athrvk commited on
Commit
3dc34fb
·
verified ·
1 Parent(s): 99e34c5

Upload folder using huggingface_hub

Browse files
gold_metrics_json_float32.json CHANGED
@@ -1,32 +1,32 @@
1
  {
2
  "per_file": {
3
  "tests/gold.jsonl": {
4
- "examples": 220,
5
- "spans": 194,
6
- "precision": 0.9585492227979274,
7
- "recall": 0.9536082474226805,
8
- "f1": 0.9560723514211886,
9
- "value_acc": 0.9536082474226805,
10
  "categories": {
11
  "digits": {
12
- "spans": 54,
13
- "value_acc": 0.9629629629629629,
14
- "exact_match_rate": 0.9629629629629629
15
  },
16
  "words": {
17
- "spans": 140,
18
- "value_acc": 0.95,
19
- "exact_match_rate": 0.95
20
  },
21
  "prefix": {
22
  "spans": 52,
23
  "value_acc": 0.9423076923076923,
24
- "exact_match_rate": 0.9423076923076923
25
  },
26
  "range": {
27
- "spans": 14,
28
- "value_acc": 0.9285714285714286,
29
- "exact_match_rate": 0.9285714285714286
30
  },
31
  "currency": {
32
  "spans": 28,
@@ -34,9 +34,9 @@
34
  "exact_match_rate": 1.0
35
  },
36
  "multi_unit": {
37
- "spans": 14,
38
- "value_acc": 1.0,
39
- "exact_match_rate": 1.0
40
  },
41
  "symbol_unit": {
42
  "spans": 19,
@@ -44,24 +44,24 @@
44
  "exact_match_rate": 1.0
45
  },
46
  "long": {
47
- "spans": 10,
48
- "value_acc": 0.8,
49
- "exact_match_rate": 0.8
50
  }
51
  },
52
  "negatives": {
53
- "examples": 35,
54
  "false_positives": 0,
55
  "fp_rate": 0.0
56
  }
57
  },
58
  "tests/gold_deva.jsonl": {
59
- "examples": 182,
60
- "spans": 151,
61
- "precision": 0.9865771812080537,
62
- "recall": 0.9735099337748344,
63
- "f1": 0.98,
64
- "value_acc": 0.9735099337748344,
65
  "categories": {
66
  "digits": {
67
  "spans": 37,
@@ -69,9 +69,9 @@
69
  "exact_match_rate": 1.0
70
  },
71
  "words": {
72
- "spans": 114,
73
- "value_acc": 0.9649122807017544,
74
- "exact_match_rate": 0.9649122807017544
75
  },
76
  "prefix": {
77
  "spans": 29,
@@ -79,18 +79,18 @@
79
  "exact_match_rate": 1.0
80
  },
81
  "range": {
82
- "spans": 10,
83
- "value_acc": 1.0,
84
  "exact_match_rate": 1.0
85
  },
86
  "currency": {
87
  "spans": 20,
88
- "value_acc": 0.95,
89
- "exact_match_rate": 0.95
90
  },
91
  "multi_unit": {
92
- "spans": 7,
93
- "value_acc": 1.0,
94
  "exact_match_rate": 1.0
95
  },
96
  "symbol_unit": {
@@ -102,6 +102,11 @@
102
  "spans": 2,
103
  "value_acc": 1.0,
104
  "exact_match_rate": 1.0
 
 
 
 
 
105
  }
106
  },
107
  "negatives": {
@@ -112,50 +117,50 @@
112
  }
113
  },
114
  "combined": {
115
- "examples": 402,
116
- "spans": 345,
117
- "precision": 0.9707602339181286,
118
- "recall": 0.9623188405797102,
119
- "f1": 0.9665211062590976,
120
- "value_acc": 0.9623188405797102,
121
  "miss_summary": {
122
- "missed_span": 4,
123
  "spurious_span": 0,
124
- "wrong_value": 0,
125
- "wrong_boundary": 9,
126
  "total": 13
127
  },
128
  "categories": {
129
  "digits": {
130
- "spans": 91,
131
- "value_acc": 0.978021978021978,
132
- "exact_match_rate": 0.978021978021978
133
  },
134
  "words": {
135
- "spans": 254,
136
- "value_acc": 0.9566929133858267,
137
- "exact_match_rate": 0.9566929133858267
138
  },
139
  "prefix": {
140
  "spans": 81,
141
  "value_acc": 0.9629629629629629,
142
- "exact_match_rate": 0.9629629629629629
143
  },
144
  "range": {
145
- "spans": 24,
146
- "value_acc": 0.9583333333333334,
147
- "exact_match_rate": 0.9583333333333334
148
  },
149
  "currency": {
150
  "spans": 48,
151
- "value_acc": 0.9791666666666666,
152
- "exact_match_rate": 0.9791666666666666
153
- },
154
- "multi_unit": {
155
- "spans": 21,
156
  "value_acc": 1.0,
157
  "exact_match_rate": 1.0
158
  },
 
 
 
 
 
159
  "symbol_unit": {
160
  "spans": 33,
161
  "value_acc": 1.0,
@@ -167,13 +172,13 @@
167
  "exact_match_rate": 1.0
168
  },
169
  "long": {
170
- "spans": 10,
171
- "value_acc": 0.8,
172
- "exact_match_rate": 0.8
173
  }
174
  },
175
  "negatives": {
176
- "examples": 73,
177
  "false_positives": 0,
178
  "fp_rate": 0.0
179
  }
 
1
  {
2
  "per_file": {
3
  "tests/gold.jsonl": {
4
+ "examples": 225,
5
+ "spans": 197,
6
+ "precision": 0.949748743718593,
7
+ "recall": 0.9593908629441624,
8
+ "f1": 0.9545454545454546,
9
+ "value_acc": 0.9543147208121827,
10
  "categories": {
11
  "digits": {
12
+ "spans": 55,
13
+ "value_acc": 0.9454545454545454,
14
+ "exact_match_rate": 0.9454545454545454
15
  },
16
  "words": {
17
+ "spans": 142,
18
+ "value_acc": 0.9577464788732394,
19
+ "exact_match_rate": 0.9647887323943662
20
  },
21
  "prefix": {
22
  "spans": 52,
23
  "value_acc": 0.9423076923076923,
24
+ "exact_match_rate": 0.9615384615384616
25
  },
26
  "range": {
27
+ "spans": 15,
28
+ "value_acc": 0.8666666666666667,
29
+ "exact_match_rate": 0.8666666666666667
30
  },
31
  "currency": {
32
  "spans": 28,
 
34
  "exact_match_rate": 1.0
35
  },
36
  "multi_unit": {
37
+ "spans": 15,
38
+ "value_acc": 0.9333333333333333,
39
+ "exact_match_rate": 0.9333333333333333
40
  },
41
  "symbol_unit": {
42
  "spans": 19,
 
44
  "exact_match_rate": 1.0
45
  },
46
  "long": {
47
+ "spans": 11,
48
+ "value_acc": 0.7272727272727273,
49
+ "exact_match_rate": 0.7272727272727273
50
  }
51
  },
52
  "negatives": {
53
+ "examples": 37,
54
  "false_positives": 0,
55
  "fp_rate": 0.0
56
  }
57
  },
58
  "tests/gold_deva.jsonl": {
59
+ "examples": 185,
60
+ "spans": 154,
61
+ "precision": 0.9869281045751634,
62
+ "recall": 0.9805194805194806,
63
+ "f1": 0.9837133550488599,
64
+ "value_acc": 0.974025974025974,
65
  "categories": {
66
  "digits": {
67
  "spans": 37,
 
69
  "exact_match_rate": 1.0
70
  },
71
  "words": {
72
+ "spans": 117,
73
+ "value_acc": 0.9658119658119658,
74
+ "exact_match_rate": 0.9743589743589743
75
  },
76
  "prefix": {
77
  "spans": 29,
 
79
  "exact_match_rate": 1.0
80
  },
81
  "range": {
82
+ "spans": 12,
83
+ "value_acc": 0.9166666666666666,
84
  "exact_match_rate": 1.0
85
  },
86
  "currency": {
87
  "spans": 20,
88
+ "value_acc": 1.0,
89
+ "exact_match_rate": 1.0
90
  },
91
  "multi_unit": {
92
+ "spans": 9,
93
+ "value_acc": 0.8888888888888888,
94
  "exact_match_rate": 1.0
95
  },
96
  "symbol_unit": {
 
