Hugging Face's logo

Datasets:
Akahsizrr
/
alea-runs

Dataset card Files
xet
 Community
1
Akahsizrr commited on
Commit
535bfcd
·
verified ·
1 Parent(s): a2f2c11

alea sync

Browse files
Files changed (1) hide show
  1. runs/loop_r2.log +423 -0
runs/loop_r2.log ADDED
@@ -0,0 +1,423 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Repo card metadata block was not found. Setting CardData to empty.
2
+ shares: {'gen_stoch': 0.9, 'gen_enterprise': 1.0, 'gen_structured': 0.8, 'gen_adversarial': 0.6, 'gen_numeric': 0.7, 'gen_long': 0.5, 'base': 0.4, 'gen_kaggle_cls': 1.0, 'gen_kaggle_reg': 0.7, 'gen_hf': 1.3}
3
+ base: 152
4
+ gen_adversarial: 228
5
+ gen_enterprise: 380
6
+ gen_hf: 494
7
+ gen_kaggle_cls: 380
8
+ gen_kaggle_reg: 266
9
+ gen_long: 190
10
+ gen_numeric: 266
11
+ gen_stoch: 342
12
+ gen_structured: 304
13
+ wrote 3002
14
+ === BASELINE EVAL ===
15
+ `torch_dtype` is deprecated! Use `dtype` instead!
16
+ {
17
+ "n_choice": 967,
18
+ "acc": 0.6070320579110652,
19
+ "ece10": 0.049875218101469936,
20
+ "known_q_l2": 0.36100139945847076,
21
+ "noul_brier": 0.16841900903910512,
22
+ "regress_r2": 3.9983444396818335e-05,
23
+ "n_regress": 251
24
+ }
25
+ active sources: {'gen_stoch': 0.9, 'gen_enterprise': 1.0, 'gen_structured': 0.8, 'gen_adversarial': 0.6, 'gen_numeric': 0.7, 'gen_long': 0.5, 'base': 0.4, 'gen_kaggle_cls': 1.0, 'gen_kaggle_reg': 0.7, 'gen_hf': 1.3}
26
+
27
+ ===== ROUND 0 (init=runs/alea_v3_final.pt) =====
28
+ Repo card metadata block was not found. Setting CardData to empty.
29
+ base: 4051
30
+ gen_adversarial: 6076
31
+ gen_enterprise: 10127
32
+ gen_hf: 13165
33
+ gen_kaggle_cls: 10127
34
+ gen_kaggle_reg: 7089
35
+ gen_long: 5063
36
+ gen_numeric: 7089
37
+ gen_stoch: 9114
38
+ gen_structured: 8101
39
+ corpus: 80002 rows -> data/loop_train_r0.jsonl
40
+ + /usr/bin/python3 -m alea.train --train data/loop_train_r0.jsonl --dev data/test_r2.jsonl --out runs/loop_r0 --init-from runs/alea_v3_final.pt --train-mode full --epochs 1.0 --batch 32 --accum 1 --lr-head 0.0001 --lr-backbone 1e-05 --attn sdpa --save-every 100000 --max-len 4096
41
+ `torch_dtype` is deprecated! Use `dtype` instead!
42
+ resumed from runs/alea_v3_final.pt
43
+ pre-packing dataset...
44
+ packed 80002 (skipped 0) trainable=632.6M
45
+ `use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.
46
+ step 50/2647 loss=21.4867 1.28s/step peak=7.3GB
47
+ step 100/2647 loss=23.3560 1.08s/step peak=7.3GB
48
+ step 150/2647 loss=4.3046 0.98s/step peak=7.3GB
49
+ step 200/2647 loss=1.3038 1.09s/step peak=7.4GB
50
+ step 250/2647 loss=6.3390 1.13s/step peak=7.4GB
51
+ step 300/2647 loss=0.6348 1.14s/step peak=7.4GB
52
+ step 350/2647 loss=0.2200 1.05s/step peak=7.4GB
53
+ step 400/2647 loss=3.1978 1.09s/step peak=7.4GB
54
+ step 450/2647 loss=1.9922 1.13s/step peak=7.4GB
55
+ step 500/2647 loss=0.9770 1.12s/step peak=7.4GB
56
+ step 550/2647 loss=1.5852 1.10s/step peak=7.4GB
57
+ step 600/2647 loss=2.2041 1.13s/step peak=7.4GB
58
+ step 650/2647 loss=1.9030 1.15s/step peak=7.4GB
59
+ step 700/2647 loss=1.5189 1.16s/step peak=7.4GB
60
+ step 750/2647 loss=1.7761 1.16s/step peak=7.4GB
61
+ step 800/2647 loss=0.3994 1.16s/step peak=7.4GB
62
