "epoch": 7.0,
205
+ "eval_accuracy": 0.6638117647058823,
206
+ "eval_loss": 1.2457345724105835,
207
+ "eval_runtime": 6.7577,
208
+ "eval_samples_per_second": 73.99,
209
+ "eval_steps_per_second": 9.323,
210
+ "step": 1312
211
+ },
212
+ {
213
+ "epoch": 7.0,
214
+ "eval_exact_match": 10.6,
215
+ "eval_f1": 15.30586844512479,
216
+ "eval_qa_bleu": 0.94556412718174,
217
+ "eval_qa_exact_match": 0.086,
218
+ "eval_recite_bleu": 3.1558565534273777,
219
+ "eval_recite_exact_match": 0.0,
220
+ "step": 1312
221
+ },
222
+ {
223
+ "epoch": 7.47,
224
+ "learning_rate": 3e-05,
225
+ "loss": 0.9453,
226
+ "step": 1400
227
+ },
228
+ {
229
+ "epoch": 8.0,
230
+ "learning_rate": 3e-05,
231
+ "loss": 0.9364,
232
+ "step": 1500
233
+ },
234
+ {
235
+ "epoch": 8.0,
236
+ "eval_accuracy": 0.6615921568627451,
237
+ "eval_loss": 1.2816376686096191,
238
+ "eval_runtime": 7.5902,
239
+ "eval_samples_per_second": 65.875,
240
+ "eval_steps_per_second": 8.3,
241
+ "step": 1500
242
+ },
243
+ {
244
+ "epoch": 8.0,
245
+ "eval_exact_match": 9.2,
246
+ "eval_f1": 12.989213707726288,
247
+ "eval_qa_bleu": 1.9797940311992752,
248
+ "eval_qa_exact_match": 0.074,
249
+ "eval_recite_bleu": 2.551621610315755,
250
+ "eval_recite_exact_match": 0.0,
251
+ "step": 1500
252
+ },
253
+ {
254
+ "epoch": 8.53,
255
+ "learning_rate": 3e-05,
256
+ "loss": 0.8889,
257
+ "step": 1600
258
+ },
259
+ {
260
+ "epoch": 9.0,
261
+ "eval_accuracy": 0.6606196078431372,
262
+ "eval_loss": 1.3036144971847534,
263
+ "eval_runtime": 7.4845,
264
+ "eval_samples_per_second": 66.805,
265
+ "eval_steps_per_second": 8.417,
266
+ "step": 1687
267
+ },
268
+ {
269
+ "epoch": 9.0,
270
+ "eval_exact_match": 14.2,
271
+ "eval_f1": 19.66897045942666,
272
+ "eval_qa_bleu": 3.3530752528407004,
273
+ "eval_qa_exact_match": 0.112,
274
+ "eval_recite_bleu": 5.3293362575452425,
275
+ "eval_recite_exact_match": 0.002,
276
+ "step": 1687
277
+ },
278
+ {
279
+ "epoch": 9.07,
280
+ "learning_rate": 3e-05,
281
+ "loss": 0.8853,
282
+ "step": 1700
283
+ },
284
+ {
285
+ "epoch": 9.6,
286
+ "learning_rate": 3e-05,
287
+ "loss": 0.8373,
288
+ "step": 1800
289
+ },
290
+ {
291
+ "epoch": 10.0,
292
+ "eval_accuracy": 0.6587450980392157,
293
+ "eval_loss": 1.3337457180023193,
294
+ "eval_runtime": 6.6361,
295
+ "eval_samples_per_second": 75.345,
296
+ "eval_steps_per_second": 9.493,
297
+ "step": 1875
298
+ },
299
+ {
300
+ "epoch": 10.0,
301
+ "eval_exact_match": 14.4,
302
+ "eval_f1": 21.20886352105531,
303
+ "eval_qa_bleu": 2.50339041197013,
304
+ "eval_qa_exact_match": 0.106,
305
+ "eval_recite_bleu": 6.648771889251858,
306
+ "eval_recite_exact_match": 0.0,
307
+ "step": 1875
308
+ },
309
+ {
310
+ "epoch": 10.13,
311
+ "learning_rate": 3e-05,
312
+ "loss": 0.8353,
313
+ "step": 1900
314
+ },
315
+ {
316
+ "epoch": 10.67,
317
+ "learning_rate": 3e-05,
318
+ "loss": 0.7949,
319
+ "step": 2000
320
+ },
321
+ {
322
+ "epoch": 11.0,
323
+ "eval_accuracy": 0.6575058823529412,
324
+ "eval_loss": 1.367775797843933,
325
+ "eval_runtime": 7.4976,
326
+ "eval_samples_per_second": 66.688,
327
+ "eval_steps_per_second": 8.403,
328
+ "step": 2062
329
+ },
330
+ {
331
+ "epoch": 11.0,
332
+ "eval_exact_match": 15.4,
333
+ "eval_f1": 21.670913900913916,
334
+ "eval_qa_bleu": 4.131625289761948,
335
+ "eval_qa_exact_match": 0.104,
336
+ "eval_recite_bleu": 8.111027123751208,
337
+ "eval_recite_exact_match": 0.0,
338
+ "step": 2062
339
+ },
340
+ {
341
+ "epoch": 11.2,
342
+ "learning_rate": 3e-05,
343