102
  "spans": 2,
103
  "value_acc": 1.0,
104
  "exact_match_rate": 1.0
105
+ },
106
+ "long": {
107
+ "spans": 1,
108
+ "value_acc": 0.0,
109
+ "exact_match_rate": 1.0
110
  }
111
  },
112
  "negatives": {
 
117
  }
118
  },
119
  "combined": {
120
+ "examples": 410,
121
+ "spans": 351,
122
+ "precision": 0.9659090909090909,
123
+ "recall": 0.9686609686609686,
124
+ "f1": 0.9672830725462305,
125
+ "value_acc": 0.9629629629629629,
126
  "miss_summary": {
127
+ "missed_span": 1,
128
  "spurious_span": 0,
129
+ "wrong_value": 2,
130
+ "wrong_boundary": 10,
131
  "total": 13
132
  },
133
  "categories": {
134
  "digits": {
135
+ "spans": 92,
136
+ "value_acc": 0.967391304347826,
137
+ "exact_match_rate": 0.967391304347826
138
  },
139
  "words": {
140
+ "spans": 259,
141
+ "value_acc": 0.9613899613899614,
142
+ "exact_match_rate": 0.9691119691119691
143
  },
144
  "prefix": {
145
  "spans": 81,
146
  "value_acc": 0.9629629629629629,
147
+ "exact_match_rate": 0.9753086419753086
148
  },
149
  "range": {
150
+ "spans": 27,
151
+ "value_acc": 0.8888888888888888,
152
+ "exact_match_rate": 0.9259259259259259
153
  },
154
  "currency": {
155
  "spans": 48,
 
 
 
 
 
156
  "value_acc": 1.0,
157
  "exact_match_rate": 1.0
158
  },
159
+ "multi_unit": {
160
+ "spans": 24,
161
+ "value_acc": 0.9166666666666666,
162
+ "exact_match_rate": 0.9583333333333334
163
+ },
164
  "symbol_unit": {
165
  "spans": 33,
166
  "value_acc": 1.0,
 
172
  "exact_match_rate": 1.0
173
  },
174
  "long": {
175
+ "spans": 12,
176
+ "value_acc": 0.6666666666666666,
177
+ "exact_match_rate": 0.75
178
  }
179
  },
180
  "negatives": {
181
+ "examples": 75,
182
  "false_positives": 0,
183
  "fp_rate": 0.0
184
  }
gold_metrics_json_int8.json CHANGED
@@ -1,32 +1,32 @@
1
  {
2
  "per_file": {
3
  "tests/gold.jsonl": {
4
- "examples": 220,
5
- "spans": 194,
6
- "precision": 0.9585492227979274,
7
- "recall": 0.9536082474226805,
8
- "f1": 0.9560723514211886,
9
- "value_acc": 0.9536082474226805,
10
  "categories": {
11
  "digits": {
12
- "spans": 54,
13
- "value_acc": 0.9629629629629629,
14
- "exact_match_rate": 0.9629629629629629
15
  },
16
  "words": {
17
- "spans": 140,
18
- "value_acc": 0.95,
19
- "exact_match_rate": 0.95
20
  },
21
  "prefix": {
22
  "spans": 52,
23
  "value_acc": 0.9423076923076923,
24
- "exact_match_rate": 0.9423076923076923
25
  },
26
  "range": {
27
- "spans": 14,
28
- "value_acc": 0.9285714285714286,
29
- "exact_match_rate": 0.9285714285714286
30
  },
31
  "currency": {
32
  "spans": 28,
@@ -34,9 +34,9 @@
34
  "exact_match_rate": 1.0
35
  },
36
  "multi_unit": {
37
- "spans": 14,
38
- "value_acc": 1.0,
39
- "exact_match_rate": 1.0
40
  },
41
  "symbol_unit": {
42
  "spans": 19,
@@ -44,24 +44,24 @@
44
  "exact_match_rate": 1.0
45
  },
46
  "long": {
47
- "spans": 10,
48
- "value_acc": 0.8,
49
- "exact_match_rate": 0.8
50
  }
51
  },
52
  "negatives": {
53
- "examples": 35,
54
  "false_positives": 0,
55
  "fp_rate": 0.0
56
  }
57
  },
58
  "tests/gold_deva.jsonl": {
59
- "examples": 182,
60
- "spans": 151,
61
- "precision": 0.9866666666666667,
62
- "recall": 0.9801324503311258,
63
- "f1": 0.983388704318937,
64
- "value_acc": 0.9801324503311258,
65
  "categories": {
66
  "digits": {
67
  "spans": 37,
@@ -69,9 +69,9 @@
69
  "exact_match_rate": 1.0
70
  },
71
  "words": {
72
- "spans": 114,
73
- "value_acc": 0.9736842105263158,
74
- "exact_match_rate": 0.9736842105263158
75
  },
76
  "prefix": {
77
  "spans": 29,
@@ -79,8 +79,8 @@
79
  "exact_match_rate": 1.0
80
  },
81
  "range": {
82
- "spans": 10,
83
- "value_acc": 1.0,
84
  "exact_match_rate": 1.0
85
  },
86
  "currency": {
@@ -89,8 +89,8 @@
89
  "exact_match_rate": 1.0
90
  },
91
  "multi_unit": {
92
- "spans": 7,
93
- "value_acc": 1.0,
94
  "exact_match_rate": 1.0
95
  },
96
  "symbol_unit": {
@@ -102,6 +102,11 @@
102
  "spans": 2,
103
  "value_acc": 1.0,
104
  "exact_match_rate": 1.0
 
 
 
 
 
105
  }
106
  },
107
  "negatives": {
@@ -112,39 +117,39 @@
112
  }
113
  },
114
  "combined": {
115
- "examples": 402,
116
- "spans": 345,
117
- "precision": 0.9708454810495627,
118
- "recall": 0.9652173913043478,
119
- "f1": 0.9680232558139534,
120
- "value_acc": 0.9652173913043478,
121
  "miss_summary": {
122
- "missed_span": 3,
123
- "spurious_span": 0,
124
- "wrong_value": 0,
125
- "wrong_boundary": 9,
126
- "total": 12
127
  },
128
  "categories": {
129
  "digits": {
130
- "spans": 91,
131
- "value_acc": 0.978021978021978,
132
- "exact_match_rate": 0.978021978021978
133
  },
134
  "words": {
135
- "spans": 254,
136
- "value_acc": 0.9606299212598425,
137
- "exact_match_rate": 0.9606299212598425
138
  },
139
  "prefix": {
140
  "spans": 81,
141
  "value_acc": 0.9629629629629629,
142
- "exact_match_rate": 0.9629629629629629
143
  },
144
  "range": {
145
- "spans": 24,
146
- "value_acc": 0.9583333333333334,
147
- "exact_match_rate": 0.9583333333333334
148
  },
149
  "currency": {
150
  "spans": 48,
@@ -152,9 +157,9 @@
152
  "exact_match_rate": 1.0
153
  },
154
  "multi_unit": {
155
- "spans": 21,
156
- "value_acc": 1.0,
157
- "exact_match_rate": 1.0
158
  },
159
  "symbol_unit": {
160
  "spans": 33,
@@ -167,13 +172,13 @@
167
  "exact_match_rate": 1.0
168
  },
169
  "long": {
170
- "spans": 10,
171
- "value_acc": 0.8,
172
- "exact_match_rate": 0.8
173
  }
174
  },
175
  "negatives": {
176
- "examples": 73,
177
  "false_positives": 0,
178
  "fp_rate": 0.0
179
  }
 
1
  {
2
  "per_file": {
3
  "tests/gold.jsonl": {
4
+ "examples": 225,
5
+ "spans": 197,
6
+ "precision": 0.949748743718593,
7
+ "recall": 0.9593908629441624,
8
+ "f1": 0.9545454545454546,
9
+ "value_acc": 0.9543147208121827,
10
  "categories": {
11
  "digits": {
12
+ "spans": 55,
13
+ "value_acc": 0.9454545454545454,
14
+ "exact_match_rate": 0.9454545454545454
15
  },
16
  "words": {
17
+ "spans": 142,
18
+ "value_acc": 0.9577464788732394,
19
+ "exact_match_rate": 0.9647887323943662
20
  },
21
  "prefix": {
22
  "spans": 52,
23
  "value_acc": 0.9423076923076923,
24
+ "exact_match_rate": 0.9615384615384616
25
  },
26
  "range": {
27
+ "spans": 15,
28
+ "value_acc": 0.8666666666666667,
29
+ "exact_match_rate": 0.8666666666666667
30
  },
31
  "currency": {
32
  "spans": 28,
 