+ step 850/2647 loss=1.0297 1.18s/step peak=7.4GB
63
+ step 900/2647 loss=1.5730 1.17s/step peak=7.4GB
64
+ step 950/2647 loss=0.5846 1.16s/step peak=7.4GB
65
+ step 1000/2647 loss=0.8669 1.17s/step peak=7.4GB
66
+ step 1050/2647 loss=1.6609 1.15s/step peak=7.4GB
67
+ step 1100/2647 loss=0.7196 1.14s/step peak=7.4GB
68
+ step 1150/2647 loss=0.4430 1.14s/step peak=7.4GB
69
+ step 1200/2647 loss=2.4744 1.15s/step peak=7.4GB
70
+ step 1250/2647 loss=0.6147 1.15s/step peak=7.4GB
71
+ step 1300/2647 loss=0.4429 1.15s/step peak=7.4GB
72
+ step 1350/2647 loss=0.5603 1.14s/step peak=7.4GB
73
+ step 1400/2647 loss=0.2032 1.12s/step peak=7.4GB
74
+ step 1450/2647 loss=0.8660 1.12s/step peak=7.4GB
75
+ step 1500/2647 loss=0.6613 1.13s/step peak=7.4GB
76
+ step 1550/2647 loss=0.8071 1.13s/step peak=7.4GB
77
+ step 1600/2647 loss=0.3520 1.13s/step peak=7.4GB
78
+ step 1650/2647 loss=0.2462 1.12s/step peak=7.4GB
79
+ step 1700/2647 loss=0.4969 1.12s/step peak=7.4GB
80
+ step 1750/2647 loss=0.4343 1.12s/step peak=7.4GB
81
+ step 1800/2647 loss=1.6418 1.12s/step peak=7.4GB
82
+ step 1850/2647 loss=0.3946 1.12s/step peak=7.4GB
83
+ step 1900/2647 loss=0.3291 1.12s/step peak=7.4GB
84
+ step 1950/2647 loss=0.2169 1.10s/step peak=7.4GB
85
+ step 2000/2647 loss=0.5659 1.11s/step peak=7.4GB
86
+ step 2050/2647 loss=0.3423 1.12s/step peak=7.4GB
87
+ step 2100/2647 loss=0.2803 1.12s/step peak=7.4GB
88
+ step 2150/2647 loss=0.2973 1.11s/step peak=7.4GB
89
+ step 2200/2647 loss=0.3349 1.12s/step peak=7.4GB
90
+ step 2250/2647 loss=0.2262 1.12s/step peak=7.4GB
91
+ step 2300/2647 loss=0.3057 1.13s/step peak=7.4GB
92
+ step 2350/2647 loss=0.2993 1.13s/step peak=7.4GB
93
+ step 2400/2647 loss=0.2185 1.12s/step peak=7.4GB
94
+ step 2450/2647 loss=0.2952 1.12s/step peak=7.4GB
95
+ step 2500/2647 loss=0.3061 1.12s/step peak=7.4GB
96
+ step 2550/2647 loss=0.3593 1.13s/step peak=7.4GB
97
+ step 2600/2647 loss=0.2195 1.12s/step peak=7.4GB
98
+ saved runs/loop_r0/step2647.pt
99
+ done
100
+ + /usr/bin/python3 -m alea.eval --ckpt runs/loop_r0/final.pt --data data/test_r2.jsonl --max-items 1200 --skip-structural
101
+ ROUND 0 METRICS {"n_choice": 967, "acc": 0.6938986556359876, "ece10": 0.027870251981092692, "known_q_l2": 0.27051023337388863, "noul_brier": 0.09407291528631966, "regress_r2": 4.96643331993063e-05, "n_regress": 251}
102
+
103
+ ===== ROUND 1 (init=runs/loop_r0/final.pt) =====
104
+ base: 4051
105
+ gen_adversarial: 6076
106
+ gen_enterprise: 10127
107
+ gen_hf: 13165
108
+ gen_kaggle_cls: 10127
109
+ gen_kaggle_reg: 7089
110
+ gen_long: 5063
111
+ gen_numeric: 7089
112
+ gen_stoch: 9114
113
+ gen_structured: 8101
114
+ corpus: 80002 rows -> data/loop_train_r1.jsonl
115
+ + /usr/bin/python3 -m alea.train --train data/loop_train_r1.jsonl --dev data/test_r2.jsonl --out runs/loop_r1 --init-from runs/loop_r0/final.pt --train-mode full --epochs 1.0 --batch 32 --accum 1 --lr-head 0.0001 --lr-backbone 1e-05 --attn sdpa --save-every 100000 --max-len 4096
116
+ `torch_dtype` is deprecated! Use `dtype` instead!
117
+ resumed from runs/loop_r0/final.pt
118
+ pre-packing dataset...
119
+ packed 80002 (skipped 0) trainable=632.6M
120
+ `use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.