+ "loss": 0.7758,
344
+ "step": 2100
345
+ },
346
+ {
347
+ "epoch": 11.73,
348
+ "learning_rate": 3e-05,
349
+ "loss": 0.7539,
350
+ "step": 2200
351
+ },
352
+ {
353
+ "epoch": 12.0,
354
+ "eval_accuracy": 0.6553725490196078,
355
+ "eval_loss": 1.3984336853027344,
356
+ "eval_runtime": 7.6892,
357
+ "eval_samples_per_second": 65.026,
358
+ "eval_steps_per_second": 8.193,
359
+ "step": 2250
360
+ },
361
+ {
362
+ "epoch": 12.0,
363
+ "eval_exact_match": 16.2,
364
+ "eval_f1": 23.074922854922868,
365
+ "eval_qa_bleu": 10.80394366395148,
366
+ "eval_qa_exact_match": 0.132,
367
+ "eval_recite_bleu": 9.158207459098465,
368
+ "eval_recite_exact_match": 0.002,
369
+ "step": 2250
370
+ },
371
+ {
372
+ "epoch": 12.27,
373
+ "learning_rate": 3e-05,
374
+ "loss": 0.7415,
375
+ "step": 2300
376
+ },
377
+ {
378
+ "epoch": 12.8,
379
+ "learning_rate": 3e-05,
380
+ "loss": 0.7133,
381
+ "step": 2400
382
+ },
383
+ {
384
+ "epoch": 13.0,
385
+ "eval_accuracy": 0.6538196078431373,
386
+ "eval_loss": 1.4471032619476318,
387
+ "eval_runtime": 6.7177,
388
+ "eval_samples_per_second": 74.431,
389
+ "eval_steps_per_second": 9.378,
390
+ "step": 2437
391
+ },
392
+ {
393
+ "epoch": 13.0,
394
+ "eval_exact_match": 16.4,
395
+ "eval_f1": 23.343965796845062,
396
+ "eval_qa_bleu": 4.473837470175788,
397
+ "eval_qa_exact_match": 0.136,
398
+ "eval_recite_bleu": 10.58390244263022,
399
+ "eval_recite_exact_match": 0.002,
400
+ "step": 2437
401
+ },
402
+ {
403
+ "epoch": 13.33,
404
+ "learning_rate": 3e-05,
405
+ "loss": 0.682,
406
+ "step": 2500
407
+ },
408
+ {
409
+ "epoch": 13.87,
410
+ "learning_rate": 3e-05,
411
+ "loss": 0.6704,
412
+ "step": 2600
413
+ },
414
+ {
415
+ "epoch": 14.0,
416
+ "eval_accuracy": 0.6524862745098039,
417
+ "eval_loss": 1.4829775094985962,
418
+ "eval_runtime": 6.6339,
419
+ "eval_samples_per_second": 75.37,
420
+ "eval_steps_per_second": 9.497,
421
+ "step": 2625
422
+ },
423
+ {
424
+ "epoch": 14.0,
425
+ "eval_exact_match": 15.6,
426
+ "eval_f1": 23.615572621393074,
427
+ "eval_qa_bleu": 7.563435997390864,
428
+ "eval_qa_exact_match": 0.13,
429
+ "eval_recite_bleu": 11.120038652898813,
430
+ "eval_recite_exact_match": 0.008,
431
+ "step": 2625
432
+ },
433
+ {
434
+ "epoch": 14.4,
435
+ "learning_rate": 3e-05,
436
+ "loss": 0.6412,
437
+ "step": 2700
438
+ },
439
+ {
440
+ "epoch": 14.93,
441
+ "learning_rate": 3e-05,
442
+ "loss": 0.6436,
443
+ "step": 2800
444
+ },
445
+ {
446
+ "epoch": 15.0,
447
+ "eval_accuracy": 0.6507529411764705,
448
+ "eval_loss": 1.524308681488037,
449
+ "eval_runtime": 6.6481,
450
+ "eval_samples_per_second": 75.209,
451
+ "eval_steps_per_second": 9.476,
452
+ "step": 2812
453
+ },
454
+ {
455
+ "epoch": 15.0,
456
+ "eval_exact_match": 17.8,
457
+ "eval_f1": 27.00021164021165,
458
+ "eval_qa_bleu": 7.90349748901623,
459
+ "eval_qa_exact_match": 0.15,
460
+ "eval_recite_bleu": 12.277483381668418,
461
+ "eval_recite_exact_match": 0.008,
462
+ "step": 2812
463
+ },
464
+ {
465
+ "epoch": 15.47,
466
+ "learning_rate": 3e-05,
467
+ "loss": 0.5969,
468
+ "step": 2900
469
+ },
470
+ {
471
+ "epoch": 16.0,
472
+ "learning_rate": 3e-05,
473
+ "loss": 0.6073,
474
+ "step": 3000
475
+ },
476
+ {
477
+ "epoch": 16.0,
478
+ "eval_accuracy": 0.6499764705882353,
479
+ "eval_loss": 1.5640995502471924,
480
+ "eval_runtime": 7.7119,
481
+ "eval_samples_per_second": 64.835,
482
+ "eval_steps_per_second": 8.169,
483
+ "step": 3000
484
+ },
485
+ {
486
+ "epoch": 16.0,
487
+ "eval_exact_match": 19.2,