34
  "exact_match_rate": 1.0
35
  },
36
  "multi_unit": {
37
+ "spans": 15,
38
+ "value_acc": 0.9333333333333333,
39
+ "exact_match_rate": 0.9333333333333333
40
  },
41
  "symbol_unit": {
42
  "spans": 19,
 
44
  "exact_match_rate": 1.0
45
  },
46
  "long": {
47
+ "spans": 11,
48
+ "value_acc": 0.7272727272727273,
49
+ "exact_match_rate": 0.7272727272727273
50
  }
51
  },
52
  "negatives": {
53
+ "examples": 37,
54
  "false_positives": 0,
55
  "fp_rate": 0.0
56
  }
57
  },
58
  "tests/gold_deva.jsonl": {
59
+ "examples": 185,
60
+ "spans": 154,
61
+ "precision": 0.9805194805194806,
62
+ "recall": 0.9805194805194806,
63
+ "f1": 0.9805194805194806,
64
+ "value_acc": 0.974025974025974,
65
  "categories": {
66
  "digits": {
67
  "spans": 37,
 
69
  "exact_match_rate": 1.0
70
  },
71
  "words": {
72
+ "spans": 117,
73
+ "value_acc": 0.9658119658119658,
74
+ "exact_match_rate": 0.9743589743589743
75
  },
76
  "prefix": {
77
  "spans": 29,
 
79
  "exact_match_rate": 1.0
80
  },
81
  "range": {
82
+ "spans": 12,
83
+ "value_acc": 0.9166666666666666,
84
  "exact_match_rate": 1.0
85
  },
86
  "currency": {
 
89
  "exact_match_rate": 1.0
90
  },
91
  "multi_unit": {
92
+ "spans": 9,
93
+ "value_acc": 0.8888888888888888,
94
  "exact_match_rate": 1.0
95
  },
96
  "symbol_unit": {
 
102
  "spans": 2,
103
  "value_acc": 1.0,
104
  "exact_match_rate": 1.0
105
+ },
106
+ "long": {
107
+ "spans": 1,
108
+ "value_acc": 0.0,
109
+ "exact_match_rate": 1.0
110
  }
111
  },
112
  "negatives": {
 
117
  }
118
  },
119
  "combined": {
120
+ "examples": 410,
121
+ "spans": 351,
122
+ "precision": 0.9631728045325779,
123
+ "recall": 0.9686609686609686,
124
+ "f1": 0.9659090909090909,
125
+ "value_acc": 0.9629629629629629,
126
  "miss_summary": {
127
+ "missed_span": 1,
128
+ "spurious_span": 1,
129
+ "wrong_value": 2,
130
+ "wrong_boundary": 10,
131
+ "total": 14
132
  },
133
  "categories": {
134
  "digits": {
135
+ "spans": 92,
136
+ "value_acc": 0.967391304347826,
137
+ "exact_match_rate": 0.967391304347826
138
  },
139
  "words": {
140
+ "spans": 259,
141
+ "value_acc": 0.9613899613899614,
142
+ "exact_match_rate": 0.9691119691119691
143
  },
144
  "prefix": {
145
  "spans": 81,
146
  "value_acc": 0.9629629629629629,
147
+ "exact_match_rate": 0.9753086419753086
148
  },
149
  "range": {
150
+ "spans": 27,
151
+ "value_acc": 0.8888888888888888,
152
+ "exact_match_rate": 0.9259259259259259
153
  },
154
  "currency": {
155
  "spans": 48,
 
157
  "exact_match_rate": 1.0
158
  },
159
  "multi_unit": {
160
+ "spans": 24,
161
+ "value_acc": 0.9166666666666666,
162
+ "exact_match_rate": 0.9583333333333334
163
  },
164
  "symbol_unit": {
165
  "spans": 33,
 
172
  "exact_match_rate": 1.0
173
  },
174
  "long": {
175
+ "spans": 12,
176
+ "value_acc": 0.6666666666666666,
177
+ "exact_match_rate": 0.75
178
  }
179
  },
180
  "negatives": {
181
+ "examples": 75,
182
  "false_positives": 0,
183
  "fp_rate": 0.0
184
  }
gold_metrics_torch.json CHANGED
@@ -1,32 +1,32 @@
1
  {
2
  "per_file": {
3
  "tests/gold.jsonl": {
4
- "examples": 220,
5
- "spans": 194,
6
- "precision": 0.9585492227979274,
7
- "recall": 0.9536082474226805,
8
- "f1": 0.9560723514211886,
9
- "value_acc": 0.9536082474226805,
10
  "categories": {
11
  "digits": {
12
- "spans": 54,
13
- "value_acc": 0.9629629629629629,
14
- "exact_match_rate": 0.9629629629629629
15
  },
16
  "words": {
17
- "spans": 140,
18
- "value_acc": 0.95,
19
- "exact_match_rate": 0.95
20
  },
21
  "prefix": {
22
  "spans": 52,
23
  "value_acc": 0.9423076923076923,
24
- "exact_match_rate": 0.9423076923076923
25
  },
26
  "range": {
27
- "spans": 14,
28
- "value_acc": 0.9285714285714286,
29
- "exact_match_rate": 0.9285714285714286
30
  },
31
  "currency": {
32
  "spans": 28,
@@ -34,9 +34,9 @@
34
  "exact_match_rate": 1.0
35
  },
36
  "multi_unit": {
37
- "spans": 14,
38
- "value_acc": 1.0,
39
- "exact_match_rate": 1.0
40
  },
41
  "symbol_unit": {
42
  "spans": 19,
@@ -44,24 +44,24 @@
44
  "exact_match_rate": 1.0
45
  },
46
  "long": {
47
- "spans": 10,
48
- "value_acc": 0.8,
49
- "exact_match_rate": 0.8
50
  }
51
  },
52
  "negatives": {
53
- "examples": 35,
54
  "false_positives": 0,
55
  "fp_rate": 0.0
56
  }
57
  },
58
  "tests/gold_deva.jsonl": {
59
- "examples": 182,
60
- "spans": 151,
61
- "precision": 0.9865771812080537,
62
- "recall": 0.9735099337748344,
63
- "f1": 0.98,
64
- "value_acc": 0.9735099337748344,
65
  "categories": {
66
  "digits": {
67
  "spans": 37,
@@ -69,9 +69,9 @@
69
  "exact_match_rate": 1.0
70
  },
71
  "words": {
72
- "spans": 114,
73
- "value_acc": 0.9649122807017544,
74
- "exact_match_rate": 0.9649122807017544
75
  },
76
  "prefix": {
77
  "spans": 29,
@@ -79,18 +79,18 @@
79
  "exact_match_rate": 1.0
80
  },
81
  "range": {
82
- "spans": 10,
83
- "value_acc": 1.0,
84
  "exact_match_rate": 1.0
85
  },
86
  "currency": {
87
  "spans": 20,
88
- "value_acc": 0.95,
89
- "exact_match_rate": 0.95
90
  },
91
  "multi_unit": {
92
- "spans": 7,
93
- "value_acc": 1.0,
94
  "exact_match_rate": 1.0
95
  },
96
  "symbol_unit": {
@@ -102,6 +102,11 @@
102
  "spans": 2,
103
  "value_acc": 1.0,
104
  "exact_match_rate": 1.0
 
 
 
 
 