121
+ step 50/2647 loss=0.2663 1.30s/step peak=7.3GB
122
+ step 100/2647 loss=0.7273 1.09s/step peak=7.3GB
123
+ step 150/2647 loss=0.6140 0.98s/step peak=7.3GB
124
+ step 200/2647 loss=0.5171 1.09s/step peak=7.3GB
125
+ step 250/2647 loss=0.8826 1.13s/step peak=7.3GB
126
+ step 300/2647 loss=1.3685 1.13s/step peak=7.4GB
127
+ step 350/2647 loss=0.1940 1.05s/step peak=7.4GB
128
+ step 400/2647 loss=1.4546 1.09s/step peak=7.4GB
129
+ step 450/2647 loss=1.5070 1.12s/step peak=7.4GB
130
+ step 500/2647 loss=0.4732 1.11s/step peak=7.4GB
131
+ step 550/2647 loss=0.7498 1.10s/step peak=7.4GB
132
+ step 600/2647 loss=0.8806 1.12s/step peak=7.4GB
133
+ step 650/2647 loss=1.4748 1.15s/step peak=7.4GB
134
+ step 700/2647 loss=1.3534 1.16s/step peak=7.4GB
135
+ step 750/2647 loss=0.8737 1.16s/step peak=7.4GB
136
+ step 800/2647 loss=0.4007 1.15s/step peak=7.4GB
137
+ step 850/2647 loss=0.5923 1.18s/step peak=7.4GB
138
+ step 900/2647 loss=0.4387 1.18s/step peak=7.4GB
139
+ step 950/2647 loss=0.5409 1.17s/step peak=7.4GB
140
+ step 1000/2647 loss=0.8012 1.18s/step peak=7.4GB
141
+ step 1050/2647 loss=0.8855 1.15s/step peak=7.4GB
142
+ step 1100/2647 loss=0.3865 1.14s/step peak=7.4GB
143
+ step 1150/2647 loss=0.4147 1.15s/step peak=7.4GB
144
+ step 1200/2647 loss=1.1627 1.15s/step peak=7.4GB
145
+ step 1250/2647 loss=0.4786 1.16s/step peak=7.4GB
146
+ step 1300/2647 loss=0.3288 1.16s/step peak=7.4GB
147
+ step 1350/2647 loss=0.3831 1.14s/step peak=7.4GB
148
+ step 1400/2647 loss=0.1818 1.13s/step peak=7.4GB
149
+ step 1450/2647 loss=0.7967 1.13s/step peak=7.4GB
150
+ step 1500/2647 loss=0.5727 1.13s/step peak=7.4GB
151
+ step 1550/2647 loss=0.6766 1.14s/step peak=7.4GB
152
+ step 1600/2647 loss=0.4859 1.14s/step peak=7.4GB
153
+ step 1650/2647 loss=0.3537 1.13s/step peak=7.4GB
154
+ step 1700/2647 loss=0.3501 1.12s/step peak=7.4GB
155
+ step 1750/2647 loss=0.3621 1.13s/step peak=7.4GB
156
+ step 1800/2647 loss=1.0274 1.12s/step peak=7.4GB
157
+ step 1850/2647 loss=0.3334 1.12s/step peak=7.4GB
158
+ step 1900/2647 loss=0.2964 1.12s/step peak=7.4GB
159
+ step 1950/2647 loss=0.1941 1.11s/step peak=7.4GB
160
+ step 2000/2647 loss=0.4524 1.12s/step peak=7.4GB
161
+ step 2050/2647 loss=0.2826 1.12s/step peak=7.4GB
162
+ step 2100/2647 loss=0.2631 1.12s/step peak=7.4GB
163
+ step 2150/2647 loss=0.2782 1.12s/step peak=7.4GB
164
+ step 2200/2647 loss=0.2753 1.12s/step peak=7.4GB
165
+ step 2250/2647 loss=0.2164 1.12s/step peak=7.4GB
166
+ step 2300/2647 loss=0.2716 1.14s/step peak=7.4GB
167
+ step 2350/2647 loss=0.2423 1.13s/step peak=7.4GB
168
+ step 2400/2647 loss=0.1978 1.12s/step peak=7.4GB
169
+ step 2450/2647 loss=0.2608 1.12s/step peak=7.4GB
170
+ step 2500/2647 loss=0.2652 1.13s/step peak=7.4GB
171
+ step 2550/2647 loss=0.3004 1.13s/step peak=7.4GB
172
+ step 2600/2647 loss=0.2218 1.13s/step peak=7.4GB
173
+ saved runs/loop_r1/step2647.pt
174
+ done
175
+ + /usr/bin/python3 -m alea.eval --ckpt runs/loop_r1/final.pt --data data/test_r2.jsonl --max-items 1200 --skip-structural
176
+ ROUND 1 METRICS {"n_choice": 967, "acc": 0.7073422957600828, "ece10": 0.040300781880009386, "known_q_l2": 0.2513496713112938, "noul_brier": 0.08588543889394269, "regress_r2": 0.00011140263420139629, "n_regress": 251, "regress_r2_byfam": {"big_json": -0.0123, "binomial_count": 0.981, "deterministic_total": 0.7586, "dice_sum": 0.9967, "exp_wait": 0.9984, "hf_stsb": -1.7926, "hf_yelp_reg": 0.3256, "housing_price": 0.0001, "insurance_charges": 0.0818, "kc_house_price": 0.0008, "life_expectancy": 0.9376, "long_log": 0.7897, "mixture_delay": 0.931, "normal_metric": 0.87, "triangular": 0.9722, "uniform_window": 0.9912}, "regress_r2_fam_mean": 0.4894}
177
+
178
+ ===== ROUND 2 (init=runs/loop_r1/final.pt) =====
179
+ base: 4051
180
+ gen_adversarial: 6076
181
+ gen_enterprise: 10127
182
+ gen_hf: 13165
183
+ gen_kaggle_cls: 10127
184
+ gen_kaggle_reg: 7089
185
+ gen_long: 5063
186
+ gen_numeric: 7089
187
+ gen_stoch: 9114
188
+ gen_structured: 8101
189
+ corpus: 80002 rows -> data/loop_train_r2.jsonl
190
+ + /usr/bin/python3 -m alea.train --train data/loop_train_r2.jsonl --dev data/test_r2.jsonl --out runs/loop_r2 --init-from runs/loop_r1/final.pt --train-mode full --epochs 1.0 --batch 32 --accum 1 --lr-head 0.0001 --lr-backbone 1e-05 --attn sdpa --save-every 100000 --max-len 4096
191
+ `torch_dtype` is deprecated! Use `dtype` instead!
192
+ resumed from runs/loop_r1/final.pt
193
+ pre-packing dataset...
194
+ packed 80002 (skipped 0) trainable=632.6M
195
+ `use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.