488
+ "eval_f1": 27.747719298245627,
489
+ "eval_qa_bleu": 9.793788404463436,
490
+ "eval_qa_exact_match": 0.14,
491
+ "eval_recite_bleu": 12.835117221544726,
492
+ "eval_recite_exact_match": 0.01,
493
+ "step": 3000
494
+ },
495
+ {
496
+ "epoch": 16.53,
497
+ "learning_rate": 3e-05,
498
+ "loss": 0.5567,
499
+ "step": 3100
500
+ },
501
+ {
502
+ "epoch": 17.0,
503
+ "eval_accuracy": 0.6486588235294117,
504
+ "eval_loss": 1.6058622598648071,
505
+ "eval_runtime": 7.7727,
506
+ "eval_samples_per_second": 64.328,
507
+ "eval_steps_per_second": 8.105,
508
+ "step": 3187
509
+ },
510
+ {
511
+ "epoch": 17.0,
512
+ "eval_exact_match": 19.6,
513
+ "eval_f1": 26.446776949066113,
514
+ "eval_qa_bleu": 10.294098070146385,
515
+ "eval_qa_exact_match": 0.156,
516
+ "eval_recite_bleu": 12.34815708077889,
517
+ "eval_recite_exact_match": 0.004,
518
+ "step": 3187
519
+ },
520
+ {
521
+ "epoch": 17.07,
522
+ "learning_rate": 3e-05,
523
+ "loss": 0.5654,
524
+ "step": 3200
525
+ },
526
+ {
527
+ "epoch": 17.6,
528
+ "learning_rate": 3e-05,
529
+ "loss": 0.5204,
530
+ "step": 3300
531
+ },
532
+ {
533
+ "epoch": 18.0,
534
+ "eval_accuracy": 0.6472705882352942,
535
+ "eval_loss": 1.6425700187683105,
536
+ "eval_runtime": 7.1347,
537
+ "eval_samples_per_second": 70.08,
538
+ "eval_steps_per_second": 8.83,
539
+ "step": 3375
540
+ },
541
+ {
542
+ "epoch": 18.0,
543
+ "eval_exact_match": 18.4,
544
+ "eval_f1": 26.083665223665236,
545
+ "eval_qa_bleu": 12.978225559731756,
546
+ "eval_qa_exact_match": 0.142,
547
+ "eval_recite_bleu": 12.629003802964625,
548
+ "eval_recite_exact_match": 0.006,
549
+ "step": 3375
550
+ },
551
+ {
552
+ "epoch": 18.13,
553
+ "learning_rate": 3e-05,
554
+ "loss": 0.5348,
555
+ "step": 3400
556
+ },
557
+ {
558
+ "epoch": 18.67,
559
+ "learning_rate": 3e-05,
560
+ "loss": 0.5019,
561
+ "step": 3500
562
+ },
563
+ {
564
+ "epoch": 19.0,
565
+ "eval_accuracy": 0.6459843137254901,
566
+ "eval_loss": 1.6983567476272583,
567
+ "eval_runtime": 7.0372,
568
+ "eval_samples_per_second": 71.051,
569
+ "eval_steps_per_second": 8.952,
570
+ "step": 3562
571
+ },
572
+ {
573
+ "epoch": 19.0,
574
+ "eval_exact_match": 19.4,
575
+ "eval_f1": 27.658198757763987,
576
+ "eval_qa_bleu": 10.765170656210177,
577
+ "eval_qa_exact_match": 0.146,
578
+ "eval_recite_bleu": 12.874816956011239,
579
+ "eval_recite_exact_match": 0.006,
580
+ "step": 3562
581
+ },
582
+ {
583
+ "epoch": 19.2,
584
+ "learning_rate": 3e-05,
585
+ "loss": 0.4888,
586
+ "step": 3600
587
+ },
588
+ {
589
+ "epoch": 19.73,
590
+ "learning_rate": 3e-05,
591
+ "loss": 0.4676,
592
+ "step": 3700
593
+ },
594
+ {
595
+ "epoch": 20.0,
596
+ "eval_accuracy": 0.6450431372549019,
597
+ "eval_loss": 1.74748957157135,
598
+ "eval_runtime": 6.6386,
599
+ "eval_samples_per_second": 75.317,
600
+ "eval_steps_per_second": 9.49,
601
+ "step": 3750
602
+ },
603
+ {
604
+ "epoch": 20.0,
605
+ "eval_exact_match": 17.4,
606
+ "eval_f1": 25.5869051909052,
607
+ "eval_qa_bleu": 7.975401244418332,
608
+ "eval_qa_exact_match": 0.144,
609
+ "eval_recite_bleu": 11.978547399136012,
610
+ "eval_recite_exact_match": 0.004,
611
+ "step": 3750
612
+ },
613
+ {
614
+ "epoch": 20.27,
615
+ "learning_rate": 3e-05,
616
+ "loss": 0.4528,
617
+ "step": 3800
618
+ },
619
+ {
620
+ "epoch": 20.8,
621
+ "learning_rate": 3e-05,
622
+ "loss": 0.4408,
623
+ "step": 3900
624
+ },
625
+ {
626
+ "epoch": 21.0,
627
+ "eval_accuracy": 0.6450274509803922,
628
+ "eval_loss": 1.7759969234466553,
629
