Hugging Face's logo

Datasets:
qingxin1999
/
pepforge-fig-data

Tasks:
Text Generation
Languages:
English
Size:
1K<n<10K
Tags:
peptide
HELM
chemistry
drug-discovery
antimicrobial-peptide
evaluation
License:
Dataset card Files
xet
 Community
qingxin1999 commited on
Commit
cac08fd
·
verified ·
1 Parent(s): 4faad18

Delete batch 4 (73 files)

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run1/infilling_report.json +0 -21
  2. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run1/results.jsonl +0 -3
  3. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run2/eval_report.json +0 -418
  4. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run2/eval_summary.txt +0 -57
  5. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run2/infilling_report.json +0 -21
  6. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run2/results.jsonl +0 -3
  7. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run3/eval_report.json +0 -414
  8. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run3/eval_summary.txt +0 -56
  9. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run3/infilling_report.json +0 -21
  10. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run3/results.jsonl +0 -3
  11. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run4/eval_report.json +0 -418
  12. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run4/eval_summary.txt +0 -57
  13. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run4/infilling_report.json +0 -21
  14. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run4/results.jsonl +0 -3
  15. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run5/eval_report.json +0 -414
  16. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run5/eval_summary.txt +0 -56
  17. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run5/infilling_report.json +0 -21
  18. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run5/results.jsonl +0 -3
  19. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/comparison_summary_run1.json +0 -67
  20. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/comparison_summary_run2.json +0 -67
  21. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/comparison_summary_run3.json +0 -67
  22. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/comparison_summary_run4.json +0 -67
  23. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/comparison_summary_run5.json +0 -67
  24. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run1/eval_report.json +0 -414
  25. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run1/eval_summary.txt +0 -56
  26. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run1/infilling_report.json +0 -21
  27. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run1/results.jsonl +0 -3
  28. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run2/eval_report.json +0 -418
  29. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run2/eval_summary.txt +0 -57
  30. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run2/infilling_report.json +0 -21
  31. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run2/results.jsonl +0 -3
  32. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run3/eval_report.json +0 -418
  33. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run3/eval_summary.txt +0 -57
  34. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run3/infilling_report.json +0 -21
  35. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run3/results.jsonl +0 -3
  36. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run4/eval_report.json +0 -417
  37. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run4/eval_summary.txt +0 -57
  38. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run4/infilling_report.json +0 -21
  39. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run4/results.jsonl +0 -3
  40. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run5/eval_report.json +0 -418
  41. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run5/eval_summary.txt +0 -57
  42. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run5/infilling_report.json +0 -21
  43. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run5/results.jsonl +0 -3
  44. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run1/eval_report.json +0 -413
  45. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run1/eval_summary.txt +0 -56
  46. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run1/infilling_report.json +0 -21
  47. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run1/results.jsonl +0 -3
