Hugging Face's logo

Spaces:
ronitraj
/
QuantumScribe
Sleeping

App Files Community
ronitraj commited on
Commit
dc54205
·
verified ·
1 Parent(s): 921ae14

Upload data/sft_dataset_analysis.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. data/sft_dataset_analysis.json +353 -0
data/sft_dataset_analysis.json ADDED
@@ -0,0 +1,353 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "eval_windows": {
3
+ "100": {
4
+ "format_pct": 100.0,
5
+ "level_counts": {
6
+ "L1_warmup": 40,
7
+ "L2_target": 50,
8
+ "L3_stretch": 10
9
+ },
10
+ "nonempty_pct": 69.0
11
+ },
12
+ "30": {
13
+ "format_pct": 100.0,
14
+ "level_counts": {
15
+ "L1_warmup": 10,
16
+ "L2_target": 16,
17
+ "L3_stretch": 4
18
+ },
19
+ "nonempty_pct": 70.0
20
+ },
21
+ "50": {
22
+ "format_pct": 100.0,
23
+ "level_counts": {
24
+ "L1_warmup": 20,
25
+ "L2_target": 24,
26
+ "L3_stretch": 6
27
+ },
28
+ "nonempty_pct": 64.0
29
+ }
30
+ },
31
+ "overlap": {
32
+ "prompt_overlap_count": 38,
33
+ "prompt_overlap_val_pct": 70.37037037037037,
34
+ "record_overlap_count": 38,
35
+ "record_overlap_val_pct": 70.37037037037037
36
+ },
37
+ "train": {
38
+ "completion_len": {
39
+ "max": 37,
40
+ "median": 24.0,
41
+ "min": 23,
42
+ "p25": 23.0,
43
+ "p75": 24.0
44
+ },
45
+ "duplicate_prompt_rows": 2433,
46
+ "duplicate_record_rows": 2433,
47
+ "format_ok": 3000,
48
+ "format_pct": 100.0,
49
+ "had_syndrome": 2092,
50
+ "invalid_count": 0,
51
+ "invalid_examples": [],
52
+ "level_pct": {
53
+ "L1_warmup": 40.0,
54
+ "L2_target": 50.0,
55
+ "L3_stretch": 10.0
56
+ },
57
+ "levels": {
58
+ "L1_warmup": 1200,
59
+ "L2_target": 1500,
60
+ "L3_stretch": 300
61
+ },
62
+ "nonempty": 2070,
63
+ "nonempty_pct": 69.0,
64
+ "per_level": {
65
+ "L1_warmup": {
66
+ "error_count_hist": {
67
+ "0": 600,
68
+ "1": 592,
69
+ "2": 8
70
+ },
71
+ "nonempty": 600,
72
+ "nonempty_pct": 50.0,
73
+ "rows": 1200,
74
+ "syndrome": 602,
75
+ "syndrome_pct": 50.166666666666664
76
+ },
77
+ "L2_target": {
78
+ "error_count_hist": {
79
+ "0": 300,
80
+ "1": 995,
81
+ "2": 183,
82
+ "3": 20,
83
+ "4": 1,
84
+ "5": 1
85
+ },
86
+ "nonempty": 1200,
87
+ "nonempty_pct": 80.0,
88
+ "rows": 1500,
89
+ "syndrome": 1217,
90
+ "syndrome_pct": 81.13333333333334
91
+ },
92
+ "L3_stretch": {
93
+ "error_count_hist": {
94
+ "0": 30,
95
+ "1": 126,
96
+ "2": 83,
97
+ "3": 40,
98
+ "4": 13,
99
+ "5": 7,
100
+ "6": 1
101
+ },
102
+ "nonempty": 270,
103
+ "nonempty_pct": 90.0,
104
+ "rows": 300,
105
+ "syndrome": 273,
106
+ "syndrome_pct": 91.0
107
+ }
108
+ },
109
+ "prompt_len": {
110
+ "max": 1501,
111
+ "median": 1213.0,
112
+ "min": 1114,
113
+ "p25": 1114.0,
114
+ "p75": 1213.0
115
+ },
116
+ "rows": 3000,
117
+ "syndrome_pct": 69.73333333333333,
118
+ "total_error_count_hist": {
119
+ "0": 930,
120
+ "1": 1713,
121
+ "2": 274,
122
+ "3": 60,
123
+ "4": 14,
124
+ "5": 8,
125
+ "6": 1
126
+ },
127
+ "total_qubits": {
128
+ "0": 362,
129
+ "1": 192,
130
+ "2": 120,
131
+ "3": 574,
132
+ "4": 608,
133
+ "5": 145,
134
+ "6": 29,
135
+ "7": 181,
136
+ "8": 17,
137
+ "9": 2,
138
+ "10": 18,
139
+ "11": 30,
140
+ "12": 30,
141
+ "13": 29,
142
+ "14": 22,
143
+ "15": 39,
144
+ "16": 34,
145
+ "17": 26,
146
+ "18": 33,
147
+ "19": 9,
148
+ "21": 28,
149
+ "22": 1,
150
+ "23": 14
151
+ },
152
+ "unique_completions": 219,
153
+ "unique_prompts": 567,
154
+ "x_count_hist": {
155
+ "0": 1479,
156
+ "1": 1435,
157