+ "eval_runtime": 7.7092,
630
+ "eval_samples_per_second": 64.857,
631
+ "eval_steps_per_second": 8.172,
632
+ "step": 3937
633
+ },
634
+ {
635
+ "epoch": 21.0,
636
+ "eval_exact_match": 19.0,
637
+ "eval_f1": 27.044218949435447,
638
+ "eval_qa_bleu": 5.961857744132503,
639
+ "eval_qa_exact_match": 0.158,
640
+ "eval_recite_bleu": 12.078600703350125,
641
+ "eval_recite_exact_match": 0.004,
642
+ "step": 3937
643
+ },
644
+ {
645
+ "epoch": 21.33,
646
+ "learning_rate": 3e-05,
647
+ "loss": 0.4191,
648
+ "step": 4000
649
+ },
650
+ {
651
+ "epoch": 21.87,
652
+ "learning_rate": 3e-05,
653
+ "loss": 0.4217,
654
+ "step": 4100
655
+ },
656
+ {
657
+ "epoch": 22.0,
658
+ "eval_accuracy": 0.644078431372549,
659
+ "eval_loss": 1.7986315488815308,
660
+ "eval_runtime": 7.6507,
661
+ "eval_samples_per_second": 65.353,
662
+ "eval_steps_per_second": 8.235,
663
+ "step": 4125
664
+ },
665
+ {
666
+ "epoch": 22.0,
667
+ "eval_exact_match": 18.2,
668
+ "eval_f1": 25.124761904761915,
669
+ "eval_qa_bleu": 15.392825669562168,
670
+ "eval_qa_exact_match": 0.14,
671
+ "eval_recite_bleu": 11.50341909728548,
672
+ "eval_recite_exact_match": 0.006,
673
+ "step": 4125
674
+ },
675
+ {
676
+ "epoch": 22.4,
677
+ "learning_rate": 3e-05,
678
+ "loss": 0.393,
679
+ "step": 4200
680
+ },
681
+ {
682
+ "epoch": 22.93,
683
+ "learning_rate": 3e-05,
684
+ "loss": 0.3886,
685
+ "step": 4300
686
+ },
687
+ {
688
+ "epoch": 23.0,
689
+ "eval_accuracy": 0.6439529411764706,
690
+ "eval_loss": 1.8378911018371582,
691
+ "eval_runtime": 7.4779,
692
+ "eval_samples_per_second": 66.864,
693
+ "eval_steps_per_second": 8.425,
694
+ "step": 4312
695
+ },
696
+ {
697
+ "epoch": 23.0,
698
+ "eval_exact_match": 14.8,
699
+ "eval_f1": 21.914444444444456,
700
+ "eval_qa_bleu": 10.979013053975578,
701
+ "eval_qa_exact_match": 0.128,
702
+ "eval_recite_bleu": 11.265801595707984,
703
+ "eval_recite_exact_match": 0.004,
704
+ "step": 4312
705
+ },
706
+ {
707
+ "epoch": 23.47,
708
+ "learning_rate": 3e-05,
709
+ "loss": 0.3605,
710
+ "step": 4400
711
+ },
712
+ {
713
+ "epoch": 24.0,
714
+ "learning_rate": 3e-05,
715
+ "loss": 0.3729,
716
+ "step": 4500
717
+ },
718
+ {
719
+ "epoch": 24.0,
720
+ "eval_accuracy": 0.6420470588235294,
721
+ "eval_loss": 1.927443504333496,
722
+ "eval_runtime": 7.4721,
723
+ "eval_samples_per_second": 66.915,
724
+ "eval_steps_per_second": 8.431,
725
+ "step": 4500
726
+ },
727
+ {
728
+ "epoch": 24.0,
729
+ "eval_exact_match": 15.6,
730
+ "eval_f1": 22.679302325581407,
731
+ "eval_qa_bleu": 7.33716058981409,
732
+ "eval_qa_exact_match": 0.132,
733
+ "eval_recite_bleu": 11.564503717897392,
734
+ "eval_recite_exact_match": 0.002,
735
+ "step": 4500
736
+ },
737
+ {
738
+ "epoch": 24.53,
739
+ "learning_rate": 3e-05,
740
+ "loss": 0.3316,
741
+ "step": 4600
742
+ },
743
+ {
744
+ "epoch": 25.0,
745
+ "eval_accuracy": 0.641921568627451,
746
+ "eval_loss": 1.9378185272216797,
747
+ "eval_runtime": 7.6806,
748
+ "eval_samples_per_second": 65.099,
749
+ "eval_steps_per_second": 8.203,
750
+ "step": 4687
751
+ },
752
+ {
753
+ "epoch": 25.0,
754
+ "eval_exact_match": 16.8,
755
+ "eval_f1": 23.502857142857145,
756
+ "eval_qa_bleu": 8.430315553769542,
757
+ "eval_qa_exact_match": 0.13,
758
+ "eval_recite_bleu": 11.821782294825612,
759
+ "eval_recite_exact_match": 0.006,
760
+ "step": 4687
761
+ },
762
+ {
763
+ "epoch": 25.07,
764
+ "learning_rate": 3e-05,
765
+ "loss": 0.3454,
766
+ "step": 4700
767
+ },
768
+ {
769
+ "epoch": 25.6,
770