105
  }
106
  },
107
  "negatives": {
@@ -112,50 +117,50 @@
112
  }
113
  },
114
  "combined": {
115
- "examples": 402,
116
- "spans": 345,
117
- "precision": 0.9707602339181286,
118
- "recall": 0.9623188405797102,
119
- "f1": 0.9665211062590976,
120
- "value_acc": 0.9623188405797102,
121
  "miss_summary": {
122
- "missed_span": 4,
123
  "spurious_span": 0,
124
- "wrong_value": 0,
125
- "wrong_boundary": 9,
126
  "total": 13
127
  },
128
  "categories": {
129
  "digits": {
130
- "spans": 91,
131
- "value_acc": 0.978021978021978,
132
- "exact_match_rate": 0.978021978021978
133
  },
134
  "words": {
135
- "spans": 254,
136
- "value_acc": 0.9566929133858267,
137
- "exact_match_rate": 0.9566929133858267
138
  },
139
  "prefix": {
140
  "spans": 81,
141
  "value_acc": 0.9629629629629629,
142
- "exact_match_rate": 0.9629629629629629
143
  },
144
  "range": {
145
- "spans": 24,
146
- "value_acc": 0.9583333333333334,
147
- "exact_match_rate": 0.9583333333333334
148
  },
149
  "currency": {
150
  "spans": 48,
151
- "value_acc": 0.9791666666666666,
152
- "exact_match_rate": 0.9791666666666666
153
- },
154
- "multi_unit": {
155
- "spans": 21,
156
  "value_acc": 1.0,
157
  "exact_match_rate": 1.0
158
  },
 
 
 
 
 
159
  "symbol_unit": {
160
  "spans": 33,
161
  "value_acc": 1.0,
@@ -167,13 +172,13 @@
167
  "exact_match_rate": 1.0
168
  },
169
  "long": {
170
- "spans": 10,
171
- "value_acc": 0.8,
172
- "exact_match_rate": 0.8
173
  }
174
  },
175
  "negatives": {
176
- "examples": 73,
177
  "false_positives": 0,
178
  "fp_rate": 0.0
179
  }
 
1
  {
2
  "per_file": {
3
  "tests/gold.jsonl": {
4
+ "examples": 225,
5
+ "spans": 197,
6
+ "precision": 0.949748743718593,
7
+ "recall": 0.9593908629441624,
8
+ "f1": 0.9545454545454546,
9
+ "value_acc": 0.9543147208121827,
10
  "categories": {
11
  "digits": {
12
+ "spans": 55,
13
+ "value_acc": 0.9454545454545454,
14
+ "exact_match_rate": 0.9454545454545454
15
  },
16
  "words": {
17
+ "spans": 142,
18
+ "value_acc": 0.9577464788732394,
19
+ "exact_match_rate": 0.9647887323943662
20
  },
21
  "prefix": {
22
  "spans": 52,
23
  "value_acc": 0.9423076923076923,
24
+ "exact_match_rate": 0.9615384615384616
25
  },
26
  "range": {
27
+ "spans": 15,
28
+ "value_acc": 0.8666666666666667,
29
+ "exact_match_rate": 0.8666666666666667
30
  },
31
  "currency": {
32
  "spans": 28,
 
34
  "exact_match_rate": 1.0
35
  },
36
  "multi_unit": {
37
+ "spans": 15,
38
+ "value_acc": 0.9333333333333333,
39
+ "exact_match_rate": 0.9333333333333333
40
  },
41
  "symbol_unit": {
42
  "spans": 19,
 
44
  "exact_match_rate": 1.0
45
  },
46
  "long": {
47
+ "spans": 11,
48
+ "value_acc": 0.7272727272727273,
49
+ "exact_match_rate": 0.7272727272727273
50
  }
51
  },
52
  "negatives": {
53
+ "examples": 37,
54
  "false_positives": 0,
55
  "fp_rate": 0.0
56
  }
57
  },
58
  "tests/gold_deva.jsonl": {
59
+ "examples": 185,
60
+ "spans": 154,
61
+ "precision": 0.9869281045751634,
62
+ "recall": 0.9805194805194806,
63
+ "f1": 0.9837133550488599,
64
+ "value_acc": 0.974025974025974,
65
  "categories": {
66
  "digits": {
67
  "spans": 37,
 
69
  "exact_match_rate": 1.0
70
  },
71
  "words": {
72
+ "spans": 117,
73
+ "value_acc": 0.9658119658119658,
74
+ "exact_match_rate": 0.9743589743589743
75
  },
76
  "prefix": {
77
  "spans": 29,
 
79
  "exact_match_rate": 1.0
80
  },
81
  "range": {
82
+ "spans": 12,
83
+ "value_acc": 0.9166666666666666,
84
  "exact_match_rate": 1.0
85
  },
86
  "currency": {
87
  "spans": 20,
88
+ "value_acc": 1.0,
89
+ "exact_match_rate": 1.0
90
  },
91
  "multi_unit": {
92
+ "spans": 9,
93
+ "value_acc": 0.8888888888888888,
94
  "exact_match_rate": 1.0
95
  },
96
  "symbol_unit": {
 
102
  "spans": 2,
103
  "value_acc": 1.0,
104
  "exact_match_rate": 1.0
105
+ },
106
+ "long": {
107
+ "spans": 1,
108
+ "value_acc": 0.0,
109
+ "exact_match_rate": 1.0
110
  }
111
  },
112
  "negatives": {
 
117
  }
118
  },
119
  "combined": {
120
+ "examples": 410,
121
+ "spans": 351,
122
+ "precision": 0.9659090909090909,
123
+ "recall": 0.9686609686609686,
124
+ "f1": 0.9672830725462305,
125
+ "value_acc": 0.9629629629629629,
126
  "miss_summary": {
127
+ "missed_span": 1,
128
  "spurious_span": 0,
129
+ "wrong_value": 2,
130
+ "wrong_boundary": 10,
131
  "total": 13
132
  },
133
  "categories": {
134
  "digits": {
135
+ "spans": 92,
136
+ "value_acc": 0.967391304347826,
137
+ "exact_match_rate": 0.967391304347826
138
  },
139
  "words": {
140
+ "spans": 259,
141
+ "value_acc": 0.9613899613899614,
142
+ "exact_match_rate": 0.9691119691119691
143
  },
144
  "prefix": {
145
  "spans": 81,
146
  "value_acc": 0.9629629629629629,
147
+ "exact_match_rate": 0.9753086419753086
148
  },
149
  "range": {
150
+ "spans": 27,
151
+ "value_acc": 0.8888888888888888,
152
+ "exact_match_rate": 0.9259259259259259
153
  },
154
  "currency": {
155
  "spans": 48,
 
 
 
 
 
156
  "value_acc": 1.0,
157
  "exact_match_rate": 1.0
158
  },
159
+ "multi_unit": {
160
+ "spans": 24,
161
+ "value_acc": 0.9166666666666666,
162
+ "exact_match_rate": 0.9583333333333334
163
+ },
164
  "symbol_unit": {
165
  "spans": 33,
166
  "value_acc": 1.0,
 
172
  "exact_match_rate": 1.0
173
  },
174
  "long": {
175
+ "spans": 12,
176
+ "value_acc": 0.6666666666666666,
177
+ "exact_match_rate": 0.75
178
  }
179
  },
180
  "negatives": {
181
+ "examples": 75,
182
  "false_positives": 0,
183
  "fp_rate": 0.0
184
  }
kaggle_metrics.json CHANGED
@@ -10,40 +10,41 @@
10
  "epochs": 20,
11
  "arch": "v2",
12
  "channels": 48,
13
- "seed": 2,
14
- "matrix": "v2:48:0,v2:48:1,v2:48:2",
15
- "smoke": false
 