196
+ step 50/2646 loss=0.4038 1.34s/step peak=7.3GB
197
+ step 100/2646 loss=0.9341 1.31s/step peak=7.3GB
198
+ step 150/2646 loss=1.8418 1.16s/step peak=7.3GB
199
+ step 200/2646 loss=0.4021 1.32s/step peak=7.4GB
200
+ step 250/2646 loss=2.5120 1.34s/step peak=7.4GB
201
+ step 300/2646 loss=0.7406 1.35s/step peak=7.4GB
202
+ step 350/2646 loss=0.2012 1.26s/step peak=7.4GB
203
+ step 400/2646 loss=0.5700 1.27s/step peak=7.4GB
204
+ step 450/2646 loss=0.8610 1.26s/step peak=7.4GB
205
+ step 500/2646 loss=0.4161 1.25s/step peak=7.4GB
206
+ step 550/2646 loss=1.8217 1.23s/step peak=7.4GB
207
+ step 600/2646 loss=0.3533 1.24s/step peak=7.4GB
208
+ step 650/2646 loss=0.9112 1.26s/step peak=7.4GB
209
+ step 700/2646 loss=0.6751 1.25s/step peak=7.4GB
210
+ step 750/2646 loss=0.4325 1.23s/step peak=7.4GB
211
+ step 800/2646 loss=0.6604 1.22s/step peak=7.4GB
212
+ step 850/2646 loss=1.2062 1.21s/step peak=7.4GB
213
+ step 900/2646 loss=0.3963 1.20s/step peak=7.4GB
214
+ step 950/2646 loss=0.4578 1.18s/step peak=7.4GB
215
+ step 1000/2646 loss=0.5357 1.19s/step peak=7.4GB
216
+ step 1050/2646 loss=0.9767 1.17s/step peak=7.4GB
217
+ step 1100/2646 loss=1.1105 1.16s/step peak=7.4GB
218
+ step 1150/2646 loss=0.5048 1.16s/step peak=7.4GB
219
+ step 1200/2646 loss=0.4417 1.15s/step peak=7.4GB
220
+ step 1250/2646 loss=0.2097 1.15s/step peak=7.4GB
221
+ step 1300/2646 loss=0.3400 1.14s/step peak=7.4GB
222
+ step 1350/2646 loss=0.3156 1.13s/step peak=7.4GB
223
+ step 1400/2646 loss=0.3240 1.12s/step peak=7.4GB
224
+ step 1450/2646 loss=0.2628 1.12s/step peak=7.4GB
225
+ step 1500/2646 loss=0.3396 1.11s/step peak=7.4GB
226
+ step 1550/2646 loss=0.2949 1.11s/step peak=7.4GB
227
+ step 1600/2646 loss=0.2443 1.11s/step peak=7.4GB
228
+ step 1650/2646 loss=0.3016 1.11s/step peak=7.4GB
229
+ step 1700/2646 loss=0.2460 1.10s/step peak=7.4GB
230
+ step 1750/2646 loss=0.3518 1.11s/step peak=7.4GB
231
+ step 1800/2646 loss=0.2432 1.11s/step peak=7.4GB
232
+ step 1850/2646 loss=0.3738 1.12s/step peak=7.4GB
233
+ step 1900/2646 loss=0.2353 1.11s/step peak=7.4GB
234
+ step 1950/2646 loss=0.1853 1.10s/step peak=7.4GB
235
+ step 2000/2646 loss=0.3053 1.11s/step peak=7.4GB
236
+ step 2050/2646 loss=0.3384 1.11s/step peak=7.4GB
237
+ step 2100/2646 loss=0.2207 1.11s/step peak=7.4GB
238
+ step 2150/2646 loss=0.2541 1.11s/step peak=7.4GB
239
+ step 2200/2646 loss=0.2301 1.11s/step peak=7.4GB
240
+ step 2250/2646 loss=0.2889 1.12s/step peak=7.4GB
241
+ step 2300/2646 loss=0.2526 1.13s/step peak=7.4GB
242
+ step 2350/2646 loss=0.2492 1.13s/step peak=7.4GB
243
+ step 2400/2646 loss=0.1824 1.11s/step peak=7.4GB
244
+ step 2450/2646 loss=0.2311 1.11s/step peak=7.4GB
245
+ step 2500/2646 loss=0.2338 1.12s/step peak=7.4GB
246
+ step 2550/2646 loss=0.2782 1.13s/step peak=7.4GB
247
+ step 2600/2646 loss=0.2091 1.13s/step peak=7.4GB
248
+ saved runs/loop_r2/step2646.pt
249
+ done
250
+ + /usr/bin/python3 -m alea.eval --ckpt runs/loop_r2/final.pt --data data/test_r2.jsonl --max-items 1200 --skip-structural
251
+ ROUND 2 METRICS {"n_choice": 967, "acc": 0.703205791106515, "ece10": 0.038129413009244634, "known_q_l2": 0.24084140018049513, "noul_brier": 0.08056891109212592, "acc_byfam": {"approval_threshold": 0.545, "bayesian_update": 0.875, "big_json": 0.059, "binomial_tail": 1.0, "churn_renewal": 0.5, "compound": 0.875, "conditional_joint": 0.556, "content_mod": 0.857, "creditcard": 0.5, "decoy_shift": 0.913, "eligibility": 0.7, "fraud_review": 0.692, "hf_ag_news": 0.9, "hf_arc_easy": 0.125, "hf_dbpedia": 0.556, "hf_winogrande": 0.875, "hf_yelp": 0.222, "incident_report": 1.0, "insurance_charges": 0.889, "json_records": 0.406, "life_expectancy": 0.75, "log_lines": 0.818, "long_log": 0.842, "markov_chain": 0.889, "mixed_state": 1.0, "multi_choice_stoch": 0.833, "near_duplicate": 1.0, "negation": 1.0, "nested_json": 0.263, "news_cat": 0.6, "none_of_above": 1.0, "numeric_compare": 0.4, "policy_doc": 0.333, "policy_routing": 0.87, "reference_conflict": 1.0, "route": 1.0, "sla_breach": 0.567, "sms_spam": 1.0, "sycophancy_bait": 1.0, "tables": 0.571, "temporal": 0.375, "triage_score": 0.818}, "brier_byfam": {"approval_threshold": 