+ "2": 71,
158
+ "3": 14,
159
+ "4": 1
160
+ },
161
+ "x_qubits": {
162
+ "0": 212,
163
+ "2": 101,
164
+ "3": 394,
165
+ "4": 578,
166
+ "5": 124,
167
+ "6": 22,
168
+ "7": 7,
169
+ "8": 15,
170
+ "9": 2,
171
+ "10": 17,
172
+ "11": 10,
173
+ "12": 22,
174
+ "13": 8,
175
+ "14": 22,
176
+ "15": 14,
177
+ "16": 32,
178
+ "17": 6,
179
+ "18": 28,
180
+ "19": 9
181
+ },
182
+ "z_count_hist": {
183
+ "0": 2165,
184
+ "1": 759,
185
+ "2": 68,
186
+ "3": 7,
187
+ "4": 1
188
+ },
189
+ "z_qubits": {
190
+ "0": 150,
191
+ "1": 192,
192
+ "2": 19,
193
+ "3": 180,
194
+ "4": 30,
195
+ "5": 21,
196
+ "6": 7,
197
+ "7": 174,
198
+ "8": 2,
199
+ "10": 1,
200
+ "11": 20,
201
+ "12": 8,
202
+ "13": 21,
203
+ "15": 25,
204
+ "16": 2,
205
+ "17": 20,
206
+ "18": 5,
207
+ "21": 28,
208
+ "22": 1,
209
+ "23": 14
210
+ }
211
+ },
212
+ "val": {
213
+ "completion_len": {
214
+ "max": 36,
215
+ "median": 24.0,
216
+ "min": 23,
217
+ "p25": 23.0,
218
+ "p75": 24.0
219
+ },
220
+ "duplicate_prompt_rows": 46,
221
+ "duplicate_record_rows": 46,
222
+ "format_ok": 100,
223
+ "format_pct": 100.0,
224
+ "had_syndrome": 71,
225
+ "invalid_count": 0,
226
+ "invalid_examples": [],
227
+ "level_pct": {
228
+ "L1_warmup": 40.0,
229
+ "L2_target": 50.0,
230
+ "L3_stretch": 10.0
231
+ },
232
+ "levels": {
233
+ "L1_warmup": 40,
234
+ "L2_target": 50,
235
+ "L3_stretch": 10
236
+ },
237
+ "nonempty": 69,
238
+ "nonempty_pct": 69.0,
239
+ "per_level": {
240
+ "L1_warmup": {
241
+ "error_count_hist": {
242
+ "0": 20,
243
+ "1": 20
244
+ },
245
+ "nonempty": 20,
246
+ "nonempty_pct": 50.0,
247
+ "rows": 40,
248
+ "syndrome": 21,
249
+ "syndrome_pct": 52.5
250
+ },
251
+ "L2_target": {
252
+ "error_count_hist": {
253
+ "0": 10,
254
+ "1": 32,
255
+ "2": 7,
256
+ "3": 1
257
+ },
258
+ "nonempty": 40,
259
+ "nonempty_pct": 80.0,
260
+ "rows": 50,
261
+ "syndrome": 40,
262
+ "syndrome_pct": 80.0
263
+ },
264
+ "L3_stretch": {
265
+ "error_count_hist": {
266
+ "0": 1,
267
+ "1": 3,
268
+ "2": 3,
269
+ "3": 1,
270
+ "5": 2
271
+ },
272
+ "nonempty": 9,
273
+ "nonempty_pct": 90.0,
274
+ "rows": 10,
275
+ "syndrome": 10,
276
+ "syndrome_pct": 100.0
277
+ }
278
+ },
279
+ "prompt_len": {
280
+ "max": 1501,
281
+ "median": 1213.0,
282
+ "min": 1114,
283
+ "p25": 1114.0,
284
+ "p75": 1213.0
285
+ },
286
+ "rows": 100,
287
+ "syndrome_pct": 71.0,
288
+ "total_error_count_hist": {
289
+ "0": 31,
290
+ "1": 55,
291
+ "2": 10,
292
+ "3": 2,
293
+ "5": 2
294
+ },
295
+ "total_qubits": {
296
+ "0": 17,
297
+ "1": 5,
298
+ "2": 7,
299
+ "3": 18,
300
+ "4": 22,
301
+ "5": 6,
302
+ "7": 3,
303
+ "8": 1,
304
+ "11": 2,
305
+ "12": 1,
306
+ "13": 1,
307
+ "14": 2,
308
+ "15": 1,
309
+ "16": 1,
310
+ "18": 2,
311
+ "23": 2
312
+ },
313
+ "unique_completions": 25,
314
+ "unique_prompts": 54,
315
+ "x_count_hist": {
316
+ "0": 45,
317
+ "1": 51,
318
+ "2": 3,
319
+ "3": 1
320
+ },
321
+ "x_qubits": {
322
+ "0": 9,
323
+ "2": 6,
324
+ "3": 11,
325
+ "4": 22,
326
+ "5": 5,
327
+ "8": 1,
328
+ "14": 2,
329
+ "15": 1,
330
+ "16": 1,
331
+ "18": 2
332
+ },
333
+ "z_count_hist": {
334
+ "0": 75,
335
+ "1": 20,
336
+ "2": 4,
337
+ "3": 1
338
+ },
339
+ "z_qubits": {
340
+ "0": 8,
341
+ "1": 5,
342
+ "2": 1,
343
+ "3": 7,
344
+ "5": 1,
345
+ "7": 3,
346
+ "11": 2,
347
+ "12": 1,
348
+ "13": 1,
349
+ "23": 2
350
+ }
351
+ },
352
+ "verdict": "SFT_READY_WITH_MINOR_DUPLICATE_CAVEAT"
353
+ }