16
  },
17
  "gold": {
18
  "torch": {
19
  "per_file": {
20
  "tests/gold.jsonl": {
21
- "examples": 220,
22
- "spans": 194,
23
- "precision": 0.9585492227979274,
24
- "recall": 0.9536082474226805,
25
- "f1": 0.9560723514211886,
26
- "value_acc": 0.9536082474226805,
27
  "categories": {
28
  "digits": {
29
- "spans": 54,
30
- "value_acc": 0.9629629629629629,
31
- "exact_match_rate": 0.9629629629629629
32
  },
33
  "words": {
34
- "spans": 140,
35
- "value_acc": 0.95,
36
- "exact_match_rate": 0.95
37
  },
38
  "prefix": {
39
  "spans": 52,
40
  "value_acc": 0.9423076923076923,
41
- "exact_match_rate": 0.9423076923076923
42
  },
43
  "range": {
44
- "spans": 14,
45
- "value_acc": 0.9285714285714286,
46
- "exact_match_rate": 0.9285714285714286
47
  },
48
  "currency": {
49
  "spans": 28,
@@ -51,9 +52,9 @@
51
  "exact_match_rate": 1.0
52
  },
53
  "multi_unit": {
54
- "spans": 14,
55
- "value_acc": 1.0,
56
- "exact_match_rate": 1.0
57
  },
58
  "symbol_unit": {
59
  "spans": 19,
@@ -61,24 +62,24 @@
61
  "exact_match_rate": 1.0
62
  },
63
  "long": {
64
- "spans": 10,
65
- "value_acc": 0.8,
66
- "exact_match_rate": 0.8
67
  }
68
  },
69
  "negatives": {
70
- "examples": 35,
71
  "false_positives": 0,
72
  "fp_rate": 0.0
73
  }
74
  },
75
  "tests/gold_deva.jsonl": {
76
- "examples": 182,
77
- "spans": 151,
78
- "precision": 0.9865771812080537,
79
- "recall": 0.9735099337748344,
80
- "f1": 0.98,
81
- "value_acc": 0.9735099337748344,
82
  "categories": {
83
  "digits": {
84
  "spans": 37,
@@ -86,9 +87,9 @@
86
  "exact_match_rate": 1.0
87
  },
88
  "words": {
89
- "spans": 114,
90
- "value_acc": 0.9649122807017544,
91
- "exact_match_rate": 0.9649122807017544
92
  },
93
  "prefix": {
94
  "spans": 29,
@@ -96,18 +97,18 @@
96
  "exact_match_rate": 1.0
97
  },
98
  "range": {
99
- "spans": 10,
100
- "value_acc": 1.0,
101
  "exact_match_rate": 1.0
102
  },
103
  "currency": {
104
  "spans": 20,
105
- "value_acc": 0.95,
106
- "exact_match_rate": 0.95
107
  },
108
  "multi_unit": {
109
- "spans": 7,
110
- "value_acc": 1.0,
111
  "exact_match_rate": 1.0
112
  },
113
  "symbol_unit": {
@@ -119,6 +120,11 @@
119
  "spans": 2,
120
  "value_acc": 1.0,
121
  "exact_match_rate": 1.0
 
 
 
 
 
122
  }
123
  },
124
  "negatives": {
@@ -129,50 +135,50 @@
129
  }
130
  },
131
  "combined": {
132
- "examples": 402,
133
- "spans": 345,
134
- "precision": 0.9707602339181286,
135
- "recall": 0.9623188405797102,
136
- "f1": 0.9665211062590976,
137
- "value_acc": 0.9623188405797102,
138
  "miss_summary": {
139
- "missed_span": 4,
140
  "spurious_span": 0,
141
- "wrong_value": 0,
142
- "wrong_boundary": 9,
143
  "total": 13
144
  },
145
  "categories": {
146
  "digits": {
147
- "spans": 91,
148
- "value_acc": 0.978021978021978,
149
- "exact_match_rate": 0.978021978021978
150
  },
151
  "words": {
152
- "spans": 254,
153
- "value_acc": 0.9566929133858267,
154
- "exact_match_rate": 0.9566929133858267
155
  },
156
  "prefix": {
157
  "spans": 81,
158
  "value_acc": 0.9629629629629629,
159
- "exact_match_rate": 0.9629629629629629
160
  },
161
  "range": {
162
- "spans": 24,
163
- "value_acc": 0.9583333333333334,
164
- "exact_match_rate": 0.9583333333333334
165
  },
166
  "currency": {
167
  "spans": 48,
168
- "value_acc": 0.9791666666666666,
169
- "exact_match_rate": 0.9791666666666666
170
- },
171
- "multi_unit": {
172
- "spans": 21,
173
  "value_acc": 1.0,
174
  "exact_match_rate": 1.0
175
  },
 
 
 
 
 
176
  "symbol_unit": {
177
  "spans": 33,
178
  "value_acc": 1.0,
@@ -184,13 +190,13 @@
184
  "exact_match_rate": 1.0
185
  },
186
  "long": {
187
- "spans": 10,
188
- "value_acc": 0.8,
189
- "exact_match_rate": 0.8
190
  }
191
  },
192
  "negatives": {
193
- "examples": 73,
194
  "false_positives": 0,
195
  "fp_rate": 0.0
196
  }
@@ -199,32 +205,32 @@
199
  "json_float32": {
200
  "per_file": {
201
  "tests/gold.jsonl": {
202
- "examples": 220,
203
- "spans": 194,
204
- "precision": 0.9585492227979274,
205
- "recall": 0.9536082474226805,
206
- "f1": 0.9560723514211886,
207
- "value_acc": 0.9536082474226805,
208
  "categories": {
209
  "digits": {
210
- "spans": 54,
211
- "value_acc": 0.9629629629629629,
212
- "exact_match_rate": 0.9629629629629629
213
  },
214
  "words": {
215
- "spans": 140,
216
- "value_acc": 0.95,
217
- "exact_match_rate": 0.95
218
  },
219
  "prefix": {
220
  "spans": 52,
221
  "value_acc": 0.9423076923076923,
222
- "exact_match_rate": 0.9423076923076923
223
  },
224
  "range": {
225
- "spans": 14,
226
- "value_acc": 0.9285714285714286,
227
- "exact_match_rate": 0.9285714285714286
228
  },
229
  "currency": {
230
  "spans": 28,
@@ -232,9 +238,9 @@
232
  "exact_match_rate": 1.0
233
  },
234
  "multi_unit": {
235
- "spans": 14,
236
- "value_acc": 1.0,
237
- "exact_match_rate": 1.0
238
  },
239
  "symbol_unit": {
240
  "spans": 19,
@@ -242,24 +248,24 @@
242
  "exact_match_rate": 1.0
243
  },
244
  "long": {
245
- "spans": 10,
246
- "value_acc": 0.8,
247
- "exact_match_rate": 0.8
248
  }
249
  },
250
  "negatives": {
251
- "examples": 35,
252
  "false_positives": 0,
253
  "fp_rate": 0.0
254
  }
255
  },
256
  "tests/gold_deva.jsonl": {
257
- "examples": 182,
258
- "spans": 151,
259
- "precision": 0.9865771812080537,
260
- "recall": 0.9735099337748344,
261
- "f1": 0.98,
262
- "value_acc": 0.9735099337748344,
263
  "categories": {
264
  "digits": {
265
  "spans": 37,
@@ -267,9 +273,9 @@
267
  "exact_match_rate": 1.0
268
  },
269
  "words": {
270
- "spans": 114,
271
- "value_acc": 0.9649122807017544,
272
- "exact_match_rate": 0.9649122807017544
273
  },
274
  "prefix": {
275
  "spans": 29,
@@ -277,18 +283,18 @@
277
  "exact_match_rate": 1.0
278
  },
279
  "range": {
280
- "spans": 10,
281
- "value_acc": 1.0,
282
  "exact_match_rate": 1.0
283
  },
284
  "currency": {
285
  "spans": 20,
286
- "value_acc": 0.95,
287
- "exact_match_rate": 0.95
288
  },
289
  "multi_unit": {
290
- "spans": 7,
291
- "value_acc": 1.0,
292
  "exact_match_rate": 1.0
293
  },
294
  "symbol_unit": {
@@ -300,6 +306,11 @@
300
  "spans": 2,
301
  "value_acc": 1.0,
302
  "exact_match_rate": 1.0
 
 
 
 
 
303
  }
304
  },
305
  "negatives": {
@@ -310,50 +321,50 @@
310
  }
311
  },
312
  "combined": {
313
- "examples": 402,
314
- "spans": 345,
315
- "precision": 0.9707602339181286,
316
- "recall": 0.9623188405797102,
317
- "f1": 0.9665211062590976,
318
- "value_acc": 0.9623188405797102,
319
  "miss_summary": {
320
- "missed_span": 4,
321
  "spurious_span": 0,
322
- "wrong_value": 0,
323
- "wrong_boundary": 9,
324
  "total": 13
325
  },
326
  "categories": {
327
  "digits": {
328
- "spans": 91,
329
- "value_acc": 0.978021978021978,
330
- "exact_match_rate": 0.978021978021978
331
  },
332
  "words": {
333
- "spans": 254,
334
- "value_acc": 0.9566929133858267,
335
- "exact_match_rate": 0.9566929133858267
336
  },
337
  "prefix": {
338
  "spans": 81,
339
  "value_acc": 0.9629629629629629,
340
- "exact_match_rate": 0.9629629629629629
341
  },
342
  "range": {
343
- "spans": 24,
344
- "value_acc": 0.9583333333333334,
345
- "exact_match_rate": 0.9583333333333334
346
  },
347
  "currency": {
348
  "spans": 48,
349
- "value_acc": 0.9791666666666666,
350
- "exact_match_rate": 0.9791666666666666
351
- },
352
- "multi_unit": {
353
- "spans": 21,
354
  "value_acc": 1.0,
355
  "exact_match_rate": 1.0
356
  },
 