0.0651, "bayesian_update": 0.0024, "big_json": 0.1542, "cards": 0.0022, "churn_renewal": 0.036, "content_mod": 0.0012, "creditcard": 0.0982, "deterministic_total": 0.0012, "eligibility": 0.0062, "fake_news": 0.0243, "fraud_review": 0.0152, "heart": 0.1481, "hf_amazon": 0.2253, "hf_goemo": 0.1818, "hf_squad2": 0.0854, "hf_wnli": 0.2359, "incident_report": 0.0146, "insurance_charges": 0.0841, "json_records": 0.1013, "log_lines": 0.1191, "long_log": 0.3157, "mixed_state": 0.0, "multi_probe": 0.0013, "mushroom": 0.1234, "nested_json": 0.1535, "numeric_compare": 0.1185, "order_stats": 0.0064, "policy_doc": 0.0001, "policy_routing": 0.0212, "sensor_grid": 0.0251, "sla_breach": 0.1347, "sms_spam": 0.0033, "tables": 0.0, "telco_churn": 0.1059, "temporal": 0.05, "triage_score": 0.0052, "two_draws": 0.0032}, "regress_r2": 0.00020254476069636596, "n_regress": 251, "regress_r2_byfam": {"big_json": -0.014, "binomial_count": 0.9864, "deterministic_total": 0.7938, "dice_sum": 0.9972, "exp_wait": 0.9995, "hf_stsb": -1.7131, "hf_yelp_reg": 0.4954, "housing_price": 0.0002, "insurance_charges": 0.1377, "kc_house_price": 0.0016, "life_expectancy": 0.9462, "long_log": 0.7953, "mixture_delay": 0.9461, "normal_metric": 0.878, "triangular": 0.9734, "uniform_window": 0.9927}, "regress_r2_fam_mean": 0.5135}
252
+
253
+ ===== ROUND 3 (init=runs/loop_r2/final.pt) =====
254
+ base: 4051
255
+ gen_adversarial: 6076
256
+ gen_enterprise: 10127
257
+ gen_hf: 13165
258
+ gen_kaggle_cls: 10127
259
+ gen_kaggle_reg: 7089
260
+ gen_long: 5063
261
+ gen_numeric: 7089
262
+ gen_stoch: 9114
263
+ gen_structured: 8101
264
+ corpus: 80002 rows -> data/loop_train_r3.jsonl
265
+ + /usr/bin/python3 -m alea.train --train data/loop_train_r3.jsonl --dev data/test_r2.jsonl --out runs/loop_r3 --init-from runs/loop_r2/final.pt --train-mode full --epochs 1.0 --batch 32 --accum 1 --lr-head 0.0001 --lr-backbone 1e-05 --attn sdpa --save-every 100000 --max-len 4096
266
+ `torch_dtype` is deprecated! Use `dtype` instead!
267
+ resumed from runs/loop_r2/final.pt
268
+ pre-packing dataset...
269
+ packed 80002 (skipped 0) trainable=632.6M
270
+ `use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.
271
+ step 50/2646 loss=0.3999 1.35s/step peak=7.4GB
272
+ step 100/2646 loss=0.4863 1.30s/step peak=7.4GB
273
+ step 150/2646 loss=0.4080 1.16s/step peak=7.4GB
274
+ step 200/2646 loss=0.3899 1.32s/step peak=7.4GB
275
+ step 250/2646 loss=0.6519 1.34s/step peak=7.4GB
276
+ step 300/2646 loss=0.7638 1.35s/step peak=7.4GB
277
+ step 350/2646 loss=0.1801 1.26s/step peak=7.4GB
278
+ step 400/2646 loss=1.8339 1.27s/step peak=7.4GB
279
+ step 450/2646 loss=0.9866 1.26s/step peak=7.4GB
280
+ step 500/2646 loss=0.3167 1.25s/step peak=7.4GB
281
+ step 550/2646 loss=0.5002 1.24s/step peak=7.4GB
282
+ step 600/2646 loss=0.2191 1.24s/step peak=7.4GB
283
+ step 650/2646 loss=0.9510 1.27s/step peak=7.4GB
284
+ step 700/2646 loss=0.5072 1.26s/step peak=7.4GB
285
+ step 750/2646 loss=0.2666 1.24s/step peak=7.4GB
286
+ step 800/2646 loss=0.5514 1.22s/step peak=7.4GB
287
+ step 850/2646 loss=0.8580 1.22s/step peak=7.4GB
288
+ step 900/2646 loss=0.3441 1.21s/step peak=7.4GB
289
+ step 950/2646 loss=0.3701 1.19s/step peak=7.4GB
290
+ step 1000/2646 loss=0.4070 1.19s/step peak=7.4GB
291
+ step 1050/2646 loss=0.9568 1.18s/step peak=7.4GB
292
+ step 1100/2646 loss=0.5744 1.17s/step peak=7.4GB
293
+ step 1150/2646 loss=0.3634 1.16s/step peak=7.4GB
294
+ step 1200/2646 loss=0.4955 1.15s/step peak=7.4GB
295
+ step 1250/2646 loss=0.1806 1.15s/step peak=7.4GB
296
+ step 1300/2646 loss=0.7732 1.14s/step peak=7.4GB
297
+ step 1350/2646 loss=0.2044 1.13s/step peak=7.4GB
298
+ step 1400/2646 loss=0.2174 1.12s/step peak=7.4GB
299
+ step 1450/2646 loss=0.2585 1.12s/step peak=7.4GB
300
+ step 1500/2646 loss=0.2290 1.12s/step peak=7.4GB
301
+ step 1550/2646 loss=0.2547 1.12s/step peak=7.4GB
302
+ step 1600/2646 loss=0.2023 1.12s/step peak=7.4GB
303
+ step 1650/2646 loss=0.2545 1.11s/step peak=7.4GB
304
+ step 1700/2646 loss=0.2369 1.10s/step peak=7.4GB
305
+ step 1750/2646 loss=0.3805 1.11s/step peak=7.4GB
306
+ step 1800/2646 loss=0.2543 1.11s/step peak=7.4GB
307
