Hugging Face's logo

YagiASAFAS
/
PoliBERT-MY

Safetensors
bert
politics
Model card Files
xet
Community
YagiASAFAS commited on
Commit
dc2d381
·
verified ·
1 Parent(s): d17ea72

Training in progress, epoch 1

Browse files
Files changed (34) hide show
  1. model.safetensors +1 -1
  2. run-0/checkpoint-4800/config.json +125 -0
  3. run-0/checkpoint-4800/model.safetensors +3 -0
  4. run-0/checkpoint-4800/optimizer.pt +3 -0
  5. run-0/checkpoint-4800/rng_state.pth +3 -0
  6. run-0/checkpoint-4800/scaler.pt +3 -0
  7. run-0/checkpoint-4800/scheduler.pt +3 -0
  8. run-0/checkpoint-4800/trainer_state.json +655 -0
  9. run-0/checkpoint-4800/training_args.bin +3 -0
  10. run-0/checkpoint-5400/config.json +125 -0
  11. run-0/checkpoint-5400/model.safetensors +3 -0
  12. run-0/checkpoint-5400/optimizer.pt +3 -0
  13. run-0/checkpoint-5400/rng_state.pth +3 -0
  14. run-0/checkpoint-5400/scaler.pt +3 -0
  15. run-0/checkpoint-5400/scheduler.pt +3 -0
  16. run-0/checkpoint-5400/trainer_state.json +731 -0
  17. run-0/checkpoint-5400/training_args.bin +3 -0
  18. run-0/checkpoint-5990/config.json +125 -0
  19. run-0/checkpoint-5990/model.safetensors +3 -0
  20. run-0/checkpoint-5990/optimizer.pt +3 -0
  21. run-0/checkpoint-5990/rng_state.pth +3 -0
  22. run-0/checkpoint-5990/scaler.pt +3 -0
  23. run-0/checkpoint-5990/scheduler.pt +3 -0
  24. run-0/checkpoint-5990/trainer_state.json +800 -0
  25. run-0/checkpoint-5990/training_args.bin +3 -0
  26. run-1/checkpoint-2399/config.json +125 -0
  27. run-1/checkpoint-2399/model.safetensors +3 -0
  28. run-1/checkpoint-2399/optimizer.pt +3 -0
  29. run-1/checkpoint-2399/rng_state.pth +3 -0
  30. run-1/checkpoint-2399/scaler.pt +3 -0
  31. run-1/checkpoint-2399/scheduler.pt +3 -0
  32. run-1/checkpoint-2399/trainer_state.json +242 -0
  33. run-1/checkpoint-2399/training_args.bin +3 -0
  34. training_args.bin +1 -1
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:61b4f741e6da4a5f3ceb0090ea4bb6ff8bd986f5d0d25dba0202dc9e575ddc3b
3
  size 438100144
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc8d6559adcf9e224d38864a1b5e5a1403e35e26169912ed2c48d05a8c54ac1d
3
  size 438100144
run-0/checkpoint-4800/config.json ADDED
@@ -0,0 +1,125 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "BertForSequenceClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.1,
6
+ "classifier_dropout": null,
7
+ "gradient_checkpointing": false,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.1,
10
+ "hidden_size": 768,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47"
60
+ },
61
+ "initializer_range": 0.02,
62
+ "intermediate_size": 3072,
63
+ "label2id": {
64
+ "LABEL_0": 0,
65
+ "LABEL_1": 1,
66
+ "LABEL_10": 10,
67
+ "LABEL_11": 11,
68
+ "LABEL_12": 12,
69
+ "LABEL_13": 13,
70
+ "LABEL_14": 14,
71
+ "LABEL_15": 15,
72
+ "LABEL_16": 16,
73
+ "LABEL_17": 17,
74
+ "LABEL_18": 18,
75
+ "LABEL_19": 19,
76
+ "LABEL_2": 2,
77
+ "LABEL_20": 20,
78
+ "LABEL_21": 21,
79
+ "LABEL_22": 22,
80
+ "LABEL_23": 23,
81
+ "LABEL_24": 24,
82
+ "LABEL_25": 25,
83
+ "LABEL_26": 26,
84
+ "LABEL_27": 27,
85
+ "LABEL_28": 28,
86
+ "LABEL_29": 29,
87
+ "LABEL_3": 3,
88
+ "LABEL_30": 30,
89
+ "LABEL_31": 31,
90
+ "LABEL_32": 32,
91
+ "LABEL_33": 33,
92
+ "LABEL_34": 34,
93
+ "LABEL_35": 35,
94
+ "LABEL_36": 36,
95
+ "LABEL_37": 37,
96
+ "LABEL_38": 38,
97
+ "LABEL_39": 39,
98
+ "LABEL_4": 4,
99
+ "LABEL_40": 40,
100
+ "LABEL_41": 41,
101
+ "LABEL_42": 42,
102
+ "LABEL_43": 43,
103
+ "LABEL_44": 44,
104
+ "LABEL_45": 45,
105
+ "LABEL_46": 46,
106
+ "LABEL_47": 47,
107
+ "LABEL_5": 5,
108
+ "LABEL_6": 6,
109
+ "LABEL_7": 7,
110
+ "LABEL_8": 8,
111
+ "LABEL_9": 9
112
+ },
113
+ "layer_norm_eps": 1e-12,
114
+ "max_position_embeddings": 512,
115
+ "model_type": "bert",
116
+ "num_attention_heads": 12,
117
+ "num_hidden_layers": 12,
118
+ "pad_token_id": 0,
119
+ "position_embedding_type": "absolute",
120
+ "torch_dtype": "float32",
121
+ "transformers_version": "4.50.3",
122
+ "type_vocab_size": 2,
123
+ "use_cache": true,
124
+ "vocab_size": 30522
125
+ }
run-0/checkpoint-4800/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c2c29fa58adb10ad5952e1d5f33dbc136aac9269e6aecabd75003581aae990c
3
+ size 438100144
run-0/checkpoint-4800/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6b0b8a32b1caac53f23e6df55ab92992939086d75953a541ef008915de14f717
3
+ size 876321402
run-0/checkpoint-4800/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ffad3d9471fb61b3f42fea2e7138fc106169adb362290f6561f26126680d7263
3
+ size 14244
run-0/checkpoint-4800/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5da4ea0c1bcacc6c536f51e41f20fb1c9301dc84cb8e04333e56f06168b8cb83
3
+ size 988
run-0/checkpoint-4800/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab022a3437d5fd65ef509a1f1cc5556ec75d65ebf4da8665b93fcbcfe045108d
3
+ size 1064
run-0/checkpoint-4800/trainer_state.json ADDED
@@ -0,0 +1,655 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 4200,
3
+ "best_metric": 0.9393817348962156,
4
+ "best_model_checkpoint": "./results/run-0/checkpoint-4200",
5
+ "epoch": 8.0,
6
+ "eval_steps": 500,
7
+ "global_step": 4800,
8
+ "is_hyper_param_search": true,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.16673614005835766,
14
+ "grad_norm": 2.8743319511413574,
15
+ "learning_rate": 6e-06,
16
+ "loss": 1.4727,
17
+ "step": 100
18
+ },
19
+ {
20
+ "epoch": 0.3334722801167153,
21
+ "grad_norm": 0.8747783899307251,
22
+ "learning_rate": 1.2e-05,
23
+ "loss": 0.617,
24
+ "step": 200
25
+ },
26
+ {
27
+ "epoch": 0.5002084201750729,
28
+ "grad_norm": 0.7884620428085327,
29
+ "learning_rate": 1.8e-05,
30
+ "loss": 0.4493,
31
+ "step": 300
32
+ },
33
+ {
34
+ "epoch": 0.6669445602334306,
35
+ "grad_norm": 0.8924545645713806,
36
+ "learning_rate": 2.4e-05,
37
+ "loss": 0.3876,
38
+ "step": 400
39
+ },
40
+ {
41
+ "epoch": 0.8336807002917882,
42
+ "grad_norm": 0.8031777739524841,
43
+ "learning_rate": 3e-05,
44
+ "loss": 0.3361,
45
+ "step": 500
46
+ },
47
+ {
48
+ "epoch": 1.0,
49
+ "grad_norm": 0.5999459624290466,
50
+ "learning_rate": 2.9453551912568308e-05,
51
+ "loss": 0.3004,
52
+ "step": 600
53
+ },
54
+ {
55
+ "epoch": 1.0,
56
+ "eval_administration_accuracy": 0.9140087554721701,
57
+ "eval_administration_f1": 0.8740363354524873,
58
+ "eval_corruption_accuracy": 0.9370439858244737,
59
+ "eval_corruption_f1": 0.9192595640701061,
60
+ "eval_democracy_accuracy": 0.9347508859703981,
61
+ "eval_democracy_f1": 0.9095054688749907,
62
+ "eval_development_accuracy": 0.8763810715030227,
63
+ "eval_development_f1": 0.8253993969919758,
64
+ "eval_economy_accuracy": 0.9010840108401084,
65
+ "eval_economy_f1": 0.8737219947600353,
66
+ "eval_education_accuracy": 0.9571607254534084,
67
+ "eval_education_f1": 0.9485581438064299,
68
+ "eval_environment_accuracy": 0.9734208880550344,
69
+ "eval_environment_f1": 0.9693817386774968,
70
+ "eval_instability_accuracy": 0.9219303731498854,
71
+ "eval_instability_f1": 0.8878400653107733,
72
+ "eval_leadership_accuracy": 0.7652699603919116,
73
+ "eval_leadership_f1": 0.7080044131422123,
74
+ "eval_loss": 0.2839118242263794,
75
+ "eval_overall_accuracy": 0.9120109790841496,
76
+ "eval_overall_f1": 0.8849771384459072,
77
+ "eval_race_accuracy": 0.9283927454659162,
78
+ "eval_race_f1": 0.907176591968042,
79
+ "eval_religion_accuracy": 0.9369397540129247,
80
+ "eval_religion_f1": 0.9217332965382043,
81
+ "eval_runtime": 10.9229,
82
+ "eval_safety_accuracy": 0.8977485928705441,
83
+ "eval_safety_f1": 0.8751086517581315,
84
+ "eval_samples_per_second": 878.336,
85
+ "eval_steps_per_second": 54.93,
86
+ "step": 600
87
+ },
88
+ {
89
+ "epoch": 1.1667361400583576,
90
+ "grad_norm": 0.6991692781448364,
91
+ "learning_rate": 2.890710382513661e-05,
92
+ "loss": 0.2667,
93
+ "step": 700
94
+ },
95
+ {
96
+ "epoch": 1.3334722801167154,
97
+ "grad_norm": 0.6808522343635559,
98
+ "learning_rate": 2.836065573770492e-05,
99
+ "loss": 0.2573,
100
+ "step": 800
101
+ },
102
+ {
103
+ "epoch": 1.500208420175073,
104
+ "grad_norm": 0.8924445509910583,
105
+ "learning_rate": 2.7814207650273226e-05,
106
+ "loss": 0.2447,
107
+ "step": 900
108
+ },
109
+ {
110
+ "epoch": 1.6669445602334307,
111
+ "grad_norm": 0.9563364386558533,
112
+ "learning_rate": 2.7267759562841533e-05,
113
+ "loss": 0.2282,
114
+ "step": 1000
115
+ },
116
+ {
117
+ "epoch": 1.8336807002917883,
118
+ "grad_norm": 0.8881794810295105,
119
+ "learning_rate": 2.6721311475409837e-05,
120
+ "loss": 0.2217,
121
+ "step": 1100
122
+ },
123
+ {
124
+ "epoch": 2.0,
125
+ "grad_norm": 1.2946265935897827,
126
+ "learning_rate": 2.6174863387978144e-05,
127
+ "loss": 0.2197,
128
+ "step": 1200
129
+ },
130
+ {
131
+ "epoch": 2.0,
132
+ "eval_administration_accuracy": 0.9364185949551803,
133
+ "eval_administration_f1": 0.9264664888238818,
134
+ "eval_corruption_accuracy": 0.9562226391494684,
135
+ "eval_corruption_f1": 0.9508457468931831,
136
+ "eval_democracy_accuracy": 0.947571398790911,
137
+ "eval_democracy_f1": 0.9355668234958175,
138
+ "eval_development_accuracy": 0.9049405878674172,
139
+ "eval_development_f1": 0.8818640965909518,
140
+ "eval_economy_accuracy": 0.924119241192412,
141
+ "eval_economy_f1": 0.9144295841480942,
142
+ "eval_education_accuracy": 0.9669585157390036,
143
+ "eval_education_f1": 0.9649722301773493,
144
+ "eval_environment_accuracy": 0.981029810298103,
145
+ "eval_environment_f1": 0.9786767163906938,
146
+ "eval_instability_accuracy": 0.9422555764019178,
147
+ "eval_instability_f1": 0.9320449284299851,
148
+ "eval_leadership_accuracy": 0.8006045445069835,
149
+ "eval_leadership_f1": 0.7682063698028097,
150
+ "eval_loss": 0.21895167231559753,
151
+ "eval_overall_accuracy": 0.9317629073726635,
152
+ "eval_overall_f1": 0.9213111296186804,
153
+ "eval_race_accuracy": 0.9516364394413175,
154
+ "eval_race_f1": 0.9462871699636376,
155
+ "eval_religion_accuracy": 0.9586199708150928,
156
+ "eval_religion_f1": 0.9568464833874664,
157
+ "eval_runtime": 10.9295,
158
+ "eval_safety_accuracy": 0.9107775693141547,
159
+ "eval_safety_f1": 0.8995269173202963,
160
+ "eval_samples_per_second": 877.809,
161
+ "eval_steps_per_second": 54.897,
162
+ "step": 1200
163
+ },
164
+ {
165
+ "epoch": 2.1667361400583576,
166
+ "grad_norm": 0.8773576021194458,
167
+ "learning_rate": 2.5628415300546447e-05,
168
+ "loss": 0.1841,
169
+ "step": 1300
170
+ },
171
+ {
172
+ "epoch": 2.333472280116715,
173
+ "grad_norm": 0.7541666030883789,
174
+ "learning_rate": 2.5081967213114754e-05,
175
+ "loss": 0.1757,
176
+ "step": 1400
177
+ },
178
+ {
179
+ "epoch": 2.500208420175073,
180
+ "grad_norm": 0.9500707983970642,
181
+ "learning_rate": 2.453551912568306e-05,
182
+ "loss": 0.1776,
183
+ "step": 1500
184
+ },
185
+ {
186
+ "epoch": 2.6669445602334307,
187
+ "grad_norm": 0.8599845170974731,
188
+ "learning_rate": 2.398907103825137e-05,
189
+ "loss": 0.1757,
190
+ "step": 1600
191
+ },
192
+ {
193
+ "epoch": 2.8336807002917883,
194
+ "grad_norm": 1.0130952596664429,
195
+ "learning_rate": 2.3442622950819672e-05,
196
+ "loss": 0.1675,
197
+ "step": 1700
198
+ },
199
+ {
200
+ "epoch": 3.0,
201
+ "grad_norm": 1.4208967685699463,
202
+ "learning_rate": 2.289617486338798e-05,
203
+ "loss": 0.1664,
204
+ "step": 1800
205
+ },
206
+ {
207
+ "epoch": 3.0,
208
+ "eval_administration_accuracy": 0.940379403794038,
209
+ "eval_administration_f1": 0.9332522249114288,
210
+ "eval_corruption_accuracy": 0.9609130706691682,
211
+ "eval_corruption_f1": 0.9591571276198037,
212
+ "eval_democracy_accuracy": 0.9572649572649573,
213
+ "eval_democracy_f1": 0.951957009924363,
214
+ "eval_development_accuracy": 0.9184907233687721,