 
 
 
 
357
  "symbol_unit": {
358
  "spans": 33,
359
  "value_acc": 1.0,
@@ -365,13 +376,13 @@
365
  "exact_match_rate": 1.0
366
  },
367
  "long": {
368
- "spans": 10,
369
- "value_acc": 0.8,
370
- "exact_match_rate": 0.8
371
  }
372
  },
373
  "negatives": {
374
- "examples": 73,
375
  "false_positives": 0,
376
  "fp_rate": 0.0
377
  }
@@ -380,32 +391,32 @@
380
  "json_int8": {
381
  "per_file": {
382
  "tests/gold.jsonl": {
383
- "examples": 220,
384
- "spans": 194,
385
- "precision": 0.9585492227979274,
386
- "recall": 0.9536082474226805,
387
- "f1": 0.9560723514211886,
388
- "value_acc": 0.9536082474226805,
389
  "categories": {
390
  "digits": {
391
- "spans": 54,
392
- "value_acc": 0.9629629629629629,
393
- "exact_match_rate": 0.9629629629629629
394
  },
395
  "words": {
396
- "spans": 140,
397
- "value_acc": 0.95,
398
- "exact_match_rate": 0.95
399
  },
400
  "prefix": {
401
  "spans": 52,
402
  "value_acc": 0.9423076923076923,
403
- "exact_match_rate": 0.9423076923076923
404
  },
405
  "range": {
406
- "spans": 14,
407
- "value_acc": 0.9285714285714286,
408
- "exact_match_rate": 0.9285714285714286
409
  },
410
  "currency": {
411
  "spans": 28,
@@ -413,9 +424,9 @@
413
  "exact_match_rate": 1.0
414
  },
415
  "multi_unit": {
416
- "spans": 14,
417
- "value_acc": 1.0,
418
- "exact_match_rate": 1.0
419
  },
420
  "symbol_unit": {
421
  "spans": 19,
@@ -423,24 +434,24 @@
423
  "exact_match_rate": 1.0
424
  },
425
  "long": {
426
- "spans": 10,
427
- "value_acc": 0.8,
428
- "exact_match_rate": 0.8
429
  }
430
  },
431
  "negatives": {
432
- "examples": 35,
433
  "false_positives": 0,
434
  "fp_rate": 0.0
435
  }
436
  },
437
  "tests/gold_deva.jsonl": {
438
- "examples": 182,
439
- "spans": 151,
440
- "precision": 0.9866666666666667,
441
- "recall": 0.9801324503311258,
442
- "f1": 0.983388704318937,
443
- "value_acc": 0.9801324503311258,
444
  "categories": {
445
  "digits": {
446
  "spans": 37,
@@ -448,9 +459,9 @@
448
  "exact_match_rate": 1.0
449
  },
450
  "words": {
451
- "spans": 114,
452
- "value_acc": 0.9736842105263158,
453
- "exact_match_rate": 0.9736842105263158
454
  },
455
  "prefix": {
456
  "spans": 29,
@@ -458,8 +469,8 @@
458
  "exact_match_rate": 1.0
459
  },
460
  "range": {
461
- "spans": 10,
462
- "value_acc": 1.0,
463
  "exact_match_rate": 1.0
464
  },
465
  "currency": {
@@ -468,8 +479,8 @@
468
  "exact_match_rate": 1.0
469
  },
470
  "multi_unit": {
471
- "spans": 7,
472
- "value_acc": 1.0,
473
  "exact_match_rate": 1.0
474
  },
475
  "symbol_unit": {
@@ -481,6 +492,11 @@
481
  "spans": 2,
482
  "value_acc": 1.0,
483
  "exact_match_rate": 1.0
 
 
 
 
 
484
  }
485
  },
486
  "negatives": {
@@ -491,39 +507,39 @@
491
  }
492
  },
493
  "combined": {
494
- "examples": 402,
495
- "spans": 345,
496
- "precision": 0.9708454810495627,
497
- "recall": 0.9652173913043478,
498
- "f1": 0.9680232558139534,
499
- "value_acc": 0.9652173913043478,
500
  "miss_summary": {
501
- "missed_span": 3,
502
- "spurious_span": 0,
503
- "wrong_value": 0,
504
- "wrong_boundary": 9,
505
- "total": 12
506
  },
507
  "categories": {
508
  "digits": {
509
- "spans": 91,
510
- "value_acc": 0.978021978021978,
511
- "exact_match_rate": 0.978021978021978
512
  },
513
  "words": {
514
- "spans": 254,
515
- "value_acc": 0.9606299212598425,
516
- "exact_match_rate": 0.9606299212598425
517
  },
518
  "prefix": {
519
  "spans": 81,
520
  "value_acc": 0.9629629629629629,
521
- "exact_match_rate": 0.9629629629629629
522
  },
523
  "range": {
524
- "spans": 24,
525
- "value_acc": 0.9583333333333334,
526
- "exact_match_rate": 0.9583333333333334
527
  },
528
  "currency": {
529
  "spans": 48,
@@ -531,9 +547,9 @@
531
  "exact_match_rate": 1.0
532
  },
533
  "multi_unit": {
534
- "spans": 21,
535
- "value_acc": 1.0,
536
- "exact_match_rate": 1.0
537
  },
538
  "symbol_unit": {
539
  "spans": 33,
@@ -546,13 +562,13 @@
546
  "exact_match_rate": 1.0
547
  },
548
  "long": {
549
- "spans": 10,
550
- "value_acc": 0.8,
551
- "exact_match_rate": 0.8
552
  }
553
  },
554
  "negatives": {
555
- "examples": 73,
556
  "false_positives": 0,
557
  "fp_rate": 0.0
558
  }
 
10
  "epochs": 20,
11
  "arch": "v2",
12
  "channels": 48,
13
+ "seed": 0,
14
+ "matrix": "v2:48:0 (local CPU; see matrix.md)",
15
+ "smoke": false,
16
+ "device": "cpu"
17
  },
18
  "gold": {
19
  "torch": {
20
  "per_file": {
21
  "tests/gold.jsonl": {
22
+ "examples": 225,
23
+ "spans": 197,
24
+ "precision": 0.949748743718593,
25
+ "recall": 0.9593908629441624,
26
+ "f1": 0.9545454545454546,
27
+ "value_acc": 0.9543147208121827,
28
  "categories": {
29
  "digits": {
30
+ "spans": 55,
31
+ "value_acc": 0.9454545454545454,
32
+ "exact_match_rate": 0.9454545454545454
33
  },
34
  "words": {
35
+ "spans": 142,
36
+ "value_acc": 0.9577464788732394,
37
+ "exact_match_rate": 0.9647887323943662
38
  },
39
  "prefix": {
40
  "spans": 52,
41
  "value_acc": 0.9423076923076923,
42
+ "exact_match_rate": 0.9615384615384616
43
  },
44
  "range": {
45
+ "spans": 15,
46
+ "value_acc": 0.8666666666666667,
47
+ "exact_match_rate": 0.8666666666666667
48
  },
49
  "currency": {
50
  "spans": 28,
 
52
  "exact_match_rate": 1.0
53
  },
54
  "multi_unit": {
55
+ "spans": 15,
56
+ "value_acc": 0.9333333333333333,
57
+ "exact_match_rate": 0.9333333333333333
58
  },
59
  "symbol_unit": {
60
  "spans": 19,
 
62
  "exact_match_rate": 1.0
63
  },
64
  "long": {
65
+ "spans": 11,
66
+ "value_acc": 0.7272727272727273,
67
+ "exact_match_rate": 0.7272727272727273
68
  }
69
  },
70
  "negatives": {
71
+ "examples": 37,
72
  "false_positives": 0,
73
  "fp_rate": 0.0
74
  }
75
  },
76
  "tests/gold_deva.jsonl": {
77
+ "examples": 185,
78
+ "spans": 154,
79
+ "precision": 0.9869281045751634,
80
+ "recall": 0.9805194805194806,
81
+ "f1": 0.9837133550488599,
82
+ "value_acc": 0.974025974025974,
83
  "categories": {
84
  "digits": {
85
  "spans": 37,
 