+ "learning_rate": 3e-05,
771
+ "loss": 0.313,
772
+ "step": 4800
773
+ },
774
+ {
775
+ "epoch": 26.0,
776
+ "eval_accuracy": 0.6410274509803922,
777
+ "eval_loss": 2.0155365467071533,
778
+ "eval_runtime": 6.6459,
779
+ "eval_samples_per_second": 75.235,
780
+ "eval_steps_per_second": 9.48,
781
+ "step": 4875
782
+ },
783
+ {
784
+ "epoch": 26.0,
785
+ "eval_exact_match": 15.4,
786
+ "eval_f1": 23.056190476190483,
787
+ "eval_qa_bleu": 10.828743073480153,
788
+ "eval_qa_exact_match": 0.132,
789
+ "eval_recite_bleu": 11.70611308380729,
790
+ "eval_recite_exact_match": 0.004,
791
+ "step": 4875
792
+ },
793
+ {
794
+ "epoch": 26.13,
795
+ "learning_rate": 3e-05,
796
+ "loss": 0.316,
797
+ "step": 4900
798
+ },
799
+ {
800
+ "epoch": 26.67,
801
+ "learning_rate": 3e-05,
802
+ "loss": 0.2891,
803
+ "step": 5000
804
+ },
805
+ {
806
+ "epoch": 27.0,
807
+ "eval_accuracy": 0.6408313725490196,
808
+ "eval_loss": 2.0192761421203613,
809
+ "eval_runtime": 7.4551,
810
+ "eval_samples_per_second": 67.068,
811
+ "eval_steps_per_second": 8.451,
812
+ "step": 5062
813
+ },
814
+ {
815
+ "epoch": 27.0,
816
+ "eval_exact_match": 14.0,
817
+ "eval_f1": 22.766833842561553,
818
+ "eval_qa_bleu": 3.5222468357048804,
819
+ "eval_qa_exact_match": 0.108,
820
+ "eval_recite_bleu": 10.474195118057567,
821
+ "eval_recite_exact_match": 0.002,
822
+ "step": 5062
823
+ },
824
+ {
825
+ "epoch": 27.2,
826
+ "learning_rate": 3e-05,
827
+ "loss": 0.299,
828
+ "step": 5100
829
+ },
830
+ {
831
+ "epoch": 27.73,
832
+ "learning_rate": 3e-05,
833
+ "loss": 0.281,
834
+ "step": 5200
835
+ },
836
+ {
837
+ "epoch": 28.0,
838
+ "eval_accuracy": 0.640243137254902,
839
+ "eval_loss": 2.074819803237915,
840
+ "eval_runtime": 7.0598,
841
+ "eval_samples_per_second": 70.823,
842
+ "eval_steps_per_second": 8.924,
843
+ "step": 5250
844
+ },
845
+ {
846
+ "epoch": 28.0,
847
+ "eval_exact_match": 14.8,
848
+ "eval_f1": 22.03527508090615,
849
+ "eval_qa_bleu": 4.8943988922035935,
850
+ "eval_qa_exact_match": 0.118,
851
+ "eval_recite_bleu": 11.435320168344504,
852
+ "eval_recite_exact_match": 0.008,
853
+ "step": 5250
854
+ },
855
+ {
856
+ "epoch": 28.27,
857
+ "learning_rate": 3e-05,
858
+ "loss": 0.2637,
859
+ "step": 5300
860
+ },
861
+ {
862
+ "epoch": 28.8,
863
+ "learning_rate": 3e-05,
864
+ "loss": 0.2639,
865
+ "step": 5400
866
+ },
867
+ {
868
+ "epoch": 29.0,
869
+ "eval_accuracy": 0.6389333333333334,
870
+ "eval_loss": 2.1383912563323975,
871
+ "eval_runtime": 7.7452,
872
+ "eval_samples_per_second": 64.556,
873
+ "eval_steps_per_second": 8.134,
874
+ "step": 5437
875
+ },
876
+ {
877
+ "epoch": 29.0,
878
+ "eval_exact_match": 16.2,
879
+ "eval_f1": 24.30754076973256,
880
+ "eval_qa_bleu": 4.486401752816634,
881
+ "eval_qa_exact_match": 0.128,
882
+ "eval_recite_bleu": 11.730902648782326,
883
+ "eval_recite_exact_match": 0.004,
884
+ "step": 5437
885
+ },
886
+ {
887
+ "epoch": 29.33,
888
+ "learning_rate": 3e-05,
889
+ "loss": 0.2473,
890
+ "step": 5500
891
+ },
892
+ {
893
+ "epoch": 29.87,
894
+ "learning_rate": 3e-05,
895
+ "loss": 0.249,
896
+ "step": 5600
897
+ },
898
+ {
899
+ "epoch": 30.0,
900
+ "eval_accuracy": 0.6398509803921568,
901
+ "eval_loss": 2.125810384750366,
902
+ "eval_runtime": 7.6954,
903
+ "eval_samples_per_second": 64.974,
904
+ "eval_steps_per_second": 8.187,
905
+ "step": 5625
906
+ },
907
+ {
908
+ "epoch": 30.0,
909
+ "eval_exact_match": 15.8,
910
+ "eval_f1": 22.123535353535363,
911
+ "eval_qa_bleu": 3.2440134844965627,