215
+ "eval_development_f1": 0.909063933465515,
216
+ "eval_economy_accuracy": 0.9327704815509693,
217
+ "eval_economy_f1": 0.9291641275521808,
218
+ "eval_education_accuracy": 0.967479674796748,
219
+ "eval_education_f1": 0.965512166473417,
220
+ "eval_environment_accuracy": 0.9835313737752762,
221
+ "eval_environment_f1": 0.9820983135141137,
222
+ "eval_instability_accuracy": 0.9481967896602043,
223
+ "eval_instability_f1": 0.9444250756726008,
224
+ "eval_leadership_accuracy": 0.8285386700020846,
225
+ "eval_leadership_f1": 0.8147910628849035,
226
+ "eval_loss": 0.19577622413635254,
227
+ "eval_overall_accuracy": 0.9393283996942533,
228
+ "eval_overall_f1": 0.9346706078324559,
229
+ "eval_race_accuracy": 0.9551803210339795,
230
+ "eval_race_f1": 0.9521792365558912,
231
+ "eval_religion_accuracy": 0.9598707525536794,
232
+ "eval_religion_f1": 0.9582958411796174,
233
+ "eval_runtime": 10.979,
234
+ "eval_safety_accuracy": 0.9193245778611632,
235
+ "eval_safety_f1": 0.9161511742356354,
236
+ "eval_samples_per_second": 873.853,
237
+ "eval_steps_per_second": 54.65,
238
+ "step": 1800
239
+ },
240
+ {
241
+ "epoch": 3.1667361400583576,
242
+ "grad_norm": 0.7676725387573242,
243
+ "learning_rate": 2.2349726775956283e-05,
244
+ "loss": 0.137,
245
+ "step": 1900
246
+ },
247
+ {
248
+ "epoch": 3.333472280116715,
249
+ "grad_norm": 0.9001705050468445,
250
+ "learning_rate": 2.180327868852459e-05,
251
+ "loss": 0.1366,
252
+ "step": 2000
253
+ },
254
+ {
255
+ "epoch": 3.500208420175073,
256
+ "grad_norm": 0.9584451913833618,
257
+ "learning_rate": 2.1256830601092897e-05,
258
+ "loss": 0.1319,
259
+ "step": 2100
260
+ },
261
+ {
262
+ "epoch": 3.6669445602334307,
263
+ "grad_norm": 0.9352701306343079,
264
+ "learning_rate": 2.0710382513661204e-05,
265
+ "loss": 0.1324,
266
+ "step": 2200
267
+ },
268
+ {
269
+ "epoch": 3.8336807002917883,
270
+ "grad_norm": 0.8512120842933655,
271
+ "learning_rate": 2.0163934426229508e-05,
272
+ "loss": 0.1333,
273
+ "step": 2300
274
+ },
275
+ {
276
+ "epoch": 4.0,
277
+ "grad_norm": 0.8892608880996704,
278
+ "learning_rate": 1.9617486338797815e-05,
279
+ "loss": 0.1266,
280
+ "step": 2400
281
+ },
282
+ {
283
+ "epoch": 4.0,
284
+ "eval_administration_accuracy": 0.9418386491557224,
285
+ "eval_administration_f1": 0.9352052966447959,
286
+ "eval_corruption_accuracy": 0.9622680842193038,
287
+ "eval_corruption_f1": 0.9610755105054823,
288
+ "eval_democracy_accuracy": 0.9578903481342506,
289
+ "eval_democracy_f1": 0.9547168679700065,
290
+ "eval_development_accuracy": 0.9218261413383365,
291
+ "eval_development_f1": 0.9169350889919825,
292
+ "eval_economy_accuracy": 0.9337085678549093,
293
+ "eval_economy_f1": 0.9328708122199544,
294
+ "eval_education_accuracy": 0.9685219929122368,
295
+ "eval_education_f1": 0.9667154036014268,
296
+ "eval_environment_accuracy": 0.9815509693558474,
297
+ "eval_environment_f1": 0.9809518427248292,
298
+ "eval_instability_accuracy": 0.947571398790911,
299
+ "eval_instability_f1": 0.9441507785991868,
300
+ "eval_leadership_accuracy": 0.8358348968105066,
301
+ "eval_leadership_f1": 0.82702607509575,
302
+ "eval_loss": 0.1933656632900238,
303
+ "eval_overall_accuracy": 0.9413782919880481,
304
+ "eval_overall_f1": 0.9381715218630767,
305
+ "eval_race_accuracy": 0.9591411298728372,
306
+ "eval_race_f1": 0.9571038116721658,
307
+ "eval_religion_accuracy": 0.9621638524077548,
308
+ "eval_religion_f1": 0.9611025172696737,
309
+ "eval_runtime": 10.9996,
310
+ "eval_safety_accuracy": 0.9242234730039608,
311
+ "eval_safety_f1": 0.920204257061666,
312
+ "eval_samples_per_second": 872.211,
313
+ "eval_steps_per_second": 54.547,
314
+ "step": 2400
315
+ },
316
+ {
317
+ "epoch": 4.166736140058358,
318
+ "grad_norm": 0.6617388129234314,
319
+ "learning_rate": 1.907103825136612e-05,
320
+ "loss": 0.1039,
321
+ "step": 2500
322
+ },
323
+ {
324
+ "epoch": 4.333472280116715,
325
+ "grad_norm": 0.7575182914733887,
326
+ "learning_rate": 1.8524590163934426e-05,
327
+ "loss": 0.1105,
328
+ "step": 2600
329
+ },
330
+ {
331
+ "epoch": 4.500208420175073,
332
+ "grad_norm": 0.69528728723526,
333
+ "learning_rate": 1.7978142076502733e-05,
334
+ "loss": 0.1084,
335
+ "step": 2700
336
+ },
337
+ {
338
+ "epoch": 4.66694456023343,
339
+ "grad_norm": 0.9533807039260864,
340
+ "learning_rate": 1.743169398907104e-05,
341
+ "loss": 0.1051,
342
+ "step": 2800
343
+ },
344
+ {
345
+ "epoch": 4.833680700291788,
346
+ "grad_norm": 0.7591508030891418,
347
+ "learning_rate": 1.6885245901639344e-05,
348
+ "loss": 0.1008,
349
+ "step": 2900
350
+ },
351
+ {
352
+ "epoch": 5.0,
353
+ "grad_norm": 1.3957374095916748,
354
+ "learning_rate": 1.633879781420765e-05,
355
+ "loss": 0.1013,
356
+ "step": 3000
357
+ },
358
+ {
359
+ "epoch": 5.0,
360
+ "eval_administration_accuracy": 0.9406920992286846,
361
+ "eval_administration_f1": 0.935638525668697,
362
+ "eval_corruption_accuracy": 0.9626850114654992,
363
+ "eval_corruption_f1": 0.9611422434993832,
364
+ "eval_democracy_accuracy": 0.9496560350218887,
365
+ "eval_democracy_f1": 0.9499559979035322,
366
+ "eval_development_accuracy": 0.9201584323535543,
367
+ "eval_development_f1": 0.9155534426644083,
368
+ "eval_economy_accuracy": 0.9354805086512403,
369
+ "eval_economy_f1": 0.9325398706072688,
370
+ "eval_education_accuracy": 0.9697727746508235,
371
+ "eval_education_f1": 0.9681349456850368,
372
+ "eval_environment_accuracy": 0.9820721284135918,
373
+ "eval_environment_f1": 0.9811885272501003,
374
+ "eval_instability_accuracy": 0.9472587033562644,
375
+ "eval_instability_f1": 0.9442377846524443,
376
+ "eval_leadership_accuracy": 0.8364602876797999,
377
+ "eval_leadership_f1": 0.828026933450708,
378
+ "eval_loss": 0.19940470159053802,
379
+ "eval_overall_accuracy": 0.9405183795427697,
380
+ "eval_overall_f1": 0.9377003414537354,
381
+ "eval_race_accuracy": 0.9554930164686263,
382
+ "eval_race_f1": 0.9547403715439807,
383
+ "eval_religion_accuracy": 0.9633104023347926,
384
+ "eval_religion_f1": 0.9620625852940151,
385
+ "eval_runtime": 10.9306,
386
+ "eval_safety_accuracy": 0.923181154888472,
387
+ "eval_safety_f1": 0.9191828692252488,
388
+ "eval_samples_per_second": 877.722,
389
+ "eval_steps_per_second": 54.892,
390
+ "step": 3000
391
+ },
392
+ {
393
+ "epoch": 5.166736140058358,
394
+ "grad_norm": 0.5871491432189941,
395
+ "learning_rate": 1.5792349726775955e-05,
396
+ "loss": 0.0873,
397
+ "step": 3100
398
+ },
399
+ {
400
+ "epoch": 5.333472280116715,
401
+ "grad_norm": 0.8581504225730896,
402
+ "learning_rate": 1.5245901639344264e-05,
403
+ "loss": 0.0862,
404
+ "step": 3200
405
+ },
406
+ {
407
+ "epoch": 5.500208420175073,
408
+ "grad_norm": 0.9329476952552795,
409
+ "learning_rate": 1.4699453551912569e-05,
410
+ "loss": 0.0849,
411
+ "step": 3300
412
+ },
413
+ {
414
+ "epoch": 5.66694456023343,
415
+ "grad_norm": 0.8767560124397278,
416
+ "learning_rate": 1.4153005464480874e-05,
417
+ "loss": 0.088,
418
+ "step": 3400
419
+ },
420
+ {
421
+ "epoch": 5.833680700291788,
422
+ "grad_norm": 0.8050324320793152,
423
+ "learning_rate": 1.3606557377049181e-05,
424
+ "loss": 0.0826,
425
+ "step": 3500
426
+ },
427
+ {
428
+ "epoch": 6.0,
429
+ "grad_norm": 0.9876068830490112,
430
+ "learning_rate": 1.3060109289617487e-05,
431
+ "loss": 0.0811,
432
+ "step": 3600
433
+ },
434
+ {
435
+ "epoch": 6.0,
436
+ "eval_administration_accuracy": 0.9434021263289556,
437
+ "eval_administration_f1": 0.9387369505688391,
438
+ "eval_corruption_accuracy": 0.962997706900146,
439
+ "eval_corruption_f1": 0.9612502739505763,
440
+ "eval_democracy_accuracy": 0.9562226391494684,
441
+ "eval_democracy_f1": 0.954550595976141,
442
+ "eval_development_accuracy": 0.9203668959766521,
443
+ "eval_development_f1": 0.9144818227568411,
444
+ "eval_economy_accuracy": 0.9361058995205337,
445
+ "eval_economy_f1": 0.9313226219336369,
446
+ "eval_education_accuracy": 0.9694600792161768,
447
+ "eval_education_f1": 0.9675599527596068,
448
+ "eval_environment_accuracy": 0.9811340421096518,
449
+ "eval_environment_f1": 0.9803282887818994,
450
+ "eval_instability_accuracy": 0.9459036898061288,
451
+ "eval_instability_f1": 0.942472499866658,
452
+ "eval_leadership_accuracy": 0.8430268918073797,
453
+ "eval_leadership_f1": 0.8387770890259914,
454
+ "eval_loss": 0.20078489184379578,
455
+ "eval_overall_accuracy": 0.941647557501216,
456
+ "eval_overall_f1": 0.938780687918615,
457
+ "eval_race_accuracy": 0.9580988117573483,
458
+ "eval_race_f1": 0.9562411064132649,
459
+ "eval_religion_accuracy": 0.961955388784657,
460
+ "eval_religion_f1": 0.960819752774671,
461
+ "eval_runtime": 11.0615,
462
+ "eval_safety_accuracy": 0.9210965186574943,
463
+ "eval_safety_f1": 0.9188273002152547,
464
+ "eval_samples_per_second": 867.33,
465
+ "eval_steps_per_second": 54.242,
466
+ "step": 3600
467
+ },
468
+ {
469
+ "epoch": 6.166736140058358,
470
+ "grad_norm": 0.8045951724052429,
471
+ "learning_rate": 1.2513661202185792e-05,
472
+ "loss": 0.0701,
473
+ "step": 3700
474
+ },
475
+ {
476
+ "epoch": 6.333472280116715,
477
+ "grad_norm": 0.6562448143959045,
478
+ "learning_rate": 1.19672131147541e-05,
479
+ "loss": 0.0726,
480
+ "step": 3800
481
+ },
482
+ {
483
+ "epoch": 6.500208420175073,
484
+ "grad_norm": 0.7833376526832581,
485
+ "learning_rate": 1.1420765027322405e-05,
486
+ "loss": 0.0673,
487
+ "step": 3900
488
+ },
489
+ {
490
+ "epoch": 6.66694456023343,
491
+ "grad_norm": 0.8071606755256653,
492
+ "learning_rate": 1.087431693989071e-05,
493
+ "loss": 0.0733,
494
+ "step": 4000
495
+ },
496
+ {
497
+ "epoch": 6.833680700291788,
498
+ "grad_norm": 0.7821515798568726,
499
+ "learning_rate": 1.0327868852459017e-05,
500
+ "loss": 0.0678,
501
+ "step": 4100
502
+ },
503
+ {
504
+ "epoch": 7.0,
505
+ "grad_norm": 0.6441165804862976,
506
+ "learning_rate": 9.781420765027323e-06,
507
+ "loss": 0.0701,
508
+ "step": 4200
509
+ },
510
+ {
511
+ "epoch": 7.0,
512
+ "eval_administration_accuracy": 0.9396497811131957,
513
+ "eval_administration_f1": 0.9371884848945066,
514
+ "eval_corruption_accuracy": 0.9616426933500104,
515
+ "eval_corruption_f1": 0.9603149034436674,
516
+ "eval_democracy_accuracy": 0.9548676255993329,
517
+ "eval_democracy_f1": 0.9544642189136688,
518
+ "eval_development_accuracy": 0.919533041484261,
519
+ "eval_development_f1": 0.9174833579112073,
520
+ "eval_economy_accuracy": 0.9370439858244737,
521
+ "eval_economy_f1": 0.9350264441971636,
522
+ "eval_education_accuracy": 0.9670627475505524,
523
+ "eval_education_f1": 0.9657003505198045,
524
+ "eval_environment_accuracy": 0.9821763602251408,
525
+ "eval_environment_f1": 0.9808859370812918,
526
+ "eval_instability_accuracy": 0.9445486762559934,
527
+ "eval_instability_f1": 0.9432538618242539,
528
+ "eval_leadership_accuracy": 0.8441734417344173,
529
+ "eval_leadership_f1": 0.8426503273307849,
530
+ "eval_loss": 0.21108125150203705,
531
+ "eval_overall_accuracy": 0.9408397609617123,
532
+ "eval_overall_f1": 0.9393817348962156,
533
+ "eval_race_accuracy": 0.9566395663956639,
534
+ "eval_race_f1": 0.9559208826126899,
535
+ "eval_religion_accuracy": 0.9609130706691682,
536
+ "eval_religion_f1": 0.9608002073614549,
537
+ "eval_runtime": 10.9796,
538
+ "eval_safety_accuracy": 0.9218261413383365,
539
+ "eval_safety_f1": 0.9188918426640929,
540
+ "eval_samples_per_second": 873.8,
541
+ "eval_steps_per_second": 54.647,
542
+ "step": 4200
543
+ },
544
+ {
545
+ "epoch": 7.166736140058358,
546
+ "grad_norm": 0.6236246228218079,
547
+ "learning_rate": 9.234972677595628e-06,
548
+ "loss": 0.0597,
549
+ "step": 4300
550
+ },
551
+ {
552
+ "epoch": 7.333472280116715,