87
  "exact_match_rate": 1.0
88
  },
89
  "words": {
90
+ "spans": 117,
91
+ "value_acc": 0.9658119658119658,
92
+ "exact_match_rate": 0.9743589743589743
93
  },
94
  "prefix": {
95
  "spans": 29,
 
97
  "exact_match_rate": 1.0
98
  },
99
  "range": {
100
+ "spans": 12,
101
+ "value_acc": 0.9166666666666666,
102
  "exact_match_rate": 1.0
103
  },
104
  "currency": {
105
  "spans": 20,
106
+ "value_acc": 1.0,
107
+ "exact_match_rate": 1.0
108
  },
109
  "multi_unit": {
110
+ "spans": 9,
111
+ "value_acc": 0.8888888888888888,
112
  "exact_match_rate": 1.0
113
  },
114
  "symbol_unit": {
 
120
  "spans": 2,
121
  "value_acc": 1.0,
122
  "exact_match_rate": 1.0
123
+ },
124
+ "long": {
125
+ "spans": 1,
126
+ "value_acc": 0.0,
127
+ "exact_match_rate": 1.0
128
  }
129
  },
130
  "negatives": {
 
135
  }
136
  },
137
  "combined": {
138
+ "examples": 410,
139
+ "spans": 351,
140
+ "precision": 0.9659090909090909,
141
+ "recall": 0.9686609686609686,
142
+ "f1": 0.9672830725462305,
143
+ "value_acc": 0.9629629629629629,
144
  "miss_summary": {
145
+ "missed_span": 1,
146
  "spurious_span": 0,
147
+ "wrong_value": 2,
148
+ "wrong_boundary": 10,
149
  "total": 13
150
  },
151
  "categories": {
152
  "digits": {
153
+ "spans": 92,
154
+ "value_acc": 0.967391304347826,
155
+ "exact_match_rate": 0.967391304347826
156
  },
157
  "words": {
158
+ "spans": 259,
159
+ "value_acc": 0.9613899613899614,
160
+ "exact_match_rate": 0.9691119691119691
161
  },
162
  "prefix": {
163
  "spans": 81,
164
  "value_acc": 0.9629629629629629,
165
+ "exact_match_rate": 0.9753086419753086
166
  },
167
  "range": {
168
+ "spans": 27,
169
+ "value_acc": 0.8888888888888888,
170
+ "exact_match_rate": 0.9259259259259259
171
  },
172
  "currency": {
173
  "spans": 48,
 
 
 
 
 
174
  "value_acc": 1.0,
175
  "exact_match_rate": 1.0
176
  },
177
+ "multi_unit": {
178
+ "spans": 24,
179
+ "value_acc": 0.9166666666666666,
180
+ "exact_match_rate": 0.9583333333333334
181
+ },
182
  "symbol_unit": {
183
  "spans": 33,
184
  "value_acc": 1.0,
 
190
  "exact_match_rate": 1.0
191
  },
192
  "long": {
193
+ "spans": 12,
194
+ "value_acc": 0.6666666666666666,
195
+ "exact_match_rate": 0.75
196
  }
197
  },
198
  "negatives": {
199
+ "examples": 75,
200
  "false_positives": 0,
201
  "fp_rate": 0.0
202
  }
 
205
  "json_float32": {
206
  "per_file": {
207
  "tests/gold.jsonl": {
208
+ "examples": 225,
209
+ "spans": 197,
210
+ "precision": 0.949748743718593,
211
+ "recall": 0.9593908629441624,
212
+ "f1": 0.9545454545454546,
213
+ "value_acc": 0.9543147208121827,
214
  "categories": {
215
  "digits": {
216
+ "spans": 55,
217
+ "value_acc": 0.9454545454545454,
218
+ "exact_match_rate": 0.9454545454545454
219
  },
220
  "words": {
221
+ "spans": 142,
222
+ "value_acc": 0.9577464788732394,
223
+ "exact_match_rate": 0.9647887323943662
224
  },
225
  "prefix": {
226
  "spans": 52,
227
  "value_acc": 0.9423076923076923,
228
+ "exact_match_rate": 0.9615384615384616
229
  },
230
  "range": {
231
+ "spans": 15,
232
+ "value_acc": 0.8666666666666667,
233
+ "exact_match_rate": 0.8666666666666667
234
  },
235
  "currency": {
236
  "spans": 28,
 
238
  "exact_match_rate": 1.0
239
  },
240
  "multi_unit": {
241
+ "spans": 15,
242
+ "value_acc": 0.9333333333333333,
243
+ "exact_match_rate": 0.9333333333333333
244
  },
245
  "symbol_unit": {
246
  "spans": 19,
 
248
  "exact_match_rate": 1.0
249
  },
250
  "long": {
251
+ "spans": 11,
252
+ "value_acc": 0.7272727272727273,
253
+ "exact_match_rate": 0.7272727272727273
254
  }
255
  },
256
  "negatives": {
257
+ "examples": 37,
258
  "false_positives": 0,
259
  "fp_rate": 0.0
260
  }
261
  },
262
  "tests/gold_deva.jsonl": {
263
+ "examples": 185,
264
+ "spans": 154,
265
+ "precision": 0.9869281045751634,
266
+ "recall": 0.9805194805194806,
267
+ "f1": 0.9837133550488599,
268
+ "value_acc": 0.974025974025974,
269
  "categories": {
270
  "digits": {
271
  "spans": 37,
 
273
  "exact_match_rate": 1.0
274
  },
275
  "words": {
276
+ "spans": 117,
277
+ "value_acc": 0.9658119658119658,
278
+ "exact_match_rate": 0.9743589743589743
279
  },
280
  "prefix": {
281
  "spans": 29,
 
283
  "exact_match_rate": 1.0
284
  },
285
  "range": {
286
+ "spans": 12,
287
+ "value_acc": 0.9166666666666666,
288
  "exact_match_rate": 1.0
289
  },
290
  "currency": {
291
  "spans": 20,
292
+ "value_acc": 1.0,
293
+ "exact_match_rate": 1.0
294
  },
295
  "multi_unit": {
296
+ "spans": 9,
297
+ "value_acc": 0.8888888888888888,
298
  "exact_match_rate": 1.0
299
  },
300
  "symbol_unit": {
 
306
  "spans": 2,
307
  "value_acc": 1.0,
308
  "exact_match_rate": 1.0
309
+ },
310
+ "long": {
311
+ "spans": 1,
312
+ "value_acc": 0.0,
313
+ "exact_match_rate": 1.0
314
  }
315
  },
316
  "negatives": {
 
321
  }
322
  },
323
  "combined": {
324
+ "examples": 410,
325
+ "spans": 351,
326
+ "precision": 0.9659090909090909,
327
+ "recall": 0.9686609686609686,
328
+ "f1": 0.9672830725462305,
329
+ "value_acc": 0.9629629629629629,
330
  "miss_summary": {
331
+ "missed_span": 1,
332
  "spurious_span": 0,
333
+ "wrong_value": 2,
334
+ "wrong_boundary": 10,
335
  "total": 13
336
  },
337
  "categories": {
338
  "digits": {
339
+ "spans": 92,
340
+ "value_acc": 0.967391304347826,
341
+ "exact_match_rate": 0.967391304347826
342
  },
343
  "words": {
344
+ "spans": 259,
345
+ "value_acc": 0.9613899613899614,
346
+ "exact_match_rate": 0.9691119691119691
347
  },
348
  "prefix": {
349
  "spans": 81,
350
  "value_acc": 0.9629629629629629,
351
+ "exact_match_rate": 0.9753086419753086
352
  },
353
  "range": {
354
+ "spans": 27,
355
+ "value_acc": 0.8888888888888888,
356
+ "exact_match_rate": 0.9259259259259259
357
  },
358
  "currency": {
359
  "spans": 48,
 
 
 
 
 
360
  "value_acc": 1.0,
361
  "exact_match_rate": 1.0
362
  },
363
+ "multi_unit": {
364
+ "spans": 24,
365
+ "value_acc": 0.9166666666666666,
366
+ "exact_match_rate": 0.9583333333333334
367
+ },
368
  "symbol_unit": {
369
  "spans": 33,
370
  "value_acc": 1.0,
 