912
+ "eval_qa_exact_match": 0.126,
913
+ "eval_recite_bleu": 11.76325197849696,
914
+ "eval_recite_exact_match": 0.004,
915
+ "step": 5625
916
+ },
917
+ {
918
+ "epoch": 30.4,
919
+ "learning_rate": 3e-05,
920
+ "loss": 0.2263,
921
+ "step": 5700
922
+ },
923
+ {
924
+ "epoch": 30.93,
925
+ "learning_rate": 3e-05,
926
+ "loss": 0.236,
927
+ "step": 5800
928
+ },
929
+ {
930
+ "epoch": 31.0,
931
+ "eval_accuracy": 0.6378666666666667,
932
+ "eval_loss": 2.2261879444122314,
933
+ "eval_runtime": 7.6462,
934
+ "eval_samples_per_second": 65.392,
935
+ "eval_steps_per_second": 8.239,
936
+ "step": 5812
937
+ },
938
+ {
939
+ "epoch": 31.0,
940
+ "eval_exact_match": 17.6,
941
+ "eval_f1": 25.120476190476193,
942
+ "eval_qa_bleu": 5.240120013213656,
943
+ "eval_qa_exact_match": 0.13,
944
+ "eval_recite_bleu": 11.888730171325669,
945
+ "eval_recite_exact_match": 0.008,
946
+ "step": 5812
947
+ },
948
+ {
949
+ "epoch": 31.47,
950
+ "learning_rate": 3e-05,
951
+ "loss": 0.2062,
952
+ "step": 5900
953
+ },
954
+ {
955
+ "epoch": 32.0,
956
+ "learning_rate": 3e-05,
957
+ "loss": 0.2235,
958
+ "step": 6000
959
+ },
960
+ {
961
+ "epoch": 32.0,
962
+ "eval_accuracy": 0.6381254901960784,
963
+ "eval_loss": 2.2489171028137207,
964
+ "eval_runtime": 7.0486,
965
+ "eval_samples_per_second": 70.936,
966
+ "eval_steps_per_second": 8.938,
967
+ "step": 6000
968
+ },
969
+ {
970
+ "epoch": 32.0,
971
+ "eval_exact_match": 16.6,
972
+ "eval_f1": 25.16174603174605,
973
+ "eval_qa_bleu": 12.336439673952102,
974
+ "eval_qa_exact_match": 0.134,
975
+ "eval_recite_bleu": 11.520322065980684,
976
+ "eval_recite_exact_match": 0.008,
977
+ "step": 6000
978
+ },
979
+ {
980
+ "epoch": 32.53,
981
+ "learning_rate": 3e-05,
982
+ "loss": 0.1952,
983
+ "step": 6100
984
+ },
985
+ {
986
+ "epoch": 33.0,
987
+ "eval_accuracy": 0.6373098039215687,
988
+ "eval_loss": 2.2877373695373535,
989
+ "eval_runtime": 7.7488,
990
+ "eval_samples_per_second": 64.526,
991
+ "eval_steps_per_second": 8.13,
992
+ "step": 6187
993
+ },
994
+ {
995
+ "epoch": 33.0,
996
+ "eval_exact_match": 15.2,
997
+ "eval_f1": 23.447550852439747,
998
+ "eval_qa_bleu": 3.977016577169375,
999
+ "eval_qa_exact_match": 0.126,
1000
+ "eval_recite_bleu": 13.076208653752236,
1001
+ "eval_recite_exact_match": 0.016,
1002
+ "step": 6187
1003
+ },
1004
+ {
1005
+ "epoch": 33.07,
1006
+ "learning_rate": 3e-05,
1007
+ "loss": 0.2023,
1008
+ "step": 6200
1009
+ },
1010
+ {
1011
+ "epoch": 33.6,
1012
+ "learning_rate": 3e-05,
1013
+ "loss": 0.1857,
1014
+ "step": 6300
1015
+ },
1016
+ {
1017
+ "epoch": 34.0,
1018
+ "eval_accuracy": 0.6367921568627452,
1019
+ "eval_loss": 2.309786319732666,
1020
+ "eval_runtime": 6.6393,
1021
+ "eval_samples_per_second": 75.309,
1022
+ "eval_steps_per_second": 9.489,
1023
+ "step": 6375
1024
+ },
1025
+ {
1026
+ "epoch": 34.0,
1027
+ "eval_exact_match": 14.6,
1028
+ "eval_f1": 21.406193118687113,
1029
+ "eval_qa_bleu": 3.892800891482867,
1030
+ "eval_qa_exact_match": 0.116,
1031
+ "eval_recite_bleu": 11.365741641301113,
1032
+ "eval_recite_exact_match": 0.008,
1033
+ "step": 6375
1034
+ },
1035
+ {
1036
+ "epoch": 34.13,
1037
+ "learning_rate": 3e-05,
1038
+ "loss": 0.1888,
1039
+ "step": 6400
1040
+ },
1041
+ {
1042
+ "epoch": 34.67,
1043
+ "learning_rate": 3e-05,
1044
+ "loss": 0.1739,
1045
+ "step": 6500
1046
+ },
1047
+ {
1048
+ "epoch": 35.0,
1049
+ "eval_accuracy": 0.6370196078431373,
1050
+ "eval_loss": 2.356412172317505,
1051
+ "eval_runtime": 6.7022,