+ step 1850/2646 loss=0.2693 1.12s/step peak=7.4GB
308
+ step 1900/2646 loss=0.2275 1.11s/step peak=7.4GB
309
+ step 1950/2646 loss=0.1579 1.10s/step peak=7.4GB
310
+ step 2000/2646 loss=0.2128 1.11s/step peak=7.4GB
311
+ step 2050/2646 loss=0.2446 1.11s/step peak=7.4GB
312
+ step 2100/2646 loss=0.1985 1.11s/step peak=7.4GB
313
+ step 2150/2646 loss=0.2127 1.11s/step peak=7.4GB
314
+ step 2200/2646 loss=0.1846 1.11s/step peak=7.4GB
315
+ step 2250/2646 loss=0.2033 1.12s/step peak=7.4GB
316
+ step 2300/2646 loss=0.2101 1.13s/step peak=7.4GB
317
+ step 2350/2646 loss=0.2061 1.13s/step peak=7.4GB
318
+ step 2400/2646 loss=0.1622 1.11s/step peak=7.4GB
319
+ step 2450/2646 loss=0.2059 1.11s/step peak=7.4GB
320
+ step 2500/2646 loss=0.1983 1.12s/step peak=7.4GB
321
+ step 2550/2646 loss=0.2345 1.13s/step peak=7.4GB
322
+ step 2600/2646 loss=0.1778 1.13s/step peak=7.4GB
323
+ saved runs/loop_r3/step2646.pt
324
+ done
325
+ + /usr/bin/python3 -m alea.eval --ckpt runs/loop_r3/final.pt --data data/test_r2.jsonl --max-items 1200 --skip-structural
326
+ ROUND 3 METRICS {"n_choice": 967, "acc": 0.7300930713547052, "ece10": 0.037123098160451305, "known_q_l2": 0.23414903667091624, "noul_brier": 0.07470007823292435, "acc_byfam": {"approval_threshold": 0.614, "bayesian_update": 0.875, "big_json": 0.176, "binomial_tail": 1.0, "churn_renewal": 0.531, "compound": 0.875, "conditional_joint": 0.556, "content_mod": 0.857, "creditcard": 0.75, "decoy_shift": 1.0, "eligibility": 0.7, "fraud_review": 0.712, "hf_ag_news": 0.9, "hf_arc_easy": 0.25, "hf_dbpedia": 0.667, "hf_winogrande": 0.5, "hf_yelp": 0.222, "incident_report": 1.0, "insurance_charges": 0.889, "json_records": 0.406, "life_expectancy": 0.75, "log_lines": 0.841, "long_log": 0.842, "markov_chain": 1.0, "mixed_state": 1.0, "multi_choice_stoch": 0.917, "near_duplicate": 1.0, "negation": 1.0, "nested_json": 0.474, "news_cat": 0.6, "none_of_above": 1.0, "numeric_compare": 0.4, "policy_doc": 0.444, "policy_routing": 0.913, "reference_conflict": 1.0, "route": 1.0, "sla_breach": 0.6, "sms_spam": 1.0, "sycophancy_bait": 1.0, "tables": 0.571, "temporal": 0.438, "triage_score": 0.841}, "brier_byfam": {"approval_threshold": 0.0646, "bayesian_update": 0.0011, "big_json": 0.1349, "cards": 0.0006, "churn_renewal": 0.0313, "content_mod": 0.0015, "creditcard": 0.0859, "deterministic_total": 0.0105, "eligibility": 0.0022, "fake_news": 0.0069, "fraud_review": 0.014, "heart": 0.1373, "hf_amazon": 0.2239, "hf_goemo": 0.1577, "hf_squad2": 0.08, "hf_wnli": 0.2448, "incident_report": 0.0012, "insurance_charges": 0.0999, "json_records": 0.0933, "log_lines": 0.1282, "long_log": 0.3056, "mixed_state": 0.0, "multi_probe": 0.001, "mushroom": 0.08, "nested_json": 0.1547, "numeric_compare": 0.0993, "order_stats": 0.0089, "policy_doc": 0.0001, "policy_routing": 0.0223, "sensor_grid": 0.0249, "sla_breach": 0.1296, "sms_spam": 0.0007, "tables": 0.0, "telco_churn": 0.0939, "temporal": 0.0453, "triage_score": 0.0048, "two_draws": 0.0041}, "regress_r2": 0.0009622800054200908, "n_regress": 251, "regress_r2_byfam": {"big_json": -0.0125, "binomial_count": 0.9886, "deterministic_total": 0.6409, "dice_sum": 0.9974, "exp_wait": 0.9981, "hf_stsb": 0.0052, "hf_yelp_reg": -1.2646, "housing_price": 0.0008, "insurance_charges": 0.245, "kc_house_price": 0.0071, "life_expectancy": 0.9476, "long_log": 0.7535, "mixture_delay": 0.9212, "normal_metric": 0.7965, "triangular": 0.9728, "uniform_window": 0.9933}, "regress_r2_fam_mean": 0.4994}
327
+
328
+ ===== ROUND 4 (init=runs/loop_r3/final.pt) =====
329
+ base: 4051
330
+ gen_adversarial: 6076
331
+ gen_enterprise: 10127
332
+ gen_hf: 13165
333
+ gen_kaggle_cls: 10127
334
+ gen_kaggle_reg: 7089
335
+ gen_long: 5063
336
+ gen_numeric: 7089
337
+ gen_stoch: 9114
338
+ gen_structured: 8101
339
+ corpus: 80002 rows -> data/loop_train_r4.jsonl
340
+ + /usr/bin/python3 -m alea.train --train data/loop_train_r4.jsonl --dev data/test_r2.jsonl --out runs/loop_r4 --init-from runs/loop_r3/final.pt --train-mode full --epochs 1.0 --batch 32 --accum 1 --lr-head 0.0001 --lr-backbone 1e-05 --attn sdpa --save-every 100000 --max-len 4096
341
+ `torch_dtype` is deprecated! Use `dtype` instead!