376
  "exact_match_rate": 1.0
377
  },
378
  "long": {
379
+ "spans": 12,
380
+ "value_acc": 0.6666666666666666,
381
+ "exact_match_rate": 0.75
382
  }
383
  },
384
  "negatives": {
385
+ "examples": 75,
386
  "false_positives": 0,
387
  "fp_rate": 0.0
388
  }
 
391
  "json_int8": {
392
  "per_file": {
393
  "tests/gold.jsonl": {
394
+ "examples": 225,
395
+ "spans": 197,
396
+ "precision": 0.949748743718593,
397
+ "recall": 0.9593908629441624,
398
+ "f1": 0.9545454545454546,
399
+ "value_acc": 0.9543147208121827,
400
  "categories": {
401
  "digits": {
402
+ "spans": 55,
403
+ "value_acc": 0.9454545454545454,
404
+ "exact_match_rate": 0.9454545454545454
405
  },
406
  "words": {
407
+ "spans": 142,
408
+ "value_acc": 0.9577464788732394,
409
+ "exact_match_rate": 0.9647887323943662
410
  },
411
  "prefix": {
412
  "spans": 52,
413
  "value_acc": 0.9423076923076923,
414
+ "exact_match_rate": 0.9615384615384616
415
  },
416
  "range": {
417
+ "spans": 15,
418
+ "value_acc": 0.8666666666666667,
419
+ "exact_match_rate": 0.8666666666666667
420
  },
421
  "currency": {
422
  "spans": 28,
 
424
  "exact_match_rate": 1.0
425
  },
426
  "multi_unit": {
427
+ "spans": 15,
428
+ "value_acc": 0.9333333333333333,
429
+ "exact_match_rate": 0.9333333333333333
430
  },
431
  "symbol_unit": {
432
  "spans": 19,
 
434
  "exact_match_rate": 1.0
435
  },
436
  "long": {
437
+ "spans": 11,
438
+ "value_acc": 0.7272727272727273,
439
+ "exact_match_rate": 0.7272727272727273
440
  }
441
  },
442
  "negatives": {
443
+ "examples": 37,
444
  "false_positives": 0,
445
  "fp_rate": 0.0
446
  }
447
  },
448
  "tests/gold_deva.jsonl": {
449
+ "examples": 185,
450
+ "spans": 154,
451
+ "precision": 0.9805194805194806,
452
+ "recall": 0.9805194805194806,
453
+ "f1": 0.9805194805194806,
454
+ "value_acc": 0.974025974025974,
455
  "categories": {
456
  "digits": {
457
  "spans": 37,
 
459
  "exact_match_rate": 1.0
460
  },
461
  "words": {
462
+ "spans": 117,
463
+ "value_acc": 0.9658119658119658,
464
+ "exact_match_rate": 0.9743589743589743
465
  },
466
  "prefix": {
467
  "spans": 29,
 
469
  "exact_match_rate": 1.0
470
  },
471
  "range": {
472
+ "spans": 12,
473
+ "value_acc": 0.9166666666666666,
474
  "exact_match_rate": 1.0
475
  },
476
  "currency": {
 
479
  "exact_match_rate": 1.0
480
  },
481
  "multi_unit": {
482
+ "spans": 9,
483
+ "value_acc": 0.8888888888888888,
484
  "exact_match_rate": 1.0
485
  },
486
  "symbol_unit": {
 
492
  "spans": 2,
493
  "value_acc": 1.0,
494
  "exact_match_rate": 1.0
495
+ },
496
+ "long": {
497
+ "spans": 1,
498
+ "value_acc": 0.0,
499
+ "exact_match_rate": 1.0
500
  }
501
  },
502
  "negatives": {
 
507
  }
508
  },
509
  "combined": {
510
+ "examples": 410,
511
+ "spans": 351,
512
+ "precision": 0.9631728045325779,
513
+ "recall": 0.9686609686609686,
514
+ "f1": 0.9659090909090909,
515
+ "value_acc": 0.9629629629629629,
516
  "miss_summary": {
517
+ "missed_span": 1,
518
+ "spurious_span": 1,
519
+ "wrong_value": 2,
520
+ "wrong_boundary": 10,
521
+ "total": 14
522
  },
523
  "categories": {
524
  "digits": {
525
+ "spans": 92,
526
+ "value_acc": 0.967391304347826,
527
+ "exact_match_rate": 0.967391304347826
528
  },
529
  "words": {
530
+ "spans": 259,
531
+ "value_acc": 0.9613899613899614,
532
+ "exact_match_rate": 0.9691119691119691
533
  },
534
  "prefix": {
535
  "spans": 81,
536
  "value_acc": 0.9629629629629629,
537
+ "exact_match_rate": 0.9753086419753086
538
  },
539
  "range": {
540
+ "spans": 27,
541
+ "value_acc": 0.8888888888888888,
542
+ "exact_match_rate": 0.9259259259259259
543
  },
544
  "currency": {
545
  "spans": 48,
 
547
  "exact_match_rate": 1.0
548
  },
549
  "multi_unit": {
550
+ "spans": 24,
551
+ "value_acc": 0.9166666666666666,
552
+ "exact_match_rate": 0.9583333333333334
553
  },
554
  "symbol_unit": {
555
  "spans": 33,
 
562
  "exact_match_rate": 1.0
563
  },
564
  "long": {
565
+ "spans": 12,
566
+ "value_acc": 0.6666666666666666,
567
+ "exact_match_rate": 0.75
568
  }
569
  },
570
  "negatives": {
571
+ "examples": 75,
572
  "false_positives": 0,
573
  "fp_rate": 0.0
574
  }
matrix.md CHANGED
@@ -1,7 +1,10 @@
1
- | arch | channels | seed | val_value_acc | gold_torch_value_acc | gold_json_float32_value_acc | gold_int8_value_acc | gold_int8_f1 | winner |
2
  |---|---|---|---|---|---|---|---|---|
3
- | v2 | 48 | 0 | 0.9217 | 0.9594 | 0.9594 | 0.9594 | 0.9667 | |
4
- | v2 | 48 | 1 | 0.9227 | 0.9565 | 0.9565 | 0.9565 | 0.9651 | |
5
- | v2 | 48 | 2 | 0.9190 | 0.9623 | 0.9623 | 0.9652 | 0.9680 | <-- winner |
 
6
 
 
 
7
  winning config: **v2:48**
 
1
+ | arch | channels | seed | data | val_value_acc | gold_int8_hi_latn | gold_int8_hi_deva | gold_int8_f1 | winner |
2
  |---|---|---|---|---|---|---|---|---|
3
+ | v2 | 48 | 2 | glued | 0.9243 | 0.9442 | 0.9545 | 0.9656 | |
4
+ | v2 | 48 | 2 | no glue (SANKHYA_JOIN_WORDS_P=0) | 0.9158 | 0.9340 | 0.9545 | — | |
5
+ | v2 | 48 | 0 | glued | 0.9242 | 0.9543 | 0.9740 | 0.9805 | <-- winner (shipped 0.3.4) |
6
+ | (0.3.3 shipped, Kaggle seed 2) | 48 | 2 | pre-glue data | 0.9190 | 0.9492 | 0.9740 | 0.9740 | |
7
 
8
+ Gold = 410 examples (225 Hinglish + 185 Devanagari, incl. the 8 lexicon-gap cases added in 0.3.4).
9
+ All 0.3.4 runs trained locally on CPU (~75 s/epoch, 20 epochs, 200k synthetic + LLM corpora at 20%); no Kaggle quota used.
10
  winning config: **v2:48**
sankhya.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e9533a380b332971896c984d45869693edeac6d1c7b1ca56875a471c8ee7915c
3
  size 161024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:37368cbc1b27d32a27b359257c78cfae11230c56de6ecfaaa58978928111c5b9
3
  size 161024
sankhya.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:946d02681dc80d0dad0a5735611e4060cd286a5acc1d0ebd420e584fd8294bea
3
- size 168310
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:283ca23493a081078ac6dd23a617470c44d6362325073202d0dc202944272eb8
3
+ size 167990
sankhya.weights.int8.json CHANGED
The diff for this file is too large to render. See raw diff
 
sankhya.weights.json CHANGED
The diff for this file is too large to render. See raw diff