553
+ "grad_norm": 0.6468791365623474,
554
+ "learning_rate": 8.688524590163935e-06,
555
+ "loss": 0.0568,
556
+ "step": 4400
557
+ },
558
+ {
559
+ "epoch": 7.500208420175073,
560
+ "grad_norm": 0.7850766777992249,
561
+ "learning_rate": 8.14207650273224e-06,
562
+ "loss": 0.0586,
563
+ "step": 4500
564
+ },
565
+ {
566
+ "epoch": 7.66694456023343,
567
+ "grad_norm": 0.9113643169403076,
568
+ "learning_rate": 7.595628415300546e-06,
569
+ "loss": 0.0611,
570
+ "step": 4600
571
+ },
572
+ {
573
+ "epoch": 7.833680700291788,
574
+ "grad_norm": 0.9405547976493835,
575
+ "learning_rate": 7.049180327868853e-06,
576
+ "loss": 0.0636,
577
+ "step": 4700
578
+ },
579
+ {
580
+ "epoch": 8.0,
581
+ "grad_norm": 0.5414742827415466,
582
+ "learning_rate": 6.502732240437159e-06,
583
+ "loss": 0.0588,
584
+ "step": 4800
585
+ },
586
+ {
587
+ "epoch": 8.0,
588
+ "eval_administration_accuracy": 0.9413174900979779,
589
+ "eval_administration_f1": 0.9368824702703686,
590
+ "eval_corruption_accuracy": 0.9614342297269126,
591
+ "eval_corruption_f1": 0.9602768664022149,
592
+ "eval_democracy_accuracy": 0.9560141755263707,
593
+ "eval_democracy_f1": 0.9543585350199038,
594
+ "eval_development_accuracy": 0.9192203460496143,
595
+ "eval_development_f1": 0.9152605716450639,
596
+ "eval_economy_accuracy": 0.9381905357515113,
597
+ "eval_economy_f1": 0.935106524044037,
598
+ "eval_education_accuracy": 0.9688346883468835,
599
+ "eval_education_f1": 0.9673054511905975,
600
+ "eval_environment_accuracy": 0.9822805920366896,
601
+ "eval_environment_f1": 0.980929327817527,
602
+ "eval_instability_accuracy": 0.9455909943714822,
603
+ "eval_instability_f1": 0.9437005892876725,
604
+ "eval_leadership_accuracy": 0.8438607462997707,
605
+ "eval_leadership_f1": 0.8410810165791273,
606
+ "eval_loss": 0.215087428689003,
607
+ "eval_overall_accuracy": 0.9414651518310054,
608
+ "eval_overall_f1": 0.93924957060805,
609
+ "eval_race_accuracy": 0.9572649572649573,
610
+ "eval_race_f1": 0.9565627258185468,
611
+ "eval_religion_accuracy": 0.9615384615384616,
612
+ "eval_religion_f1": 0.9610239728595763,
613
+ "eval_runtime": 10.9675,
614
+ "eval_safety_accuracy": 0.9220346049614342,
615
+ "eval_safety_f1": 0.9185067963619639,
616
+ "eval_samples_per_second": 874.763,
617
+ "eval_steps_per_second": 54.707,
618
+ "step": 4800
619
+ }
620
+ ],
621
+ "logging_steps": 100,
622
+ "max_steps": 5990,
623
+ "num_input_tokens_seen": 0,
624
+ "num_train_epochs": 10,
625
+ "save_steps": 500,
626
+ "stateful_callbacks": {
627
+ "EarlyStoppingCallback": {
628
+ "args": {
629
+ "early_stopping_patience": 2,
630
+ "early_stopping_threshold": 0.0
631
+ },
632
+ "attributes": {
633
+ "early_stopping_patience_counter": 1
634
+ }
635
+ },
636
+ "TrainerControl": {
637
+ "args": {
638
+ "should_epoch_stop": false,
639
+ "should_evaluate": false,
640
+ "should_log": false,
641
+ "should_save": true,
642
+ "should_training_stop": false
643
+ },
644
+ "attributes": {}
645
+ }
646
+ },
647
+ "total_flos": 8.080213808303309e+16,
648
+ "train_batch_size": 16,
649
+ "trial_name": null,
650
+ "trial_params": {
651
+ "gradient_accumulation_steps": 4,
652
+ "learning_rate": 3e-05,
653
+ "num_train_epochs": 10
654
+ }
655
+ }
run-0/checkpoint-4800/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:82c268ed019d067f52ee3b6ec06e4f32fa79f82839c235ad890040984c5ac1cf
3
+ size 5368
run-0/checkpoint-5400/config.json ADDED
@@ -0,0 +1,125 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "BertForSequenceClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.1,
6
+ "classifier_dropout": null,
7
+ "gradient_checkpointing": false,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.1,
10
+ "hidden_size": 768,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47"
60
+ },
61
+ "initializer_range": 0.02,
62
+ "intermediate_size": 3072,
63
+ "label2id": {
64
+ "LABEL_0": 0,
65
+ "LABEL_1": 1,
66
+ "LABEL_10": 10,
67
+ "LABEL_11": 11,
68
+ "LABEL_12": 12,
69
+ "LABEL_13": 13,
70
+ "LABEL_14": 14,
71
+ "LABEL_15": 15,
72
+ "LABEL_16": 16,
73
+ "LABEL_17": 17,
74
+ "LABEL_18": 18,
75
+ "LABEL_19": 19,
76
+ "LABEL_2": 2,
77
+ "LABEL_20": 20,
78
+ "LABEL_21": 21,
79
+ "LABEL_22": 22,
80
+ "LABEL_23": 23,
81
+ "LABEL_24": 24,
82
+ "LABEL_25": 25,
83
+ "LABEL_26": 26,
84
+ "LABEL_27": 27,
85
+ "LABEL_28": 28,
86
+ "LABEL_29": 29,
87
+ "LABEL_3": 3,
88
+ "LABEL_30": 30,
89
+ "LABEL_31": 31,
90
+ "LABEL_32": 32,
91
+ "LABEL_33": 33,
92
+ "LABEL_34": 34,
93
+ "LABEL_35": 35,
94
+ "LABEL_36": 36,
95
+ "LABEL_37": 37,
96
+ "LABEL_38": 38,
97
+ "LABEL_39": 39,
98
+ "LABEL_4": 4,
99
+ "LABEL_40": 40,
100
+ "LABEL_41": 41,
101
+ "LABEL_42": 42,
102
+ "LABEL_43": 43,
103
+ "LABEL_44": 44,
104
+ "LABEL_45": 45,
105
+ "LABEL_46": 46,
106
+ "LABEL_47": 47,
107
+ "LABEL_5": 5,
108
+ "LABEL_6": 6,
109
+ "LABEL_7": 7,
110
+ "LABEL_8": 8,
111
+ "LABEL_9": 9
112
+ },
113
+ "layer_norm_eps": 1e-12,
114
+ "max_position_embeddings": 512,
115
+ "model_type": "bert",
116
+ "num_attention_heads": 12,
117
+ "num_hidden_layers": 12,
118
+ "pad_token_id": 0,
119
+ "position_embedding_type": "absolute",
120
+ "torch_dtype": "float32",
121
+ "transformers_version": "4.50.3",
122
+ "type_vocab_size": 2,
123
+ "use_cache": true,
124
+ "vocab_size": 30522
125
+ }
run-0/checkpoint-5400/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a326440ce577fe1132a12ebf4a81327be2d2e573645aa74e5dd5bab7b618e01
3
+ size 438100144
run-0/checkpoint-5400/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6582319381d2846e45aa01db33bbd32baf6e27d68ebd8bfa1084e106636c70aa
3
+ size 876321402
run-0/checkpoint-5400/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9bdfa8c98b3727f6b5b88ab051ab63440a2cae5036de58802dc4de012c651ee
3
+ size 14244
run-0/checkpoint-5400/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42d56a21a2ea57f9d44b815019499ae7e6f44931fb0bfae7a8f5927e0d13f9a4
3
+ size 988
run-0/checkpoint-5400/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:483ad395cc0e39df116cda5b60b19830067092f467ad75d9686f6f8ba24c8d83
3
+ size 1064
run-0/checkpoint-5400/trainer_state.json ADDED
@@ -0,0 +1,731 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 5400,
3
+ "best_metric": 0.9394140341920821,
4
+ "best_model_checkpoint": "./results/run-0/checkpoint-5400",
5
+ "epoch": 9.0,
6
+ "eval_steps": 500,
7
+ "global_step": 5400,
8
+ "is_hyper_param_search": true,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.16673614005835766,
14
+ "grad_norm": 2.8743319511413574,
15
+ "learning_rate": 6e-06,
16
+ "loss": 1.4727,
17
+ "step": 100
18
+ },
19
+ {
20
+ "epoch": 0.3334722801167153,
21
+ "grad_norm": 0.8747783899307251,
22
+ "learning_rate": 1.2e-05,
23
+ "loss": 0.617,
24
+ "step": 200
25
+ },
26
+ {
27
+ "epoch": 0.5002084201750729,
28
+ "grad_norm": 0.7884620428085327,
29
+ "learning_rate": 1.8e-05,
30
+ "loss": 0.4493,
31
+ "step": 300
32
+ },
33
+ {
34
+ "epoch": 0.6669445602334306,
35
+ "grad_norm": 0.8924545645713806,
36
+ "learning_rate": 2.4e-05,
37
+ "loss": 0.3876,
38
+ "step": 400
39
+ },
40
+ {
41
+ "epoch": 0.8336807002917882,
42
+ "grad_norm": 0.8031777739524841,
43
+ "learning_rate": 3e-05,
44
+ "loss": 0.3361,
45
+ "step": 500
46
+ },
47
+ {
48
+ "epoch": 1.0,
49
+ "grad_norm": 0.5999459624290466,
50
+ "learning_rate": 2.9453551912568308e-05,
51
+ "loss": 0.3004,
52
+ "step": 600
53
+ },
54
+ {
55
+ "epoch": 1.0,
56
+ "eval_administration_accuracy": 0.9140087554721701,
57
+ "eval_administration_f1": 0.8740363354524873,
58
+ "eval_corruption_accuracy": 0.9370439858244737,
59
+ "eval_corruption_f1": 0.9192595640701061,
60
+ "eval_democracy_accuracy": 0.9347508859703981,
61
+ "eval_democracy_f1": 0.9095054688749907,
62
+ "eval_development_accuracy": 0.8763810715030227,
63
+ "eval_development_f1": 0.8253993969919758,
64
+ "eval_economy_accuracy": 0.9010840108401084,
65
+ "eval_economy_f1": 0.8737219947600353,
66
+ "eval_education_accuracy": 0.9571607254534084,
67
+ "eval_education_f1": 0.9485581438064299,
68
+ "eval_environment_accuracy": 0.9734208880550344,
69
+ "eval_environment_f1": 0.9693817386774968,
70
+ "eval_instability_accuracy": 0.9219303731498854,
71
+ "eval_instability_f1": 0.8878400653107733,
72
+ "eval_leadership_accuracy": 0.7652699603919116,
73
+ "eval_leadership_f1": 0.7080044131422123,
74
+ "eval_loss": 0.2839118242263794,
75
+ "eval_overall_accuracy": 0.9120109790841496,
76
+ "eval_overall_f1": 0.8849771384459072,
77
+ "eval_race_accuracy": 0.9283927454659162,
78
+ "eval_race_f1": 0.907176591968042,
79
+ "eval_religion_accuracy": 0.9369397540129247,
80
+ "eval_religion_f1": 0.9217332965382043,
81
+ "eval_runtime": 10.9229,
82
+ "eval_safety_accuracy": 0.8977485928705441,
83
+ "eval_safety_f1": 0.8751086517581315,
84
+ "eval_samples_per_second": 878.336,
85
+ "eval_steps_per_second": 54.93,
86
+ "step": 600
87
+ },
88
+ {
89
+ "epoch": 1.1667361400583576,
90
+ "grad_norm": 0.6991692781448364,
91
+ "learning_rate": 2.890710382513661e-05,
92
+ "loss": 0.2667,
93
+ "step": 700
94
+ },
95
+ {
96
+ "epoch": 1.3334722801167154,
97
+ "grad_norm": 0.6808522343635559,
98
+ "learning_rate": 2.836065573770492e-05,
99
+ "loss": 0.2573,
100
+ "step": 800
101
+ },
102
+ {
103
+ "epoch": 1.500208420175073,
104
+ "grad_norm": 0.8924445509910583,
105
+ "learning_rate": 2.7814207650273226e-05,
106
+ "loss": 0.2447,
107
+ "step": 900
108
+ },
109
+ {
110
+ "epoch": 1.6669445602334307,
111
+ "grad_norm": 0.9563364386558533,
112
+ "learning_rate": 2.7267759562841533e-05,
113
+ "loss": 0.2282,
114
+ "step": 1000
115
+ },
116
+ {
117
+ "epoch": 1.8336807002917883,
118
+ "grad_norm": 0.8881794810295105,
119
+ "learning_rate": 2.6721311475409837e-05,
120
+ "loss": 0.2217,
121
+ "step": 1100
122
+ },
123
+ {
124
+ "epoch": 2.0,
125
+ "grad_norm": 1.2946265935897827,
126
+ "learning_rate": 2.6174863387978144e-05,
127
+ "loss": 0.2197,
128
+ "step": 1200
129
+ },
130
+ {
131
+ "epoch": 2.0,
132
+ "eval_administration_accuracy": 0.9364185949551803,
133
+ "eval_administration_f1": 0.9264664888238818,
134
+ "eval_corruption_accuracy": 0.9562226391494684,
135
+ "eval_corruption_f1": 0.9508457468931831,
136
+ "eval_democracy_accuracy": 0.947571398790911,
137
+ "eval_democracy_f1": 0.9355668234958175,
138
+ "eval_development_accuracy": 0.9049405878674172,
139
+ "eval_development_f1": 0.8818640965909518,
140
+ "eval_economy_accuracy": 0.924119241192412,
141
+ "eval_economy_f1": 0.9144295841480942,
142
+ "eval_education_accuracy": 0.9669585157390036,
143
+ "eval_education_f1": 0.9649722301773493,
144
+ "eval_environment_accuracy": 0.981029810298103,
145
+ "eval_environment_f1": 0.9786767163906938,
146
+ "eval_instability_accuracy": 0.9422555764019178,
147
+ "eval_instability_f1": 0.9320449284299851,
148
+ "eval_leadership_accuracy": 0.8006045445069835,
149
+ "eval_leadership_f1": 0.7682063698028097,
150
+ "eval_loss": 0.21895167231559753,
151
+ "eval_overall_accuracy": 0.9317629073726635,
152
+ "eval_overall_f1": 0.9213111296186804,
153
+ "eval_race_accuracy": 0.9516364394413175,
154
+ "eval_race_f1": 0.9462871699636376,
155
+ "eval_religion_accuracy": 0.9586199708150928,
156
+ "eval_religion_f1": 0.9568464833874664,
157
+ "eval_runtime": 10.9295,
158
+ "eval_safety_accuracy": 0.9107775693141547,
159
+ "eval_safety_f1": 0.8995269173202963,
160
+ "eval_samples_per_second": 877.809,
161
+ "eval_steps_per_second": 54.897,
162
+ "step": 1200
163
+ },
164
+ {
165
+ "epoch": 2.1667361400583576,
166
+ "grad_norm": 0.8773576021194458,
167
+ "learning_rate": 2.5628415300546447e-05,
168
+ "loss": 0.1841,
169
+ "step": 1300
170
+ },
171
+ {
172
+ "epoch": 2.333472280116715,
173