342
+ resumed from runs/loop_r3/final.pt
343
+ pre-packing dataset...
344
+ packed 80002 (skipped 0) trainable=632.6M
345
+ `use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.
346
+ step 50/2645 loss=0.1939 0.92s/step peak=7.4GB
347
+ step 100/2645 loss=1.4186 1.15s/step peak=7.4GB
348
+ step 150/2645 loss=0.3910 1.15s/step peak=7.4GB
349
+ step 200/2645 loss=0.5887 1.21s/step peak=7.4GB
350
+ step 250/2645 loss=0.4067 1.20s/step peak=7.4GB
351
+ step 300/2645 loss=0.4147 1.21s/step peak=7.4GB
352
+ step 350/2645 loss=0.2965 1.18s/step peak=7.4GB
353
+ step 400/2645 loss=0.6344 1.21s/step peak=7.4GB
354
+ step 450/2645 loss=0.4383 1.19s/step peak=7.4GB
355
+ step 500/2645 loss=0.3427 1.18s/step peak=7.4GB
356
+ step 550/2645 loss=0.4871 1.15s/step peak=7.4GB
357
+ step 600/2645 loss=0.7094 1.16s/step peak=7.4GB
358
+ step 650/2645 loss=0.5690 1.16s/step peak=7.4GB
359
+ step 700/2645 loss=0.4274 1.16s/step peak=7.4GB
360
+ step 750/2645 loss=1.1814 1.16s/step peak=7.4GB
361
+ step 800/2645 loss=0.1674 1.15s/step peak=7.4GB
362
+ step 850/2645 loss=0.3632 1.15s/step peak=7.4GB
363
+ step 900/2645 loss=0.4075 1.16s/step peak=7.4GB
364
+ step 950/2645 loss=0.3562 1.16s/step peak=7.4GB
365
+ step 1000/2645 loss=0.2591 1.16s/step peak=7.4GB
366
+ step 1050/2645 loss=0.6965 1.15s/step peak=7.4GB
367
+ step 1100/2645 loss=0.3690 1.13s/step peak=7.4GB
368
+ step 1150/2645 loss=0.3304 1.13s/step peak=7.4GB
369
+ step 1200/2645 loss=0.2017 1.13s/step peak=7.4GB
370
+ step 1250/2645 loss=0.6494 1.13s/step peak=7.4GB
371
+ step 1300/2645 loss=0.2356 1.13s/step peak=7.4GB
372
+ step 1350/2645 loss=0.3191 1.12s/step peak=7.4GB
373
+ step 1400/2645 loss=0.1928 1.11s/step peak=7.4GB
374
+ step 1450/2645 loss=0.3241 1.10s/step peak=7.4GB
375
+ step 1500/2645 loss=0.1791 1.10s/step peak=7.4GB
376
+ step 1550/2645 loss=0.2612 1.10s/step peak=7.4GB
377
+ step 1600/2645 loss=0.1792 1.10s/step peak=7.4GB
378
+ step 1650/2645 loss=0.2443 1.11s/step peak=7.4GB
379
+ step 1700/2645 loss=0.1469 1.10s/step peak=7.4GB
380
+ step 1750/2645 loss=0.2573 1.10s/step peak=7.4GB
381
+ step 1800/2645 loss=0.2313 1.11s/step peak=7.4GB
382
+ step 1850/2645 loss=0.2551 1.11s/step peak=7.4GB
383
+ step 1900/2645 loss=0.1982 1.10s/step peak=7.4GB
384
+ step 1950/2645 loss=0.1787 1.09s/step peak=7.4GB
385
+ step 2000/2645 loss=0.2201 1.11s/step peak=7.4GB
386
+ step 2050/2645 loss=0.2350 1.11s/step peak=7.4GB
387
+ step 2100/2645 loss=0.1909 1.11s/step peak=7.4GB
388
+ step 2150/2645 loss=0.1840 1.11s/step peak=7.4GB
389
+ step 2200/2645 loss=0.2323 1.11s/step peak=7.4GB
390
+ step 2250/2645 loss=0.1907 1.11s/step peak=7.4GB
391
+ step 2300/2645 loss=0.2551 1.12s/step peak=7.4GB
392
+ step 2350/2645 loss=0.1811 1.12s/step peak=7.4GB
393
+ step 2400/2645 loss=0.1496 1.11s/step peak=7.4GB
394
+ step 2450/2645 loss=0.2092 1.11s/step peak=7.4GB
395
+ step 2500/2645 loss=0.1909 1.11s/step peak=7.4GB
396
+ step 2550/2645 loss=0.2170 1.12s/step peak=7.4GB
397
+ step 2600/2645 loss=0.1758 1.12s/step peak=7.4GB
398
+ saved runs/loop_r4/step2645.pt
399
+ done
400
+ + /usr/bin/python3 -m alea.eval --ckpt runs/loop_r4/final.pt --data data/test_r2.jsonl --max-items 1200 --skip-structural
401