1052
+ "eval_samples_per_second": 74.602,
1053
+ "eval_steps_per_second": 9.4,
1054
+ "step": 6562
1055
+ },
1056
+ {
1057
+ "epoch": 35.0,
1058
+ "eval_exact_match": 16.0,
1059
+ "eval_f1": 24.90098936662045,
1060
+ "eval_qa_bleu": 6.669415710869239,
1061
+ "eval_qa_exact_match": 0.13,
1062
+ "eval_recite_bleu": 12.340892651236608,
1063
+ "eval_recite_exact_match": 0.01,
1064
+ "step": 6562
1065
+ },
1066
+ {
1067
+ "epoch": 35.2,
1068
+ "learning_rate": 3e-05,
1069
+ "loss": 0.1771,
1070
+ "step": 6600
1071
+ },
1072
+ {
1073
+ "epoch": 35.73,
1074
+ "learning_rate": 3e-05,
1075
+ "loss": 0.1675,
1076
+ "step": 6700
1077
+ },
1078
+ {
1079
+ "epoch": 36.0,
1080
+ "eval_accuracy": 0.6365647058823529,
1081
+ "eval_loss": 2.413015127182007,
1082
+ "eval_runtime": 7.498,
1083
+ "eval_samples_per_second": 66.684,
1084
+ "eval_steps_per_second": 8.402,
1085
+ "step": 6750
1086
+ },
1087
+ {
1088
+ "epoch": 36.0,
1089
+ "eval_exact_match": 16.6,
1090
+ "eval_f1": 24.060404040404055,
1091
+ "eval_qa_bleu": 9.329376121653498,
1092
+ "eval_qa_exact_match": 0.134,
1093
+ "eval_recite_bleu": 12.441487241994658,
1094
+ "eval_recite_exact_match": 0.012,
1095
+ "step": 6750
1096
+ },
1097
+ {
1098
+ "epoch": 36.27,
1099
+ "learning_rate": 3e-05,
1100
+ "loss": 0.1623,
1101
+ "step": 6800
1102
+ },
1103
+ {
1104
+ "epoch": 36.8,
1105
+ "learning_rate": 3e-05,
1106
+ "loss": 0.158,
1107
+ "step": 6900
1108
+ },
1109
+ {
1110
+ "epoch": 37.0,
1111
+ "eval_accuracy": 0.6361333333333333,
1112
+ "eval_loss": 2.4566919803619385,
1113
+ "eval_runtime": 7.1162,
1114
+ "eval_samples_per_second": 70.263,
1115
+ "eval_steps_per_second": 8.853,
1116
+ "step": 6937
1117
+ },
1118
+ {
1119
+ "epoch": 37.0,
1120
+ "eval_exact_match": 16.0,
1121
+ "eval_f1": 22.78879744333159,
1122
+ "eval_qa_bleu": 4.048618436225478,
1123
+ "eval_qa_exact_match": 0.128,
1124
+ "eval_recite_bleu": 12.710989230469986,
1125
+ "eval_recite_exact_match": 0.014,
1126
+ "step": 6937
1127
+ },
1128
+ {
1129
+ "epoch": 37.33,
1130
+ "learning_rate": 3e-05,
1131
+ "loss": 0.1525,
1132
+ "step": 7000
1133
+ },
1134
+ {
1135
+ "epoch": 37.87,
1136
+ "learning_rate": 3e-05,
1137
+ "loss": 0.1509,
1138
+ "step": 7100
1139
+ },
1140
+ {
1141
+ "epoch": 38.0,
1142
+ "eval_accuracy": 0.6366117647058823,
1143
+ "eval_loss": 2.466296672821045,
1144
+ "eval_runtime": 6.7087,
1145
+ "eval_samples_per_second": 74.53,
1146
+ "eval_steps_per_second": 9.391,
1147
+ "step": 7125
1148
+ },
1149
+ {
1150
+ "epoch": 38.0,
1151
+ "eval_exact_match": 17.8,
1152
+ "eval_f1": 24.827619047619052,
1153
+ "eval_qa_bleu": 6.898280882518603,
1154
+ "eval_qa_exact_match": 0.142,
1155
+ "eval_recite_bleu": 13.041079892055818,
1156
+ "eval_recite_exact_match": 0.012,
1157
+ "step": 7125
1158
+ },
1159
+ {
1160
+ "epoch": 38.4,
1161
+ "learning_rate": 3e-05,
1162
+ "loss": 0.1425,
1163
+ "step": 7200
1164
+ },
1165
+ {
1166
+ "epoch": 38.93,
1167
+ "learning_rate": 3e-05,
1168
+ "loss": 0.1461,
1169
+ "step": 7300
1170
+ },
1171
+ {
1172
+ "epoch": 39.0,
1173
+ "eval_accuracy": 0.6351137254901961,
1174
+ "eval_loss": 2.5354840755462646,
1175
+ "eval_runtime": 7.4853,
1176
+ "eval_samples_per_second": 66.797,
1177
+ "eval_steps_per_second": 8.416,
1178
+ "step": 7312
1179
+ },
1180
+ {
1181
+ "epoch": 39.0,
1182
+ "eval_exact_match": 14.6,
1183
+ "eval_f1": 21.557038888678758,
1184
+ "eval_qa_bleu": 4.264242388706355,
1185
+ "eval_qa_exact_match": 0.124,
1186
+ "eval_recite_bleu": 11.656295817452541,