+ "grad_norm": 0.7541666030883789,
174
+ "learning_rate": 2.5081967213114754e-05,
175
+ "loss": 0.1757,
176
+ "step": 1400
177
+ },
178
+ {
179
+ "epoch": 2.500208420175073,
180
+ "grad_norm": 0.9500707983970642,
181
+ "learning_rate": 2.453551912568306e-05,
182
+ "loss": 0.1776,
183
+ "step": 1500
184
+ },
185
+ {
186
+ "epoch": 2.6669445602334307,
187
+ "grad_norm": 0.8599845170974731,
188
+ "learning_rate": 2.398907103825137e-05,
189
+ "loss": 0.1757,
190
+ "step": 1600
191
+ },
192
+ {
193
+ "epoch": 2.8336807002917883,
194
+ "grad_norm": 1.0130952596664429,
195
+ "learning_rate": 2.3442622950819672e-05,
196
+ "loss": 0.1675,
197
+ "step": 1700
198
+ },
199
+ {
200
+ "epoch": 3.0,
201
+ "grad_norm": 1.4208967685699463,
202
+ "learning_rate": 2.289617486338798e-05,
203
+ "loss": 0.1664,
204
+ "step": 1800
205
+ },
206
+ {
207
+ "epoch": 3.0,
208
+ "eval_administration_accuracy": 0.940379403794038,
209
+ "eval_administration_f1": 0.9332522249114288,
210
+ "eval_corruption_accuracy": 0.9609130706691682,
211
+ "eval_corruption_f1": 0.9591571276198037,
212
+ "eval_democracy_accuracy": 0.9572649572649573,
213
+ "eval_democracy_f1": 0.951957009924363,
214
+ "eval_development_accuracy": 0.9184907233687721,
215
+ "eval_development_f1": 0.909063933465515,
216
+ "eval_economy_accuracy": 0.9327704815509693,
217
+ "eval_economy_f1": 0.9291641275521808,
218
+ "eval_education_accuracy": 0.967479674796748,
219
+ "eval_education_f1": 0.965512166473417,
220
+ "eval_environment_accuracy": 0.9835313737752762,
221
+ "eval_environment_f1": 0.9820983135141137,
222
+ "eval_instability_accuracy": 0.9481967896602043,
223
+ "eval_instability_f1": 0.9444250756726008,
224
+ "eval_leadership_accuracy": 0.8285386700020846,
225
+ "eval_leadership_f1": 0.8147910628849035,
226
+ "eval_loss": 0.19577622413635254,
227
+ "eval_overall_accuracy": 0.9393283996942533,
228
+ "eval_overall_f1": 0.9346706078324559,
229
+ "eval_race_accuracy": 0.9551803210339795,
230
+ "eval_race_f1": 0.9521792365558912,
231
+ "eval_religion_accuracy": 0.9598707525536794,
232
+ "eval_religion_f1": 0.9582958411796174,
233
+ "eval_runtime": 10.979,
234
+ "eval_safety_accuracy": 0.9193245778611632,
235
+ "eval_safety_f1": 0.9161511742356354,
236
+ "eval_samples_per_second": 873.853,
237
+ "eval_steps_per_second": 54.65,
238
+ "step": 1800
239
+ },
240
+ {
241
+ "epoch": 3.1667361400583576,
242
+ "grad_norm": 0.7676725387573242,
243
+ "learning_rate": 2.2349726775956283e-05,
244
+ "loss": 0.137,
245
+ "step": 1900
246
+ },
247
+ {
248
+ "epoch": 3.333472280116715,
249
+ "grad_norm": 0.9001705050468445,
250
+ "learning_rate": 2.180327868852459e-05,
251
+ "loss": 0.1366,
252
+ "step": 2000
253
+ },
254
+ {
255
+ "epoch": 3.500208420175073,
256
+ "grad_norm": 0.9584451913833618,
257
+ "learning_rate": 2.1256830601092897e-05,
258
+ "loss": 0.1319,
259
+ "step": 2100
260
+ },
261
+ {
262
+ "epoch": 3.6669445602334307,
263
+ "grad_norm": 0.9352701306343079,
264
+ "learning_rate": 2.0710382513661204e-05,
265
+ "loss": 0.1324,
266
+ "step": 2200
267
+ },
268
+ {
269
+ "epoch": 3.8336807002917883,
270
+ "grad_norm": 0.8512120842933655,
271
+ "learning_rate": 2.0163934426229508e-05,
272
+ "loss": 0.1333,
273
+ "step": 2300
274
+ },
275
+ {
276
+ "epoch": 4.0,
277
+ "grad_norm": 0.8892608880996704,
278
+ "learning_rate": 1.9617486338797815e-05,
279
+ "loss": 0.1266,
280
+ "step": 2400
281
+ },
282
+ {
283
+ "epoch": 4.0,
284
+ "eval_administration_accuracy": 0.9418386491557224,
285
+ "eval_administration_f1": 0.9352052966447959,
286
+ "eval_corruption_accuracy": 0.9622680842193038,
287
+ "eval_corruption_f1": 0.9610755105054823,
288
+ "eval_democracy_accuracy": 0.9578903481342506,
289
+ "eval_democracy_f1": 0.9547168679700065,
290
+ "eval_development_accuracy": 0.9218261413383365,
291
+ "eval_development_f1": 0.9169350889919825,
292
+ "eval_economy_accuracy": 0.9337085678549093,
293
+ "eval_economy_f1": 0.9328708122199544,
294
+ "eval_education_accuracy": 0.9685219929122368,
295
+ "eval_education_f1": 0.9667154036014268,
296
+ "eval_environment_accuracy": 0.9815509693558474,
297
+ "eval_environment_f1": 0.9809518427248292,
298
+ "eval_instability_accuracy": 0.947571398790911,
299
+ "eval_instability_f1": 0.9441507785991868,
300
+ "eval_leadership_accuracy": 0.8358348968105066,
301
+ "eval_leadership_f1": 0.82702607509575,
302
+ "eval_loss": 0.1933656632900238,
303
+ "eval_overall_accuracy": 0.9413782919880481,
304
+ "eval_overall_f1": 0.9381715218630767,
305
+ "eval_race_accuracy": 0.9591411298728372,
306
+ "eval_race_f1": 0.9571038116721658,
307
+ "eval_religion_accuracy": 0.9621638524077548,
308
+ "eval_religion_f1": 0.9611025172696737,
309
+ "eval_runtime": 10.9996,
310
+ "eval_safety_accuracy": 0.9242234730039608,
311
+ "eval_safety_f1": 0.920204257061666,
312
+ "eval_samples_per_second": 872.211,
313
+ "eval_steps_per_second": 54.547,
314
+ "step": 2400
315
+ },
316
+ {
317
+ "epoch": 4.166736140058358,
318
+ "grad_norm": 0.6617388129234314,
319
+ "learning_rate": 1.907103825136612e-05,
320
+ "loss": 0.1039,
321
+ "step": 2500
322
+ },
323
+ {
324
+ "epoch": 4.333472280116715,
325
+ "grad_norm": 0.7575182914733887,
326
+ "learning_rate": 1.8524590163934426e-05,
327
+ "loss": 0.1105,
328
+ "step": 2600
329
+ },
330
+ {
331
+ "epoch": 4.500208420175073,
332
+ "grad_norm": 0.69528728723526,
333
+ "learning_rate": 1.7978142076502733e-05,
334
+ "loss": 0.1084,
335
+ "step": 2700
336
+ },
337
+ {
338
+ "epoch": 4.66694456023343,
339
+ "grad_norm": 0.9533807039260864,
340
+ "learning_rate": 1.743169398907104e-05,
341
+ "loss": 0.1051,
342
+ "step": 2800
343
+ },
344
+ {
345
+ "epoch": 4.833680700291788,
346
+ "grad_norm": 0.7591508030891418,
347
+ "learning_rate": 1.6885245901639344e-05,
348
+ "loss": 0.1008,
349
+ "step": 2900
350
+ },
351
+ {
352
+ "epoch": 5.0,
353
+ "grad_norm": 1.3957374095916748,
354
+ "learning_rate": 1.633879781420765e-05,
355
+ "loss": 0.1013,
356
+ "step": 3000
357
+ },
358
+ {
359
+ "epoch": 5.0,
360
+ "eval_administration_accuracy": 0.9406920992286846,
361
+ "eval_administration_f1": 0.935638525668697,
362
+ "eval_corruption_accuracy": 0.9626850114654992,
363
+ "eval_corruption_f1": 0.9611422434993832,
364
+ "eval_democracy_accuracy": 0.9496560350218887,
365
+ "eval_democracy_f1": 0.9499559979035322,
366
+ "eval_development_accuracy": 0.9201584323535543,
367
+ "eval_development_f1": 0.9155534426644083,
368
+ "eval_economy_accuracy": 0.9354805086512403,
369
+ "eval_economy_f1": 0.9325398706072688,
370
+ "eval_education_accuracy": 0.9697727746508235,
371
+ "eval_education_f1": 0.9681349456850368,
372
+ "eval_environment_accuracy": 0.9820721284135918,
373
+ "eval_environment_f1": 0.9811885272501003,
374
+ "eval_instability_accuracy": 0.9472587033562644,
375
+ "eval_instability_f1": 0.9442377846524443,
376
+ "eval_leadership_accuracy": 0.8364602876797999,
377
+ "eval_leadership_f1": 0.828026933450708,
378
+ "eval_loss": 0.19940470159053802,
379
+ "eval_overall_accuracy": 0.9405183795427697,
380
+ "eval_overall_f1": 0.9377003414537354,
381
+ "eval_race_accuracy": 0.9554930164686263,
382
+ "eval_race_f1": 0.9547403715439807,
383
+ "eval_religion_accuracy": 0.9633104023347926,
384
+ "eval_religion_f1": 0.9620625852940151,
385
+ "eval_runtime": 10.9306,
386
+ "eval_safety_accuracy": 0.923181154888472,
387
+ "eval_safety_f1": 0.9191828692252488,
388
+ "eval_samples_per_second": 877.722,
389
+ "eval_steps_per_second": 54.892,
390
+ "step": 3000
391
+ },
392
+ {
393
+ "epoch": 5.166736140058358,
394
+ "grad_norm": 0.5871491432189941,
395
+ "learning_rate": 1.5792349726775955e-05,
396
+ "loss": 0.0873,
397
+ "step": 3100
398
+ },
399
+ {
400
+ "epoch": 5.333472280116715,
401
+ "grad_norm": 0.8581504225730896,
402
+ "learning_rate": 1.5245901639344264e-05,
403
+ "loss": 0.0862,
404
+ "step": 3200
405
+ },
406
+ {
407
+ "epoch": 5.500208420175073,
408
+ "grad_norm": 0.9329476952552795,
409
+ "learning_rate": 1.4699453551912569e-05,
410
+ "loss": 0.0849,
411
+ "step": 3300
412
+ },
413
+ {
414
+ "epoch": 5.66694456023343,
415
+ "grad_norm": 0.8767560124397278,
416
+ "learning_rate": 1.4153005464480874e-05,
417
+ "loss": 0.088,
418
+ "step": 3400
419
+ },
420
+ {
421
+ "epoch": 5.833680700291788,
422
+ "grad_norm": 0.8050324320793152,
423
+ "learning_rate": 1.3606557377049181e-05,
424
+ "loss": 0.0826,
425
+ "step": 3500
426
+ },
427
+ {
428
+ "epoch": 6.0,
429
+ "grad_norm": 0.9876068830490112,
430
+ "learning_rate": 1.3060109289617487e-05,
431
+ "loss": 0.0811,
432
+ "step": 3600
433
+ },
434
+ {
435
+ "epoch": 6.0,
436
+ "eval_administration_accuracy": 0.9434021263289556,
437
+ "eval_administration_f1": 0.9387369505688391,
438
+ "eval_corruption_accuracy": 0.962997706900146,
439
+ "eval_corruption_f1": 0.9612502739505763,
440
+ "eval_democracy_accuracy": 0.9562226391494684,
441
+ "eval_democracy_f1": 0.954550595976141,
442
+ "eval_development_accuracy": 0.9203668959766521,
443
+ "eval_development_f1": 0.9144818227568411,
444
+ "eval_economy_accuracy": 0.9361058995205337,
445
+ "eval_economy_f1": 0.9313226219336369,
446
+ "eval_education_accuracy": 0.9694600792161768,
447
+ "eval_education_f1": 0.9675599527596068,
448
+ "eval_environment_accuracy": 0.9811340421096518,
449
+ "eval_environment_f1": 0.9803282887818994,
450
+ "eval_instability_accuracy": 0.9459036898061288,
451
+ "eval_instability_f1": 0.942472499866658,
452
+ "eval_leadership_accuracy": 0.8430268918073797,
453
+ "eval_leadership_f1": 0.8387770890259914,
454
+ "eval_loss": 0.20078489184379578,
455
+ "eval_overall_accuracy": 0.941647557501216,
456
+ "eval_overall_f1": 0.938780687918615,
457
+ "eval_race_accuracy": 0.9580988117573483,
458
+ "eval_race_f1": 0.9562411064132649,
459
+ "eval_religion_accuracy": 0.961955388784657,
460
+ "eval_religion_f1": 0.960819752774671,
461
+ "eval_runtime": 11.0615,
462
+ "eval_safety_accuracy": 0.9210965186574943,
463
+ "eval_safety_f1": 0.9188273002152547,
464
+ "eval_samples_per_second": 867.33,
465
+ "eval_steps_per_second": 54.242,
466
+ "step": 3600
467
+ },
468
+ {
469
+ "epoch": 6.166736140058358,
470
+ "grad_norm": 0.8045951724052429,
471
+ "learning_rate": 1.2513661202185792e-05,
472
+ "loss": 0.0701,
473
+ "step": 3700
474
+ },
475
+ {
476
+ "epoch": 6.333472280116715,
477
+ "grad_norm": 0.6562448143959045,
478
+ "learning_rate": 1.19672131147541e-05,
479
+ "loss": 0.0726,
480
+ "step": 3800
481
+ },
482
+ {
483
+ "epoch": 6.500208420175073,
484
+ "grad_norm": 0.7833376526832581,
485
+ "learning_rate": 1.1420765027322405e-05,
486
+ "loss": 0.0673,
487
+ "step": 3900
488
+ },
489
+ {
490
+ "epoch": 6.66694456023343,
491
+ "grad_norm": 0.8071606755256653,
492
+ "learning_rate": 1.087431693989071e-05,
493
+ "loss": 0.0733,
494
+ "step": 4000
495
+ },
496
+ {
497
+ "epoch": 6.833680700291788,
498
+ "grad_norm": 0.7821515798568726,
499
+ "learning_rate": 1.0327868852459017e-05,
500
+ "loss": 0.0678,
501
+ "step": 4100
502
+ },
503
+ {
504
+ "epoch": 7.0,
505
+ "grad_norm": 0.6441165804862976,
506
+ "learning_rate": 9.781420765027323e-06,
507
+ "loss": 0.0701,
508
+ "step": 4200
509
+ },
510
+ {
511
+ "epoch": 7.0,
512
+ "eval_administration_accuracy": 0.9396497811131957,
513
+ "eval_administration_f1": 0.9371884848945066,
514
+ "eval_corruption_accuracy": 0.9616426933500104,
515
+ "eval_corruption_f1": 0.9603149034436674,
516
+ "eval_democracy_accuracy": 0.9548676255993329,
517
+ "eval_democracy_f1": 0.9544642189136688,
518
+ "eval_development_accuracy": 0.919533041484261,
519
+ "eval_development_f1": 0.9174833579112073,
520
+ "eval_economy_accuracy": 0.9370439858244737,
521