+ ROUND 4 METRICS {"n_choice": 967, "acc": 0.7290589451913133, "ece10": 0.04463713796425999, "known_q_l2": 0.2255186494217856, "noul_brier": 0.07311051468800116, "acc_byfam": {"approval_threshold": 0.614, "bayesian_update": 1.0, "big_json": 0.176, "binomial_tail": 1.0, "churn_renewal": 0.531, "compound": 0.875, "conditional_joint": 0.667, "content_mod": 0.857, "creditcard": 0.875, "decoy_shift": 0.87, "eligibility": 0.733, "fraud_review": 0.692, "hf_ag_news": 0.9, "hf_arc_easy": 0.25, "hf_dbpedia": 0.556, "hf_winogrande": 0.625, "hf_yelp": 0.333, "incident_report": 1.0, "insurance_charges": 0.889, "json_records": 0.344, "life_expectancy": 0.75, "log_lines": 0.886, "long_log": 0.842, "markov_chain": 1.0, "mixed_state": 1.0, "multi_choice_stoch": 0.917, "near_duplicate": 1.0, "negation": 1.0, "nested_json": 0.421, "news_cat": 0.55, "none_of_above": 1.0, "numeric_compare": 0.55, "policy_doc": 0.333, "policy_routing": 0.87, "reference_conflict": 1.0, "route": 1.0, "sla_breach": 0.533, "sms_spam": 1.0, "sycophancy_bait": 1.0, "tables": 0.643, "temporal": 0.438, "triage_score": 0.841}, "brier_byfam": {"approval_threshold": 0.061, "bayesian_update": 0.0008, "big_json": 0.1304, "cards": 0.0003, "churn_renewal": 0.0326, "content_mod": 0.002, "creditcard": 0.0762, "deterministic_total": 0.0004, "eligibility": 0.004, "fake_news": 0.0027, "fraud_review": 0.0124, "heart": 0.1438, "hf_amazon": 0.2375, "hf_goemo": 0.152, "hf_squad2": 0.0566, "hf_wnli": 0.2476, "incident_report": 0.0003, "insurance_charges": 0.0936, "json_records": 0.0924, "log_lines": 0.0874, "long_log": 0.3106, "mixed_state": 0.0, "multi_probe": 0.0009, "mushroom": 0.0642, "nested_json": 0.1448, "numeric_compare": 0.1305, "order_stats": 0.0078, "policy_doc": 0.0, "policy_routing": 0.0222, "sensor_grid": 0.0177, "sla_breach": 0.1311, "sms_spam": 0.0004, "tables": 0.0, "telco_churn": 0.0846, "temporal": 0.0462, "triage_score": 0.0042, "two_draws": 0.0036}, "regress_r2": 0.0015058274478552125, "n_regress": 251, "regress_r2_byfam": {"big_json": -0.0016, "binomial_count": 0.9861, "deterministic_total": 0.4638, "dice_sum": 0.9973, "exp_wait": 0.9984, "hf_stsb": -1.4838, "hf_yelp_reg": -2.7089, "housing_price": 0.0012, "insurance_charges": 0.3409, "kc_house_price": 0.0108, "life_expectancy": 0.9544, "long_log": 0.6237, "mixture_delay": 0.946, "normal_metric": 0.8758, "triangular": 0.9748, "uniform_window": 0.9936}, "regress_r2_fam_mean": 0.3108}
402
+
403
+ ===== ROUND 5 (init=runs/loop_r4/final.pt) =====
404
+ base: 4051
405
+ gen_adversarial: 6076
406
+ gen_enterprise: 10127
407
+ gen_hf: 13165
408
+ gen_kaggle_cls: 10127
409
+ gen_kaggle_reg: 7089
410
+ gen_long: 5063
411
+ gen_numeric: 7089
412
+ gen_stoch: 9114
413
+ gen_structured: 8101
414
+ corpus: 80002 rows -> data/loop_train_r5.jsonl
415
+ + /usr/bin/python3 -m alea.train --train data/loop_train_r5.jsonl --dev data/test_r2.jsonl --out runs/loop_r5 --init-from runs/loop_r4/final.pt --train-mode full --epochs 1.0 --batch 32 --accum 1 --lr-head 0.0001 --lr-backbone 1e-05 --attn sdpa --save-every 100000 --max-len 4096
416
+ `torch_dtype` is deprecated! Use `dtype` instead!
417
+ resumed from runs/loop_r4/final.pt
418
+ pre-packing dataset...
419
+ packed 80002 (skipped 0) trainable=632.6M
420
+ `use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.
421
+ step 50/2646 loss=0.3339 1.35s/step peak=7.4GB
422
+ step 100/2646 loss=0.4226 1.30s/step peak=7.4GB
423
+ step 150/2646 loss=0.3175 1.15s/step peak=7.4GB