1187
+ "eval_recite_exact_match": 0.008,
1188
+ "step": 7312
1189
+ },
1190
+ {
1191
+ "epoch": 39.47,
1192
+ "learning_rate": 3e-05,
1193
+ "loss": 0.1322,
1194
+ "step": 7400
1195
+ },
1196
+ {
1197
+ "epoch": 40.0,
1198
+ "learning_rate": 3e-05,
1199
+ "loss": 0.1393,
1200
+ "step": 7500
1201
+ },
1202
+ {
1203
+ "epoch": 40.0,
1204
+ "eval_accuracy": 0.634964705882353,
1205
+ "eval_loss": 2.581481456756592,
1206
+ "eval_runtime": 6.7057,
1207
+ "eval_samples_per_second": 74.564,
1208
+ "eval_steps_per_second": 9.395,
1209
+ "step": 7500
1210
+ },
1211
+ {
1212
+ "epoch": 40.0,
1213
+ "eval_exact_match": 14.0,
1214
+ "eval_f1": 20.95174603174604,
1215
+ "eval_qa_bleu": 6.801813594380655,
1216
+ "eval_qa_exact_match": 0.116,
1217
+ "eval_recite_bleu": 12.129670692723177,
1218
+ "eval_recite_exact_match": 0.01,
1219
+ "step": 7500
1220
+ },
1221
+ {
1222
+ "epoch": 40.53,
1223
+ "learning_rate": 3e-05,
1224
+ "loss": 0.1252,
1225
+ "step": 7600
1226
+ },
1227
+ {
1228
+ "epoch": 41.0,
1229
+ "eval_accuracy": 0.6353098039215687,
1230
+ "eval_loss": 2.5514683723449707,
1231
+ "eval_runtime": 7.4812,
1232
+ "eval_samples_per_second": 66.834,
1233
+ "eval_steps_per_second": 8.421,
1234
+ "step": 7687
1235
+ },
1236
+ {
1237
+ "epoch": 41.0,
1238
+ "eval_exact_match": 15.4,
1239
+ "eval_f1": 22.577012987013,
1240
+ "eval_qa_bleu": 10.500103979458904,
1241
+ "eval_qa_exact_match": 0.134,
1242
+ "eval_recite_bleu": 11.650747183051218,
1243
+ "eval_recite_exact_match": 0.01,
1244
+ "step": 7687
1245
+ },
1246
+ {
1247
+ "epoch": 41.07,
1248
+ "learning_rate": 3e-05,
1249
+ "loss": 0.1322,
1250
+ "step": 7700
1251
+ },
1252
+ {
1253
+ "epoch": 41.6,
1254
+ "learning_rate": 3e-05,
1255
+ "loss": 0.1203,
1256
+ "step": 7800
1257
+ },
1258
+ {
1259
+ "epoch": 42.0,
1260
+ "eval_accuracy": 0.6352705882352941,
1261
+ "eval_loss": 2.5817809104919434,
1262
+ "eval_runtime": 6.6635,
1263
+ "eval_samples_per_second": 75.035,
1264
+ "eval_steps_per_second": 9.454,
1265
+ "step": 7875
1266
+ },
1267
+ {
1268
+ "epoch": 42.0,
1269
+ "eval_exact_match": 15.0,
1270
+ "eval_f1": 22.31658263305323,
1271
+ "eval_qa_bleu": 6.526035776150396,
1272
+ "eval_qa_exact_match": 0.132,
1273
+ "eval_recite_bleu": 13.143639447688551,
1274
+ "eval_recite_exact_match": 0.012,
1275
+ "step": 7875
1276
+ },
1277
+ {
1278
+ "epoch": 42.25,
1279
+ "learning_rate": 3e-05,
1280
+ "loss": 0.1094,
1281
+ "step": 7900
1282
+ },
1283
+ {
1284
+ "epoch": 42.78,
1285
+ "learning_rate": 3e-05,
1286
+ "loss": 0.1179,
1287
+ "step": 8000
1288
+ },
1289
+ {
1290
+ "epoch": 43.0,
1291
+ "eval_accuracy": 0.6354745098039216,
1292
+ "eval_loss": 2.629621982574463,
1293
+ "eval_runtime": 7.7246,
1294
+ "eval_samples_per_second": 64.728,
1295
+ "eval_steps_per_second": 8.156,
1296
+ "step": 8041
1297
+ },
1298
+ {
1299
+ "epoch": 43.0,
1300
+ "eval_exact_match": 15.0,
1301
+ "eval_f1": 22.786969696969702,
1302
+ "eval_qa_bleu": 6.542657658031823,
1303
+ "eval_qa_exact_match": 0.118,
1304
+ "eval_recite_bleu": 12.225128582410859,
1305
+ "eval_recite_exact_match": 0.01,
1306
+ "step": 8041
1307
+ }
1308
+ ],
1309
+ "logging_steps": 100,
1310
+ "max_steps": 9350,
1311
+ "num_train_epochs": 50,
1312
+ "save_steps": 500,
1313
+ "total_flos": 1.8637055884924027e+18,
1314
+ "trial_name": null,
1315
+ "trial_params": null
1316
+ }
checkpoint-8041/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3407c08abce41cb0416802afb0954e7df7ab1fc16a56c52b9d611206d5650365
3
+ size 4728