+ "eval_economy_f1": 0.9350264441971636,
522
+ "eval_education_accuracy": 0.9670627475505524,
523
+ "eval_education_f1": 0.9657003505198045,
524
+ "eval_environment_accuracy": 0.9821763602251408,
525
+ "eval_environment_f1": 0.9808859370812918,
526
+ "eval_instability_accuracy": 0.9445486762559934,
527
+ "eval_instability_f1": 0.9432538618242539,
528
+ "eval_leadership_accuracy": 0.8441734417344173,
529
+ "eval_leadership_f1": 0.8426503273307849,
530
+ "eval_loss": 0.21108125150203705,
531
+ "eval_overall_accuracy": 0.9408397609617123,
532
+ "eval_overall_f1": 0.9393817348962156,
533
+ "eval_race_accuracy": 0.9566395663956639,
534
+ "eval_race_f1": 0.9559208826126899,
535
+ "eval_religion_accuracy": 0.9609130706691682,
536
+ "eval_religion_f1": 0.9608002073614549,
537
+ "eval_runtime": 10.9796,
538
+ "eval_safety_accuracy": 0.9218261413383365,
539
+ "eval_safety_f1": 0.9188918426640929,
540
+ "eval_samples_per_second": 873.8,
541
+ "eval_steps_per_second": 54.647,
542
+ "step": 4200
543
+ },
544
+ {
545
+ "epoch": 7.166736140058358,
546
+ "grad_norm": 0.6236246228218079,
547
+ "learning_rate": 9.234972677595628e-06,
548
+ "loss": 0.0597,
549
+ "step": 4300
550
+ },
551
+ {
552
+ "epoch": 7.333472280116715,
553
+ "grad_norm": 0.6468791365623474,
554
+ "learning_rate": 8.688524590163935e-06,
555
+ "loss": 0.0568,
556
+ "step": 4400
557
+ },
558
+ {
559
+ "epoch": 7.500208420175073,
560
+ "grad_norm": 0.7850766777992249,
561
+ "learning_rate": 8.14207650273224e-06,
562
+ "loss": 0.0586,
563
+ "step": 4500
564
+ },
565
+ {
566
+ "epoch": 7.66694456023343,
567
+ "grad_norm": 0.9113643169403076,
568
+ "learning_rate": 7.595628415300546e-06,
569
+ "loss": 0.0611,
570
+ "step": 4600
571
+ },
572
+ {
573
+ "epoch": 7.833680700291788,
574
+ "grad_norm": 0.9405547976493835,
575
+ "learning_rate": 7.049180327868853e-06,
576
+ "loss": 0.0636,
577
+ "step": 4700
578
+ },
579
+ {
580
+ "epoch": 8.0,
581
+ "grad_norm": 0.5414742827415466,
582
+ "learning_rate": 6.502732240437159e-06,
583
+ "loss": 0.0588,
584
+ "step": 4800
585
+ },
586
+ {
587
+ "epoch": 8.0,
588
+ "eval_administration_accuracy": 0.9413174900979779,
589
+ "eval_administration_f1": 0.9368824702703686,
590
+ "eval_corruption_accuracy": 0.9614342297269126,
591
+ "eval_corruption_f1": 0.9602768664022149,
592
+ "eval_democracy_accuracy": 0.9560141755263707,
593
+ "eval_democracy_f1": 0.9543585350199038,
594
+ "eval_development_accuracy": 0.9192203460496143,
595
+ "eval_development_f1": 0.9152605716450639,
596
+ "eval_economy_accuracy": 0.9381905357515113,
597
+ "eval_economy_f1": 0.935106524044037,
598
+ "eval_education_accuracy": 0.9688346883468835,
599
+ "eval_education_f1": 0.9673054511905975,
600
+ "eval_environment_accuracy": 0.9822805920366896,
601
+ "eval_environment_f1": 0.980929327817527,
602
+ "eval_instability_accuracy": 0.9455909943714822,
603
+ "eval_instability_f1": 0.9437005892876725,
604
+ "eval_leadership_accuracy": 0.8438607462997707,
605
+ "eval_leadership_f1": 0.8410810165791273,
606
+ "eval_loss": 0.215087428689003,
607
+ "eval_overall_accuracy": 0.9414651518310054,
608
+ "eval_overall_f1": 0.93924957060805,
609
+ "eval_race_accuracy": 0.9572649572649573,
610
+ "eval_race_f1": 0.9565627258185468,
611
+ "eval_religion_accuracy": 0.9615384615384616,
612
+ "eval_religion_f1": 0.9610239728595763,
613
+ "eval_runtime": 10.9675,
614
+ "eval_safety_accuracy": 0.9220346049614342,
615
+ "eval_safety_f1": 0.9185067963619639,
616
+ "eval_samples_per_second": 874.763,
617
+ "eval_steps_per_second": 54.707,
618
+ "step": 4800
619
+ },
620
+ {
621
+ "epoch": 8.166736140058358,
622
+ "grad_norm": 0.5899946093559265,
623
+ "learning_rate": 5.956284153005465e-06,
624
+ "loss": 0.0543,
625
+ "step": 4900
626
+ },
627
+ {
628
+ "epoch": 8.333472280116716,
629
+ "grad_norm": 0.5821051597595215,
630
+ "learning_rate": 5.409836065573771e-06,
631
+ "loss": 0.0528,
632
+ "step": 5000
633
+ },
634
+ {
635
+ "epoch": 8.500208420175072,
636
+ "grad_norm": 0.7321776151657104,
637
+ "learning_rate": 4.863387978142076e-06,
638
+ "loss": 0.0532,
639
+ "step": 5100
640
+ },
641
+ {
642
+ "epoch": 8.66694456023343,
643
+ "grad_norm": 0.7479367256164551,
644
+ "learning_rate": 4.316939890710383e-06,
645
+ "loss": 0.0519,
646
+ "step": 5200
647
+ },
648
+ {
649
+ "epoch": 8.833680700291788,
650
+ "grad_norm": 0.8334552049636841,
651
+ "learning_rate": 3.770491803278689e-06,
652
+ "loss": 0.0505,
653
+ "step": 5300
654
+ },
655
+ {
656
+ "epoch": 9.0,
657
+ "grad_norm": 0.7611594200134277,
658
+ "learning_rate": 3.2240437158469947e-06,
659
+ "loss": 0.0528,
660
+ "step": 5400
661
+ },
662
+ {
663
+ "epoch": 9.0,
664
+ "eval_administration_accuracy": 0.9398582447362935,
665
+ "eval_administration_f1": 0.9370835475227389,
666
+ "eval_corruption_accuracy": 0.9618511569731082,
667
+ "eval_corruption_f1": 0.9605413145422365,
668
+ "eval_democracy_accuracy": 0.9559099437148217,
669
+ "eval_democracy_f1": 0.9549404791061389,
670
+ "eval_development_accuracy": 0.9212007504690432,
671
+ "eval_development_f1": 0.9170777974760537,
672
+ "eval_economy_accuracy": 0.9381905357515113,
673
+ "eval_economy_f1": 0.93592555672295,
674
+ "eval_education_accuracy": 0.9682092974775902,
675
+ "eval_education_f1": 0.9667368725950756,
676
+ "eval_environment_accuracy": 0.9813425057327496,
677
+ "eval_environment_f1": 0.9802272421649649,
678
+ "eval_instability_accuracy": 0.9472587033562644,
679
+ "eval_instability_f1": 0.9448551688210497,
680
+ "eval_leadership_accuracy": 0.8438607462997707,
681
+ "eval_leadership_f1": 0.8412354371277554,
682
+ "eval_loss": 0.21861030161380768,
683
+ "eval_overall_accuracy": 0.9413609200194567,
684
+ "eval_overall_f1": 0.9394140341920821,
685
+ "eval_race_accuracy": 0.9572649572649573,
686
+ "eval_race_f1": 0.9563061857501846,
687
+ "eval_religion_accuracy": 0.961955388784657,
688
+ "eval_religion_f1": 0.9612659459831671,
689
+ "eval_runtime": 11.1696,
690
+ "eval_safety_accuracy": 0.9194288096727121,
691
+ "eval_safety_f1": 0.9167728624926694,
692
+ "eval_samples_per_second": 858.939,
693
+ "eval_steps_per_second": 53.717,
694
+ "step": 5400
695
+ }
696
+ ],
697
+ "logging_steps": 100,
698
+ "max_steps": 5990,
699
+ "num_input_tokens_seen": 0,
700
+ "num_train_epochs": 10,
701
+ "save_steps": 500,
702
+ "stateful_callbacks": {
703
+ "EarlyStoppingCallback": {
704
+ "args": {
705
+ "early_stopping_patience": 2,
706
+ "early_stopping_threshold": 0.0
707
+ },
708
+ "attributes": {
709
+ "early_stopping_patience_counter": 0
710
+ }
711
+ },
712
+ "TrainerControl": {
713
+ "args": {
714
+ "should_epoch_stop": false,
715
+ "should_evaluate": false,
716
+ "should_log": false,
717
+ "should_save": true,
718
+ "should_training_stop": false
719
+ },
720
+ "attributes": {}
721
+ }
722
+ },
723
+ "total_flos": 9.090240534341222e+16,
724
+ "train_batch_size": 16,
725
+ "trial_name": null,
726
+ "trial_params": {
727
+ "gradient_accumulation_steps": 4,
728
+ "learning_rate": 3e-05,
729
+ "num_train_epochs": 10
730
+ }
731
+ }
run-0/checkpoint-5400/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:82c268ed019d067f52ee3b6ec06e4f32fa79f82839c235ad890040984c5ac1cf
3
+ size 5368
run-0/checkpoint-5990/config.json ADDED
@@ -0,0 +1,125 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "BertForSequenceClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.1,
6
+ "classifier_dropout": null,
7
+ "gradient_checkpointing": false,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.1,
10
+ "hidden_size": 768,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47"
60
+ },
61
+ "initializer_range": 0.02,
62
+ "intermediate_size": 3072,
63
+ "label2id": {
64
+ "LABEL_0": 0,
65
+ "LABEL_1": 1,
66
+ "LABEL_10": 10,
67
+ "LABEL_11": 11,
68
+ "LABEL_12": 12,
69
+ "LABEL_13": 13,
70
+ "LABEL_14": 14,
71
+ "LABEL_15": 15,
72
+ "LABEL_16": 16,
73
+ "LABEL_17": 17,
74
+ "LABEL_18": 18,
75
+ "LABEL_19": 19,
76
+ "LABEL_2": 2,
77
+ "LABEL_20": 20,
78
+ "LABEL_21": 21,
79
+ "LABEL_22": 22,
80
+ "LABEL_23": 23,
81
+ "LABEL_24": 24,
82
+ "LABEL_25": 25,
83
+ "LABEL_26": 26,
84
+ "LABEL_27": 27,
85
+ "LABEL_28": 28,
86
+ "LABEL_29": 29,
87
+ "LABEL_3": 3,
88
+ "LABEL_30": 30,
89
+ "LABEL_31": 31,
90
+ "LABEL_32": 32,
91
+ "LABEL_33": 33,
92
+ "LABEL_34": 34,
93
+ "LABEL_35": 35,
94
+ "LABEL_36": 36,
95
+ "LABEL_37": 37,
96
+ "LABEL_38": 38,
97
+ "LABEL_39": 39,
98
+ "LABEL_4": 4,
99
+ "LABEL_40": 40,
100
+ "LABEL_41": 41,
101
+ "LABEL_42": 42,
102
+ "LABEL_43": 43,
103
+ "LABEL_44": 44,
104
+ "LABEL_45": 45,
105
+ "LABEL_46": 46,
106
+ "LABEL_47": 47,
107
+ "LABEL_5": 5,
108
+ "LABEL_6": 6,
109
+ "LABEL_7": 7,
110
+ "LABEL_8": 8,
111
+ "LABEL_9": 9
112
+ },
113
+ "layer_norm_eps": 1e-12,
114
+ "max_position_embeddings": 512,
115
+ "model_type": "bert",
116
+ "num_attention_heads": 12,
117
+ "num_hidden_layers": 12,
118
+ "pad_token_id": 0,
119
+ "position_embedding_type": "absolute",
120
+ "torch_dtype": "float32",
121
+ "transformers_version": "4.50.3",
122
+ "type_vocab_size": 2,
123
+ "use_cache": true,
124
+ "vocab_size": 30522
125
+ }
run-0/checkpoint-5990/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c5eed0fdb488201be108caf8e4b8b6f71bdfc8ce05930b5a153bd48fa638748
3
+ size 438100144
run-0/checkpoint-5990/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d48796d4a45ff5b18bff9dd888203ac54559beb4ef9ed371964b590c155cac99
3
+ size 876321402
run-0/checkpoint-5990/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4125a41dc6b2d88e3e9fbfd89ec6713a48035ed11c30377f6097db203d06d76e
3
+ size 14244
run-0/checkpoint-5990/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a6baea4f8b9d8d35728abb8f2514128a9c1c13b5ba92ec2d440fe598ed4ad9c7
3
+ size 988
run-0/checkpoint-5990/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3326985dc7605f2a57fcb46b9e919b9daf5b7aad0fe092c0efccca80f56b78fe
3
+ size 1064
run-0/checkpoint-5990/trainer_state.json ADDED
@@ -0,0 +1,800 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 5990,
3
+ "best_metric": 0.9394484894664766,
4
+ "best_model_checkpoint": "./results/run-0/checkpoint-5990",
5
+ "epoch": 9.98374322634431,
6
+ "eval_steps": 500,
7
+ "global_step": 5990,
8
+ "is_hyper_param_search": true,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.16673614005835766,
14
+ "grad_norm": 2.8743319511413574,
15
+ "learning_rate": 6e-06,
16
+ "loss": 1.4727,
17
+ "step": 100
18
+ },
19
+ {
20
+ "epoch": 0.3334722801167153,
21
+ "grad_norm": 0.8747783899307251,
22
+ "learning_rate": 1.2e-05,
23
+ "loss": 0.617,
24
+ "step": 200
25
+ },
26
+ {
27
+ "epoch": 0.5002084201750729,
28
+ "grad_norm": 0.7884620428085327,
29
+ "learning_rate": 1.8e-05,
30
+ "loss": 0.4493,
31
+ "step": 300
32
+ },
33
+ {
34
+ "epoch": 0.6669445602334306,
35
+ "grad_norm": 0.8924545645713806,
36
+ "learning_rate": 2.4e-05,
37
+ "loss": 0.3876,
38
+ "step": 400
39
+ },
40
+ {
41
+ "epoch": 0.8336807002917882,
42
+ "grad_norm": 0.8031777739524841,
43
+ "learning_rate": 3e-05,
44
+ "loss": 0.3361,
45
+ "step": 500
46
+ },
47
+ {
48
+ "epoch": 1.0,
49
+ "grad_norm": 0.5999459624290466,
50
+ "learning_rate": 2.9453551912568308e-05,
51
+ "loss": 0.3004,
52
+ "step": 600
53
+ },
54
+ {
55
+ "epoch": 1.0,
56
+ "eval_administration_accuracy": 0.9140087554721701,
57
+ "eval_administration_f1": 0.8740363354524873,
58
+ "eval_corruption_accuracy": 0.9370439858244737,
59
+ "eval_corruption_f1": 0.9192595640701061,
60
+ "eval_democracy_accuracy": 0.9347508859703981,
61
+ "eval_democracy_f1": 0.9095054688749907,
62
+ "eval_development_accuracy": 0.8763810715030227,
63
+ "eval_development_f1": 0.8253993969919758,
64
+ "eval_economy_accuracy": 0.9010840108401084,
65
+ "eval_economy_f1": 0.8737219947600353,
66
+ "eval_education_accuracy": 0.9571607254534084,
67
+ "eval_education_f1": 0.9485581438064299,
68
+ "eval_environment_accuracy": 0.9734208880550344,
69
+ "eval_environment_f1": 0.9693817386774968,
70
+ "eval_instability_accuracy": 0.9219303731498854,
71
+ "eval_instability_f1": 0.8878400653107733,
72
+ "eval_leadership_accuracy": 0.7652699603919116,
73
+ "eval_leadership_f1": 0.7080044131422123,
74
+ "eval_loss": 0.2839118242263794,
75
+ "eval_overall_accuracy": 0.9120109790841496,
76
+ "eval_overall_f1": 0.8849771384459072,
77
+ "eval_race_accuracy": 0.9283927454659162,
78
+ "eval_race_f1": 0.907176591968042,
79
+ "eval_religion_accuracy": 0.9369397540129247,
80
+ "eval_religion_f1": 0.9217332965382043,
81
+ "eval_runtime": 10.9229,
82
+ "eval_safety_accuracy": 0.8977485928705441,
83
+ "eval_safety_f1": 0.8751086517581315,
84
+ "eval_samples_per_second": 878.336,
85
+ "eval_steps_per_second": 54.93,
86
+ "step": 600
87
+ },
88
+ {
89
+ "epoch": 1.1667361400583576,
90
+ "grad_norm": 0.6991692781448364,
91
+ "learning_rate": 2.890710382513661e-05,
92
+ "loss": 0.2667,
93
+ "step": 700
94
+ },
95
+ {
96
+ "epoch": 1.3334722801167154,
97
+ "grad_norm": 0.6808522343635559,
98
+ "learning_rate": 2.836065573770492e-05,
99
+ "loss": 0.2573,
100
+ "step": 800
101
+ },
102
+ {
103
+ "epoch": 1.500208420175073,
104
+ "grad_norm": 0.8924445509910583,
105
+ "learning_rate": 2.7814207650273226e-05,
106
+ "loss": 0.2447,
107
+ "step": 900
108
+ },
109
+ {
110
+ "epoch": 1.6669445602334307,
111
+ "grad_norm": 0.9563364386558533,
112
+ "learning_rate": 2.7267759562841533e-05,
113
+ "loss": 0.2282,
114
+ "step": 1000
115
+ },
116
+ {
117
+ "epoch": 1.8336807002917883,
118
+ "grad_norm": 0.8881794810295105,
119
+ "learning_rate": 2.6721311475409837e-05,
120
+ "loss": 0.2217,
121
+ "step": 1100
122
+ },
123
+ {
124
+ "epoch": 2.0,
125
+ "grad_norm": 1.2946265935897827,
126
+ "learning_rate": 2.6174863387978144e-05,
127
+ "loss": 0.2197,
128
+ "step": 1200
129
+ },
130
+ {
131
+ "epoch": 2.0,
132
+ "eval_administration_accuracy": 0.9364185949551803,
133
+ "eval_administration_f1": 0.9264664888238818,
134
+ "eval_corruption_accuracy": 0.9562226391494684,
135
+ "eval_corruption_f1": 0.9508457468931831,
136
+ "eval_democracy_accuracy": 0.947571398790911,
137
+ "eval_democracy_f1": 0.9355668234958175,
138
+ "eval_development_accuracy": 0.9049405878674172,
139
+ "eval_development_f1": 0.8818640965909518,
140
+ "eval_economy_accuracy": 0.924119241192412,
141
+ "eval_economy_f1": 0.9144295841480942,
142
+ "eval_education_accuracy": 0.9669585157390036,
143
+ "eval_education_f1": 0.9649722301773493,
144
+ "eval_environment_accuracy": 0.981029810298103,
145
+ "eval_environment_f1": 0.9786767163906938,
146
+ "eval_instability_accuracy": 0.9422555764019178,
147
+ "eval_instability_f1": 0.9320449284299851,
148
+ "eval_leadership_accuracy": 0.8006045445069835,
149
+ "eval_leadership_f1": 0.7682063698028097,
150
+ "eval_loss": 0.21895167231559753,
151
+ "eval_overall_accuracy": 0.9317629073726635,
152
+ "eval_overall_f1": 0.9213111296186804,
153
+ "eval_race_accuracy": 0.9516364394413175,
154
+ "eval_race_f1": 0.9462871699636376,
155
+ "eval_religion_accuracy": 0.9586199708150928,
156
+ "eval_religion_f1": 0.9568464833874664,
157
+ "eval_runtime": 10.9295,
158
+ "eval_safety_accuracy": 0.9107775693141547,
159
+ "eval_safety_f1": 0.8995269173202963,
160
+ "eval_samples_per_second": 877.809,
161
+ "eval_steps_per_second": 54.897,
162
+ "step": 1200
163
+ },
164
+ {
165
+ "epoch": 2.1667361400583576,
166
+ "grad_norm": 0.8773576021194458,
167
+ "learning_rate": 2.5628415300546447e-05,
168
+ "loss": 0.1841,
169
+ "step": 1300
170
+ },
171
+ {
172
+ "epoch": 2.333472280116715,
173
+ "grad_norm": 0.7541666030883789,
174
+ "learning_rate": 2.5081967213114754e-05,
175
+ "loss": 0.1757,
176
+ "step": 1400
177
+ },
178
+ {
179
+ "epoch": 2.500208420175073,
180
+ "grad_norm": 0.9500707983970642,
181
+ "learning_rate": 2.453551912568306e-05,
182
+ "loss": 0.1776,
183
+ "step": 1500
184
+ },
185
+ {
186
+ "epoch": 2.6669445602334307,
187
+ "grad_norm": 0.8599845170974731,
188
+ "learning_rate": 2.398907103825137e-05,
189
+ "loss": 0.1757,
190
+ "step": 1600
191
+ },
192
+ {
193
+ "epoch": 2.8336807002917883,
194
+ "grad_norm": 1.0130952596664429,
195
+ "learning_rate": 2.3442622950819672e-05,
196
+ "loss": 0.1675,
197
+ "step": 1700
198
+ },
199
+ {
200
+ "epoch": 3.0,
201
+ "grad_norm": 1.4208967685699463,
202
+ "learning_rate": 2.289617486338798e-05,
203
+ "loss": 0.1664,
204
+ "step": 1800
205
+ },
206
+ {
207
+ "epoch": 3.0,
208
+ "eval_administration_accuracy": 0.940379403794038,
209
+ "eval_administration_f1": 0.9332522249114288,
210
+ "eval_corruption_accuracy": 0.9609130706691682,
211
+ "eval_corruption_f1": 0.9591571276198037,
212
+ "eval_democracy_accuracy": 0.9572649572649573,
213
+ "eval_democracy_f1": 0.951957009924363,
214
+ "eval_development_accuracy": 0.9184907233687721,
215
+ "eval_development_f1": 0.909063933465515,
216
+ "eval_economy_accuracy": 0.9327704815509693,
217
+ "eval_economy_f1": 0.9291641275521808,
218
+ "eval_education_accuracy": 0.967479674796748,
219
+ "eval_education_f1": 0.965512166473417,
220
+ "eval_environment_accuracy": 0.9835313737752762,
221
+ "eval_environment_f1": 0.9820983135141137,
222
+ "eval_instability_accuracy": 0.9481967896602043,
223
+ "eval_instability_f1": 0.9444250756726008,
224
+ "eval_leadership_accuracy": 0.8285386700020846,
225
+ "eval_leadership_f1": 0.8147910628849035,
226
+ "eval_loss": 0.19577622413635254,
227
+ "eval_overall_accuracy": 0.9393283996942533,
228
+ "eval_overall_f1": 0.9346706078324559,
229
+ "eval_race_accuracy": 0.9551803210339795,
230
+ "eval_race_f1": 0.9521792365558912,
231
+ "eval_religion_accuracy": 0.9598707525536794,
232
+ "eval_religion_f1": 0.9582958411796174,
233
+ "eval_runtime": 10.979,
234
+ "eval_safety_accuracy": 0.9193245778611632,
235
+ "eval_safety_f1": 0.9161511742356354,
236
+ "eval_samples_per_second": 873.853,
237
+ "eval_steps_per_second": 54.65,
238
+ "step": 1800
239
+ },
240
+ {
241
+ "epoch": 3.1667361400583576,
242
+ "grad_norm": 0.7676725387573242,
243
+ "learning_rate": 2.2349726775956283e-05,
244
+ "loss": 0.137,
245
+ "step": 1900
246
+ },
247
+ {
248
+ "epoch": 3.333472280116715,
249
+ "grad_norm": 0.9001705050468445,
250
+ "learning_rate": 2.180327868852459e-05,
251
+ "loss": 0.1366,
252
+ "step": 2000
253
+ },
254
+ {
255
+ "epoch": 3.500208420175073,
256
+ "grad_norm": 0.9584451913833618,
257
+ "learning_rate": 2.1256830601092897e-05,
258
+ "loss": 0.1319,
259
+ "step": 2100
260
+ },
261
+ {
262
+ "epoch": 3.6669445602334307,
263
+ "grad_norm": 0.9352701306343079,
264
+ "learning_rate": 2.0710382513661204e-05,
265
+ "loss": 0.1324,
266
+ "step": 2200
267
+ },
268
+ {
269
+ "epoch": 3.8336807002917883,
270
+ "grad_norm": 0.8512120842933655,
271
+ "learning_rate": 2.0163934426229508e-05,
272
+ "loss": 0.1333,
273
+ "step": 2300
274
+ },
275
+ {
276
+ "epoch": 4.0,
277
+ "grad_norm": 0.8892608880996704,
278
+ "learning_rate": 1.9617486338797815e-05,
279
+ "loss": 0.1266,
280
+ "step": 2400
281
+ },
282
+ {
283
+ "epoch": 4.0,
284
+ "eval_administration_accuracy": 0.9418386491557224,
285
+ "eval_administration_f1": 0.9352052966447959,
286
+ "eval_corruption_accuracy": 0.9622680842193038,
287
+ "eval_corruption_f1": 0.9610755105054823,
288
+ "eval_democracy_accuracy": 0.9578903481342506,
289
+ "eval_democracy_f1": 0.9547168679700065,
290
+ "eval_development_accuracy": 0.9218261413383365,
291
+ "eval_development_f1": 0.9169350889919825,
292
+ "eval_economy_accuracy": 0.9337085678549093,
293
+ "eval_economy_f1": 0.9328708122199544,
294
+ "eval_education_accuracy": 0.9685219929122368,
295
+ "eval_education_f1": 0.9667154036014268,
296
+ "eval_environment_accuracy": 0.9815509693558474,
297
+ "eval_environment_f1": 0.9809518427248292,
298
+ "eval_instability_accuracy": 0.947571398790911,
299
+ "eval_instability_f1": 0.9441507785991868,
300
+ "eval_leadership_accuracy": 0.8358348968105066,
301
+ "eval_leadership_f1": 0.82702607509575,
302
+ "eval_loss": 0.1933656632900238,
303
+ "eval_overall_accuracy": 0.9413782919880481,
304
+ "eval_overall_f1": 0.9381715218630767,
305
+ "eval_race_accuracy": 0.9591411298728372,
306
+ "eval_race_f1": 0.9571038116721658,
307
+ "eval_religion_accuracy": 0.9621638524077548,
308
+ "eval_religion_f1": 0.9611025172696737,
309
+ "eval_runtime": 10.9996,
310
+ "eval_safety_accuracy": 0.9242234730039608,
311
+ "eval_safety_f1": 0.920204257061666,
312
+ "eval_samples_per_second": 872.211,
313
+ "eval_steps_per_second": 54.547,
314
+ "step": 2400
315
+ },
316
+ {
317
+ "epoch": 4.166736140058358,
318
+ "grad_norm": 0.6617388129234314,
319
+ "learning_rate": 1.907103825136612e-05,
320
+ "loss": 0.1039,
321
+ "step": 2500
322
+ },
323
+ {
324
+ "epoch": 4.333472280116715,
325
+ "grad_norm": 0.7575182914733887,
326
+ "learning_rate": 1.8524590163934426e-05,
327
+ "loss": 0.1105,
328
+ "step": 2600
329
+ },
330
+ {
331
+ "epoch": 4.500208420175073,
332
+ "grad_norm": 0.69528728723526,
333
+ "learning_rate": 1.7978142076502733e-05,
334
+ "loss": 0.1084,
335
+ "step": 2700
336
+ },
337
+ {
338
+ "epoch": 4.66694456023343,
339
+ "grad_norm": 0.9533807039260864,
340
+ "learning_rate": 1.743169398907104e-05,
341
+ "loss": 0.1051,
342
+ "step": 2800
343
+ },
344
+ {
345
+ "epoch": 4.833680700291788,
346
+ "grad_norm": 0.7591508030891418,
347
+ "learning_rate": 1.6885245901639344e-05,
348
+ "loss": 0.1008,
349
+ "step": 2900
350
+ },
351
+ {
352
+ "epoch": 5.0,
353
+ "grad_norm": 1.3957374095916748,
354
+ "learning_rate": 1.633879781420765e-05,
355
+ "loss": 0.1013,
356
+ "step": 3000
357
+ },
358
+ {
359
+ "epoch": 5.0,
360
+ "eval_administration_accuracy": 0.9406920992286846,
361
+ "eval_administration_f1": 0.935638525668697,
362
+ "eval_corruption_accuracy": 0.9626850114654992,
363
+ "eval_corruption_f1": 0.9611422434993832,
364
+ "eval_democracy_accuracy": 0.9496560350218887,
365
+ "eval_democracy_f1": 0.9499559979035322,
366
+ "eval_development_accuracy": 0.9201584323535543,
367
+ "eval_development_f1": 0.9155534426644083,
368
+ "eval_economy_accuracy": 0.9354805086512403,
369
+ "eval_economy_f1": 0.9325398706072688,
370
+ "eval_education_accuracy": 0.9697727746508235,
371
+ "eval_education_f1": 0.9681349456850368,
372
+ "eval_environment_accuracy": 0.9820721284135918,
373
+ "eval_environment_f1": 0.9811885272501003,
374
+ "eval_instability_accuracy": 0.9472587033562644,
375
+ "eval_instability_f1": 0.9442377846524443,
376
+ "eval_leadership_accuracy": 0.8364602876797999,
377
+ "eval_leadership_f1": 0.828026933450708,
378
+ "eval_loss": 0.19940470159053802,
379
+ "eval_overall_accuracy": 0.9405183795427697,
380
+ "eval_overall_f1": 0.9377003414537354,
381
+ "eval_race_accuracy": 0.9554930164686263,
382
+ "eval_race_f1": 0.9547403715439807,
383
+ "eval_religion_accuracy": 0.9633104023347926,
384
+ "eval_religion_f1": 0.9620625852940151,
385
+ "eval_runtime": 10.9306,
386
+ "eval_safety_accuracy": 0.923181154888472,
387
+ "eval_safety_f1": 0.9191828692252488,
388
+ "eval_samples_per_second": 877.722,
389
+ "eval_steps_per_second": 54.892,
390
+ "step": 3000
391
+ },
392
+ {
393
+ "epoch": 5.166736140058358,
394
+ "grad_norm": 0.5871491432189941,
395
+ "learning_rate": 1.5792349726775955e-05,
396
+ "loss": 0.0873,
397
+ "step": 3100
398
+ },
399
+ {
400
+ "epoch": 5.333472280116715,
401
+ "grad_norm": 0.8581504225730896,
402
+ "learning_rate": 1.5245901639344264e-05,
403
+ "loss": 0.0862,
404
+ "step": 3200
405
+ },
406
+ {
407
+ "epoch": 5.500208420175073,
408
+ "grad_norm": 0.9329476952552795,
409
+ "learning_rate": 1.4699453551912569e-05,
410
+ "loss": 0.0849,
411
+ "step": 3300
412
+ },
413
+ {
414
+ "epoch": 5.66694456023343,
415
+ "grad_norm": 0.8767560124397278,
416
+ "learning_rate": 1.4153005464480874e-05,
417
+ "loss": 0.088,
418
+ "step": 3400
419
+ },
420
+ {
421
+ "epoch": 5.833680700291788,
422
+ "grad_norm": 0.8050324320793152,
423
+ "learning_rate": 1.3606557377049181e-05,
424
+ "loss": 0.0826,
425
+ "step": 3500
426
+ },
427
+ {
428
+ "epoch": 6.0,
429
+ "grad_norm": 0.9876068830490112,
430
+ "learning_rate": 1.3060109289617487e-05,
431
+ "loss": 0.0811,
432
+ "step": 3600
433
+ },
434
+ {
435
+ "epoch": 6.0,
436
+ "eval_administration_accuracy": 0.9434021263289556,
437
+ "eval_administration_f1": 0.9387369505688391,
438
+ "eval_corruption_accuracy": 0.962997706900146,
439
+ "eval_corruption_f1": 0.9612502739505763,
440
+ "eval_democracy_accuracy": 0.9562226391494684,
441
+ "eval_democracy_f1": 0.954550595976141,
442
+ "eval_development_accuracy": 0.9203668959766521,
443
+ "eval_development_f1": 0.9144818227568411,
444
+ "eval_economy_accuracy": 0.9361058995205337,
445
+ "eval_economy_f1": 0.9313226219336369,
446
+ "eval_education_accuracy": 0.9694600792161768,
447
+ "eval_education_f1": 0.9675599527596068,
448
+ "eval_environment_accuracy": 0.9811340421096518,
449
+ "eval_environment_f1": 0.9803282887818994,
450
+ "eval_instability_accuracy": 0.9459036898061288,
451
+ "eval_instability_f1": 0.942472499866658,
452
+ "eval_leadership_accuracy": 0.8430268918073797,
453
+ "eval_leadership_f1": 0.8387770890259914,
454
+ "eval_loss": 0.20078489184379578,
455
+ "eval_overall_accuracy": 0.941647557501216,
456
+ "eval_overall_f1": 0.938780687918615,
457
+ "eval_race_accuracy": 0.9580988117573483,
458
+ "eval_race_f1": 0.9562411064132649,
459
+ "eval_religion_accuracy": 0.961955388784657,
460
+ "eval_religion_f1": 0.960819752774671,
461
+ "eval_runtime": 11.0615,
462
+ "eval_safety_accuracy": 0.9210965186574943,
463
+ "eval_safety_f1": 0.9188273002152547,
464
+ "eval_samples_per_second": 867.33,
465
+ "eval_steps_per_second": 54.242,
466
+ "step": 3600
467
+ },
468
+ {
469
+ "epoch": 6.166736140058358,
470
+ "grad_norm": 0.8045951724052429,
471
+ "learning_rate": 1.2513661202185792e-05,
472
+ "loss": 0.0701,
473
+ "step": 3700
474
+ },
475
+ {
476
+ "epoch": 6.333472280116715,
477
+ "grad_norm": 0.6562448143959045,
478
+ "learning_rate": 1.19672131147541e-05,
479
+ "loss": 0.0726,
480
+ "step": 3800
481
+ },
482
+ {
483
+ "epoch": 6.500208420175073,
484
+ "grad_norm": 0.7833376526832581,
485
+ "learning_rate": 1.1420765027322405e-05,
486
+ "loss": 0.0673,
487
+ "step": 3900
488
+ },
489
+ {
490
+ "epoch": 6.66694456023343,
491
+ "grad_norm": 0.8071606755256653,
492
+ "learning_rate": 1.087431693989071e-05,
493
+ "loss": 0.0733,
494
+ "step": 4000
495
+ },
496
+ {
497
+ "epoch": 6.833680700291788,
498
+ "grad_norm": 0.7821515798568726,
499
+ "learning_rate": 1.0327868852459017e-05,
500
+ "loss": 0.0678,
501
+ "step": 4100
502
+ },
503
+ {
504
+ "epoch": 7.0,
505
+ "grad_norm": 0.6441165804862976,
506
+ "learning_rate": 9.781420765027323e-06,
507
+ "loss": 0.0701,
508
+ "step": 4200
509
+ },
510
+ {
511
+ "epoch": 7.0,
512
+ "eval_administration_accuracy": 0.9396497811131957,
513
+ "eval_administration_f1": 0.9371884848945066,
514
+ "eval_corruption_accuracy": 0.9616426933500104,
515
+ "eval_corruption_f1": 0.9603149034436674,
516
+ "eval_democracy_accuracy": 0.9548676255993329,
517
+ "eval_democracy_f1": 0.9544642189136688,
518
+ "eval_development_accuracy": 0.919533041484261,
519
+ "eval_development_f1": 0.9174833579112073,
520
+ "eval_economy_accuracy": 0.9370439858244737,
521
+ "eval_economy_f1": 0.9350264441971636,
522
+ "eval_education_accuracy": 0.9670627475505524,
523
+ "eval_education_f1": 0.9657003505198045,
524
+ "eval_environment_accuracy": 0.9821763602251408,
525
+ "eval_environment_f1": 0.9808859370812918,
526
+ "eval_instability_accuracy": 0.9445486762559934,
527
+ "eval_instability_f1": 0.9432538618242539,
528
+ "eval_leadership_accuracy": 0.8441734417344173,
529
+ "eval_leadership_f1": 0.8426503273307849,
530
+ "eval_loss": 0.21108125150203705,
531
+ "eval_overall_accuracy": 0.9408397609617123,
532
+ "eval_overall_f1": 0.9393817348962156,
533
+ "eval_race_accuracy": 0.9566395663956639,
534
+ "eval_race_f1": 0.9559208826126899,
535
+ "eval_religion_accuracy": 0.9609130706691682,
536
+ "eval_religion_f1": 0.9608002073614549,
537
+ "eval_runtime": 10.9796,
538
+ "eval_safety_accuracy": 0.9218261413383365,
539
+ "eval_safety_f1": 0.9188918426640929,
540
+ "eval_samples_per_second": 873.8,
541
+ "eval_steps_per_second": 54.647,
542
+ "step": 4200
543
+ },
544
+ {
545
+ "epoch": 7.166736140058358,
546
+ "grad_norm": 0.6236246228218079,
547
+ "learning_rate": 9.234972677595628e-06,
548
+ "loss": 0.0597,
549
+ "step": 4300
550
+ },
551
+ {
552
+ "epoch": 7.333472280116715,
553
+ "grad_norm": 0.6468791365623474,
554
+ "learning_rate": 8.688524590163935e-06,
555
+ "loss": 0.0568,
556
+ "step": 4400
557
+ },
558
+ {
559
+ "epoch": 7.500208420175073,
560
+ "grad_norm": 0.7850766777992249,
561
+ "learning_rate": 8.14207650273224e-06,
562
+ "loss": 0.0586,
563
+ "step": 4500
564
+ },
565
+ {
566
+ "epoch": 7.66694456023343,
567
+ "grad_norm": 0.9113643169403076,
568
+ "learning_rate": 7.595628415300546e-06,
569
+ "loss": 0.0611,
570
+ "step": 4600
571
+ },
572
+ {
573
+ "epoch": 7.833680700291788,
574
+ "grad_norm": 0.9405547976493835,
575
+ "learning_rate": 7.049180327868853e-06,
576
+ "loss": 0.0636,
577
+ "step": 4700
578
+ },
579
+ {
580
+ "epoch": 8.0,
581
+ "grad_norm": 0.5414742827415466,
582
+ "learning_rate": 6.502732240437159e-06,
583
+ "loss": 0.0588,
584
+ "step": 4800
585
+ },
586
+ {
587
+ "epoch": 8.0,
588
+ "eval_administration_accuracy": 0.9413174900979779,
589
+ "eval_administration_f1": 0.9368824702703686,
590
+ "eval_corruption_accuracy": 0.9614342297269126,
591
+ "eval_corruption_f1": 0.9602768664022149,
592
+ "eval_democracy_accuracy": 0.9560141755263707,
593
+ "eval_democracy_f1": 0.9543585350199038,
594
+ "eval_development_accuracy": 0.9192203460496143,
595
+ "eval_development_f1": 0.9152605716450639,
596
+ "eval_economy_accuracy": 0.9381905357515113,
597
+ "eval_economy_f1": 0.935106524044037,
598
+ "eval_education_accuracy": 0.9688346883468835,
599
+ "eval_education_f1": 0.9673054511905975,
600
+ "eval_environment_accuracy": 0.9822805920366896,
601
+ "eval_environment_f1": 0.980929327817527,
602
+ "eval_instability_accuracy": 0.9455909943714822,
603
+ "eval_instability_f1": 0.9437005892876725,
604
+ "eval_leadership_accuracy": 0.8438607462997707,
605
+ "eval_leadership_f1": 0.8410810165791273,
606
+ "eval_loss": 0.215087428689003,
607
+ "eval_overall_accuracy": 0.9414651518310054,
608
+ "eval_overall_f1": 0.93924957060805,
609
+ "eval_race_accuracy": 0.9572649572649573,
610
+ "eval_race_f1": 0.9565627258185468,
611
+ "eval_religion_accuracy": 0.9615384615384616,
612
+ "eval_religion_f1": 0.9610239728595763,
613
+ "eval_runtime": 10.9675,
614
+ "eval_safety_accuracy": 0.9220346049614342,
615
+ "eval_safety_f1": 0.9185067963619639,
616
+ "eval_samples_per_second": 874.763,
617
+ "eval_steps_per_second": 54.707,
618
+ "step": 4800
619
+ },
620
+ {
621
+ "epoch": 8.166736140058358,
622
+ "grad_norm": 0.5899946093559265,
623
+ "learning_rate": 5.956284153005465e-06,
624
+ "loss": 0.0543,
625
+ "step": 4900
626
+ },
627
+ {
628
+ "epoch": 8.333472280116716,
629
+ "grad_norm": 0.5821051597595215,
630
+ "learning_rate": 5.409836065573771e-06,
631
+ "loss": 0.0528,
632
+ "step": 5000
633
+ },
634
+ {
635
+ "epoch": 8.500208420175072,
636
+ "grad_norm": 0.7321776151657104,
637
+ "learning_rate": 4.863387978142076e-06,
638
+ "loss": 0.0532,
639
+ "step": 5100
640
+ },
641
+ {
642
+ "epoch": 8.66694456023343,
643
+ "grad_norm": 0.7479367256164551,
644
+ "learning_rate": 4.316939890710383e-06,
645
+ "loss": 0.0519,
646
+ "step": 5200
647
+ },
648
+ {
649
+ "epoch": 8.833680700291788,
650
+ "grad_norm": 0.8334552049636841,
651
+ "learning_rate": 3.770491803278689e-06,
652
+ "loss": 0.0505,
653
+ "step": 5300
654
+ },
655
+ {
656
+ "epoch": 9.0,
657
+ "grad_norm": 0.7611594200134277,
658
+ "learning_rate": 3.2240437158469947e-06,
659
+ "loss": 0.0528,
660
+ "step": 5400
661
+ },
662
+ {
663
+ "epoch": 9.0,
664
+ "eval_administration_accuracy": 0.9398582447362935,
665
+ "eval_administration_f1": 0.9370835475227389,
666
+ "eval_corruption_accuracy": 0.9618511569731082,
667
+ "eval_corruption_f1": 0.9605413145422365,
668
+ "eval_democracy_accuracy": 0.9559099437148217,
669
+ "eval_democracy_f1": 0.9549404791061389,
670
+ "eval_development_accuracy": 0.9212007504690432,
671
+ "eval_development_f1": 0.9170777974760537,
672
+ "eval_economy_accuracy": 0.9381905357515113,
673
+ "eval_economy_f1": 0.93592555672295,
674
+ "eval_education_accuracy": 0.9682092974775902,
675
+ "eval_education_f1": 0.9667368725950756,
676
+ "eval_environment_accuracy": 0.9813425057327496,
677
+ "eval_environment_f1": 0.9802272421649649,
678
+ "eval_instability_accuracy": 0.9472587033562644,
679
+ "eval_instability_f1": 0.9448551688210497,
680
+ "eval_leadership_accuracy": 0.8438607462997707,
681
+ "eval_leadership_f1": 0.8412354371277554,
682
+ "eval_loss": 0.21861030161380768,
683
+ "eval_overall_accuracy": 0.9413609200194567,
684
+ "eval_overall_f1": 0.9394140341920821,
685
+ "eval_race_accuracy": 0.9572649572649573,
686
+ "eval_race_f1": 0.9563061857501846,
687
+ "eval_religion_accuracy": 0.961955388784657,
688
+ "eval_religion_f1": 0.9612659459831671,
689
+ "eval_runtime": 11.1696,
690
+ "eval_safety_accuracy": 0.9194288096727121,
691
+ "eval_safety_f1": 0.9167728624926694,
692
+ "eval_samples_per_second": 858.939,
693
+ "eval_steps_per_second": 53.717,
694
+ "step": 5400
695
+ },
696
+ {
697
+ "epoch": 9.166736140058358,
698
+ "grad_norm": 0.5734995007514954,
699
+ "learning_rate": 2.6775956284153005e-06,
700
+ "loss": 0.0483,
701
+ "step": 5500
702
+ },
703
+ {
704
+ "epoch": 9.333472280116716,
705
+ "grad_norm": 0.4083118736743927,
706
+ "learning_rate": 2.1311475409836063e-06,
707
+ "loss": 0.0491,
708
+ "step": 5600
709
+ },
710
+ {
711
+ "epoch": 9.500208420175072,
712
+ "grad_norm": 0.6585514545440674,
713
+ "learning_rate": 1.5846994535519126e-06,
714
+ "loss": 0.0488,
715
+ "step": 5700
716
+ },
717
+ {
718
+ "epoch": 9.66694456023343,
719
+ "grad_norm": 0.8815754055976868,
720
+ "learning_rate": 1.0382513661202186e-06,
721
+ "loss": 0.0485,
722
+ "step": 5800
723
+ },
724
+ {
725
+ "epoch": 9.833680700291788,
726
+ "grad_norm": 0.7910645008087158,
727
+ "learning_rate": 4.918032786885246e-07,
728
+ "loss": 0.0475,
729
+ "step": 5900
730
+ },
731
+ {
732
+ "epoch": 9.98374322634431,
733
+ "eval_administration_accuracy": 0.9398582447362935,
734
+ "eval_administration_f1": 0.9361942162273019,
735
+ "eval_corruption_accuracy": 0.9618511569731082,
736
+ "eval_corruption_f1": 0.9605486831469265,
737
+ "eval_democracy_accuracy": 0.9554930164686263,
738
+ "eval_democracy_f1": 0.9543821493687584,
739
+ "eval_development_accuracy": 0.9214092140921409,
740
+ "eval_development_f1": 0.9172870275429638,
741
+ "eval_economy_accuracy": 0.9368355222013759,
742
+ "eval_economy_f1": 0.9345244446666318,
743
+ "eval_education_accuracy": 0.9690431519699813,
744
+ "eval_education_f1": 0.9674369063891264,
745
+ "eval_environment_accuracy": 0.9807171148634564,
746
+ "eval_environment_f1": 0.9797111999641163,
747
+ "eval_instability_accuracy": 0.9462163852407754,
748
+ "eval_instability_f1": 0.9447235769030393,
749
+ "eval_leadership_accuracy": 0.8455284552845529,
750
+ "eval_leadership_f1": 0.8428711309576332,
751
+ "eval_loss": 0.22077877819538116,
752
+ "eval_overall_accuracy": 0.9413956639566394,
753
+ "eval_overall_f1": 0.9394484894664766,
754
+ "eval_race_accuracy": 0.9575776526996039,
755
+ "eval_race_f1": 0.9567385833513858,
756
+ "eval_religion_accuracy": 0.962059620596206,
757
+ "eval_religion_f1": 0.9614110923085984,
758
+ "eval_runtime": 11.0829,
759
+ "eval_safety_accuracy": 0.9201584323535543,
760
+ "eval_safety_f1": 0.9175528627712379,
761
+ "eval_samples_per_second": 865.657,
762
+ "eval_steps_per_second": 54.137,
763
+ "step": 5990
764
+ }
765
+ ],
766
+ "logging_steps": 100,
767
+ "max_steps": 5990,
768
+ "num_input_tokens_seen": 0,
769
+ "num_train_epochs": 10,
770
+ "save_steps": 500,
771
+ "stateful_callbacks": {
772
+ "EarlyStoppingCallback": {
773
+ "args": {
774
+ "early_stopping_patience": 2,
775
+ "early_stopping_threshold": 0.0
776
+ },
777
+ "attributes": {
778
+ "early_stopping_patience_counter": 0
779
+ }
780
+ },
781
+ "TrainerControl": {
782
+ "args": {
783
+ "should_epoch_stop": false,
784
+ "should_evaluate": false,
785
+ "should_log": false,
786
+ "should_save": true,
787
+ "should_training_stop": true
788
+ },
789
+ "attributes": {}
790
+ }
791
+ },
792
+ "total_flos": 9.932543652062822e+16,
793
+ "train_batch_size": 16,
794
+ "trial_name": null,
795
+ "trial_params": {
796
+ "gradient_accumulation_steps": 4,
797
+ "learning_rate": 3e-05,
798
+ "num_train_epochs": 10
799
+ }
800
+ }
run-0/checkpoint-5990/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:82c268ed019d067f52ee3b6ec06e4f32fa79f82839c235ad890040984c5ac1cf
3
+ size 5368
run-1/checkpoint-2399/config.json ADDED
@@ -0,0 +1,125 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "BertForSequenceClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.1,
6
+ "classifier_dropout": null,
7
+ "gradient_checkpointing": false,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.1,
10
+ "hidden_size": 768,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47"
60
+ },
61
+ "initializer_range": 0.02,
62
+ "intermediate_size": 3072,
63
+ "label2id": {
64
+ "LABEL_0": 0,
65
+ "LABEL_1": 1,
66
+ "LABEL_10": 10,
67
+ "LABEL_11": 11,
68
+ "LABEL_12": 12,
69
+ "LABEL_13": 13,
70
+ "LABEL_14": 14,
71
+ "LABEL_15": 15,
72
+ "LABEL_16": 16,
73
+ "LABEL_17": 17,
74
+ "LABEL_18": 18,
75
+ "LABEL_19": 19,
76
+ "LABEL_2": 2,
77
+ "LABEL_20": 20,
78
+ "LABEL_21": 21,
79
+ "LABEL_22": 22,
80
+ "LABEL_23": 23,
81
+ "LABEL_24": 24,
82
+ "LABEL_25": 25,
83
+ "LABEL_26": 26,
84
+ "LABEL_27": 27,
85
+ "LABEL_28": 28,
86
+ "LABEL_29": 29,
87
+ "LABEL_3": 3,
88
+ "LABEL_30": 30,
89
+ "LABEL_31": 31,
90
+ "LABEL_32": 32,
91
+ "LABEL_33": 33,
92
+ "LABEL_34": 34,
93
+ "LABEL_35": 35,
94
+ "LABEL_36": 36,
95
+ "LABEL_37": 37,
96
+ "LABEL_38": 38,
97
+ "LABEL_39": 39,
98
+ "LABEL_4": 4,
99
+ "LABEL_40": 40,
100
+ "LABEL_41": 41,
101
+ "LABEL_42": 42,
102
+ "LABEL_43": 43,
103
+ "LABEL_44": 44,
104
+ "LABEL_45": 45,
105
+ "LABEL_46": 46,
106
+ "LABEL_47": 47,
107
+ "LABEL_5": 5,
108
+ "LABEL_6": 6,
109
+ "LABEL_7": 7,
110
+ "LABEL_8": 8,
111
+ "LABEL_9": 9
112
+ },
113
+ "layer_norm_eps": 1e-12,
114
+ "max_position_embeddings": 512,
115
+ "model_type": "bert",
116
+ "num_attention_heads": 12,
117
+ "num_hidden_layers": 12,
118
+ "pad_token_id": 0,
119
+ "position_embedding_type": "absolute",
120
+ "torch_dtype": "float32",
121
+ "transformers_version": "4.50.3",
122
+ "type_vocab_size": 2,
123
+ "use_cache": true,
124
+ "vocab_size": 30522
125
+ }
run-1/checkpoint-2399/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc8d6559adcf9e224d38864a1b5e5a1403e35e26169912ed2c48d05a8c54ac1d
3
+ size 438100144
run-1/checkpoint-2399/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:48c69797f393304be39d4ed18ae9f8e30e427184310b41dcdf2d28ed4f7a5d08
3
+ size 876321402
run-1/checkpoint-2399/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:03f6521411fa3fb31926db64cfd06b37e9d7c0607f2eec0bf790c73df4910854
3
+ size 14244
run-1/checkpoint-2399/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:64fde29b279e6b3aa49d19d3ea139f3ca54384174f4fc03279fae226afc19459
3
+ size 988
run-1/checkpoint-2399/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e731f9ba3545902d324588c71409a08726e761a3c726411b8525d42a2e62e11
3
+ size 1064
run-1/checkpoint-2399/trainer_state.json ADDED
@@ -0,0 +1,242 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 2399,
3
+ "best_metric": 0.9244250028140656,
4
+ "best_model_checkpoint": "./results/run-1/checkpoint-2399",
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 2399,
8
+ "is_hyper_param_search": true,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.041684035014589414,
14
+ "grad_norm": 2.362386465072632,
15
+ "learning_rate": 9.900000000000002e-06,
16
+ "loss": 1.3944,
17
+ "step": 100
18
+ },
19
+ {
20
+ "epoch": 0.08336807002917883,
21
+ "grad_norm": 0.7483571767807007,
22
+ "learning_rate": 1.9900000000000003e-05,
23
+ "loss": 0.5576,
24
+ "step": 200
25
+ },
26
+ {
27
+ "epoch": 0.12505210504376824,
28
+ "grad_norm": 0.8563070297241211,
29
+ "learning_rate": 2.9900000000000002e-05,
30
+ "loss": 0.4502,
31
+ "step": 300
32
+ },
33
+ {
34
+ "epoch": 0.16673614005835766,
35
+ "grad_norm": 0.7200814485549927,
36
+ "learning_rate": 3.99e-05,
37
+ "loss": 0.388,
38
+ "step": 400
39
+ },
40
+ {
41
+ "epoch": 0.20842017507294705,
42
+ "grad_norm": 1.4553083181381226,
43
+ "learning_rate": 4.99e-05,
44
+ "loss": 0.3508,
45
+ "step": 500
46
+ },
47
+ {
48
+ "epoch": 0.25010421008753647,
49
+ "grad_norm": 0.9915399551391602,
50
+ "learning_rate": 4.978927203065135e-05,
51
+ "loss": 0.3186,
52
+ "step": 600
53
+ },
54
+ {
55
+ "epoch": 0.29178824510212586,
56
+ "grad_norm": 0.9978141188621521,
57
+ "learning_rate": 4.957641549595573e-05,
58
+ "loss": 0.3087,
59
+ "step": 700
60
+ },
61
+ {
62
+ "epoch": 0.3334722801167153,
63
+ "grad_norm": 1.3709461688995361,
64
+ "learning_rate": 4.936355896126011e-05,
65
+ "loss": 0.2976,
66
+ "step": 800
67
+ },
68
+ {
69
+ "epoch": 0.3751563151313047,
70
+ "grad_norm": 1.82745361328125,
71
+ "learning_rate": 4.91507024265645e-05,
72
+ "loss": 0.28,
73
+ "step": 900
74
+ },
75
+ {
76
+ "epoch": 0.4168403501458941,
77
+ "grad_norm": 1.2072925567626953,
78
+ "learning_rate": 4.893784589186888e-05,
79
+ "loss": 0.2751,
80
+ "step": 1000
81
+ },
82
+ {
83
+ "epoch": 0.45852438516048355,
84
+ "grad_norm": 1.0433937311172485,
85
+ "learning_rate": 4.872498935717327e-05,
86
+ "loss": 0.2712,
87
+ "step": 1100
88
+ },
89
+ {
90
+ "epoch": 0.5002084201750729,
91
+ "grad_norm": 1.1097549200057983,
92
+ "learning_rate": 4.851213282247765e-05,
93
+ "loss": 0.2909,
94
+ "step": 1200
95
+ },
96
+ {
97
+ "epoch": 0.5418924551896623,
98
+ "grad_norm": 1.2917771339416504,
99
+ "learning_rate": 4.8299276287782034e-05,
100
+ "loss": 0.2501,
101
+ "step": 1300
102
+ },
103
+ {
104
+ "epoch": 0.5835764902042517,
105
+ "grad_norm": 0.9812286496162415,
106
+ "learning_rate": 4.808641975308642e-05,
107
+ "loss": 0.2621,
108
+ "step": 1400
109
+ },
110
+ {
111
+ "epoch": 0.6252605252188412,
112
+ "grad_norm": 1.1786612272262573,
113
+ "learning_rate": 4.7873563218390804e-05,
114
+ "loss": 0.2435,
115
+ "step": 1500
116
+ },
117
+ {
118
+ "epoch": 0.6669445602334306,
119
+ "grad_norm": 1.4759145975112915,
120
+ "learning_rate": 4.766070668369519e-05,
121
+ "loss": 0.2398,
122
+ "step": 1600
123
+ },
124
+ {
125
+ "epoch": 0.70862859524802,
126
+ "grad_norm": 1.5548242330551147,
127
+ "learning_rate": 4.744785014899958e-05,
128
+ "loss": 0.2432,
129
+ "step": 1700
130
+ },
131
+ {
132
+ "epoch": 0.7503126302626094,
133
+ "grad_norm": 1.9005539417266846,
134
+ "learning_rate": 4.723499361430396e-05,
135
+ "loss": 0.2351,
136
+ "step": 1800
137
+ },
138
+ {
139
+ "epoch": 0.7919966652771988,
140
+ "grad_norm": 1.5831588506698608,
141
+ "learning_rate": 4.702213707960835e-05,
142
+ "loss": 0.2378,
143
+ "step": 1900
144
+ },
145
+ {
146
+ "epoch": 0.8336807002917882,
147
+ "grad_norm": 1.5482088327407837,
148
+ "learning_rate": 4.680928054491273e-05,
149
+ "loss": 0.2271,
150
+ "step": 2000
151
+ },
152
+ {
153
+ "epoch": 0.8753647353063777,
154
+ "grad_norm": 1.1437913179397583,
155
+ "learning_rate": 4.6596424010217114e-05,
156
+ "loss": 0.231,
157
+ "step": 2100
158
+ },
159
+ {
160
+ "epoch": 0.9170487703209671,
161
+ "grad_norm": 0.7985406517982483,
162
+ "learning_rate": 4.63835674755215e-05,
163
+ "loss": 0.228,
164
+ "step": 2200
165
+ },
166
+ {
167
+ "epoch": 0.9587328053355565,
168
+ "grad_norm": 1.9323142766952515,
169
+ "learning_rate": 4.6170710940825884e-05,
170
+ "loss": 0.2198,
171
+ "step": 2300
172
+ },
173
+ {
174
+ "epoch": 1.0,
175
+ "eval_administration_accuracy": 0.9322493224932249,
176
+ "eval_administration_f1": 0.9122911143651031,
177
+ "eval_corruption_accuracy": 0.9560141755263707,
178
+ "eval_corruption_f1": 0.9517304794081561,
179
+ "eval_democracy_accuracy": 0.9505941213258287,
180
+ "eval_democracy_f1": 0.9401838742047087,
181
+ "eval_development_accuracy": 0.9118198874296435,
182
+ "eval_development_f1": 0.9003553411809495,
183
+ "eval_economy_accuracy": 0.9264123410464874,
184
+ "eval_economy_f1": 0.9202465382357672,
185
+ "eval_education_accuracy": 0.9643527204502814,
186
+ "eval_education_f1": 0.9615509617901187,
187
+ "eval_environment_accuracy": 0.9799874921826142,
188
+ "eval_environment_f1": 0.9786142985651582,
189
+ "eval_instability_accuracy": 0.9417344173441734,
190
+ "eval_instability_f1": 0.9340921018628459,
191
+ "eval_leadership_accuracy": 0.8085261621846987,
192
+ "eval_leadership_f1": 0.7858417402549625,
193
+ "eval_loss": 0.21971678733825684,
194
+ "eval_overall_accuracy": 0.9324317281634354,
195
+ "eval_overall_f1": 0.9244250028140656,
196
+ "eval_race_accuracy": 0.9508025849489264,
197
+ "eval_race_f1": 0.9481714649378918,
198
+ "eval_religion_accuracy": 0.9530956848030019,
199
+ "eval_religion_f1": 0.9518106778632144,
200
+ "eval_runtime": 10.9084,
201
+ "eval_safety_accuracy": 0.9135918282259746,
202
+ "eval_safety_f1": 0.9082114410999095,
203
+ "eval_samples_per_second": 879.505,
204
+ "eval_steps_per_second": 55.003,
205
+ "step": 2399
206
+ }
207
+ ],
208
+ "logging_steps": 100,
209
+ "max_steps": 23990,
210
+ "num_input_tokens_seen": 0,
211
+ "num_train_epochs": 10,
212
+ "save_steps": 500,
213
+ "stateful_callbacks": {
214
+ "EarlyStoppingCallback": {
215
+ "args": {
216
+ "early_stopping_patience": 2,
217
+ "early_stopping_threshold": 0.0
218
+ },
219
+ "attributes": {
220
+ "early_stopping_patience_counter": 0
221
+ }
222
+ },
223
+ "TrainerControl": {
224
+ "args": {
225
+ "should_epoch_stop": false,
226
+ "should_evaluate": false,
227
+ "should_log": false,
228
+ "should_save": true,
229
+ "should_training_stop": false
230
+ },
231
+ "attributes": {}
232
+ }
233
+ },
234
+ "total_flos": 9686485853798400.0,
235
+ "train_batch_size": 16,
236
+ "trial_name": null,
237
+ "trial_params": {
238
+ "gradient_accumulation_steps": 1,
239
+ "learning_rate": 5e-05,
240
+ "num_train_epochs": 10
241
+ }
242
+ }
run-1/checkpoint-2399/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4312a997194e916559618a59f3bbf4e1f98c127e89ae3b7fb5162b782863b12c
3
+ size 5368
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:82c268ed019d067f52ee3b6ec06e4f32fa79f82839c235ad890040984c5ac1cf
3
  size 5368
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4312a997194e916559618a59f3bbf4e1f98c127e89ae3b7fb5162b782863b12c
3
  size 5368