  48. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run2/eval_report.json +0 -413
  49. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run2/eval_summary.txt +0 -56
  50. Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run2/infilling_report.json +0 -21
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run1/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.8687,
4
- "valid_count": 8687,
5
- "monomer_accuracy": {
6
- "mean": 0.102699750870045,
7
- "std": 0.21467870024616584,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7473,
18
- "pct_above_50": 0.057,
19
- "pct_perfect": 0.0112
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run1/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:958a5147cb393e0610b9de098603759c707ebe9d7899b6e28a0041dc9e4f2daf
3
- size 11467572
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run2/eval_report.json DELETED
@@ -1,418 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:55.291675",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 8733,
7
- "num_unique": 8694,
8
- "num_novel": 5213
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 8733,
13
- "rate": 0.8733
14
- },
15
- "uniqueness": {
16
- "num_valid": 8733,
17
- "num_unique": 8694,
18
- "rate": 0.9955341806939196
19
- },
20
- "novelty": {
21
- "num_unique": 8694,
22
- "num_novel": 5213,
23
- "rate": 0.5996089256958822
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.8838,
27
- "num_molecules": 8694,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.3964,
32
- "median": 0.395,
33
- "std": 0.1828,
34
- "percentiles": {
35
- "5": 0.0,
36
- "25": 0.2577,
37
- "50": 0.395,
38
- "75": 0.5615,
39
- "95": 0.6529
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 260,
97
- 3,
98
- 8,
99
- 13,
100
- 18,
101
- 14,
102
- 31,
103
- 37,
104
- 71,
105
- 108,
106
- 185,
107
- 271,
108
- 280,
109
- 274,
110
- 259,
111
- 197,
112
- 143,
113
- 135,
114
- 120,
115
- 100,
116
- 113,
117
- 131,
118
- 142,
119
- 120,
120
- 149,
121
- 158,
122
- 180,
123
- 210,
124
- 225,
125
- 230,
126
- 237,
127
- 206,
128
- 160,
129
- 97,
130
- 63,
131
- 31,
132
- 13,
133
- 4,
134
- 1,
135
- 3,
136
- 0,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.8868,
156
- "nn_self_distance": {
157
- "mean": 0.3757,
158
- "median": 0.3442,
159
- "std": 0.1515,
160
- "percentiles": {
161
- "5": 0.1523,
162
- "25": 0.2656,
163
- "50": 0.3442,
164
- "75": 0.5083,
165
- "95": 0.6255
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 1,
223
- 22,
224
- 39,
225
- 29,
226
- 26,
227
- 44,
228
- 53,
229
- 59,
230
- 96,
231
- 114,
232
- 188,
233
- 216,
234
- 298,
235
- 305,
236
- 314,
237
- 345,
238
- 288,
239
- 235,
240
- 228,
241
- 182,
242
- 120,
243
- 143,
244
- 115,
245
- 117,
246
- 121,
247
- 139,
248
- 142,
249
- 192,
250
- 169,
251
- 183,
252
- 194,
253
- 117,
254
- 76,
255
- 40,
256
- 30,
257
- 8,
258
- 8,
259
- 2,
260
- 2,
261
- 0,
262
- 0,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.1202,
282
- "peptide_length_mean": 6.78,
283
- "peptide_length_dist": {
284
- "2": 135,
285
- "3": 681,
286
- "4": 4405,
287
- "5": 383,
288
- "6": 403,
289
- "7": 349,
290
- "8": 320,
291
- "9": 430,
292
- "10": 276,
293
- "11": 162,
294
- "12": 138,
295
- "13": 129,
296
- "14": 91,
297
- "15": 112,
298
- "16": 72,
299
- "17": 67,
300
- "18": 62,
301
- "19": 40,
302
- "20": 108,
303
- "21": 53,
304
- "22": 27,
305
- "23": 27,
306
- "24": 30,
307
- "25": 233
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1634,
315
- "valid": 888,
316
- "generation_rate": 0.1634,
317
- "pass_rate": 0.5435
318
- },
319
- "Bz_Stapled": {
320
- "generated": 4,
321
- "valid": 1,
322
- "generation_rate": 0.0004,
323
- "pass_rate": 0.25
324
- },
325
- "Disulfide": {
326
- "generated": 231,
327
- "valid": 128,
328
- "generation_rate": 0.0231,
329
- "pass_rate": 0.5541
330
- },
331
- "Ester": {
332
- "generated": 300,
333
- "valid": 126,
334
- "generation_rate": 0.03,
335
- "pass_rate": 0.42
336
- },
337
- "Lanthionin": {
338
- "generated": 13,
339
- "valid": 5,
340
- "generation_rate": 0.0013,
341
- "pass_rate": 0.3846
342
- },
343
- "PG_Amide": {
344
- "generated": 302,
345
- "valid": 173,
346
- "generation_rate": 0.0302,
347
- "pass_rate": 0.5728
348
- },
349
- "Sulfanilamide": {
350
- "generated": 7,
351
- "valid": 5,
352
- "generation_rate": 0.0007,
353
- "pass_rate": 0.7143
354
- },
355
- "Trp-Cys": {
356
- "generated": 4,
357
- "valid": 3,
358
- "generation_rate": 0.0004,
359
- "pass_rate": 0.75
360
- },
361
- "head2tail ring": {
362
- "generated": 401,
363
- "valid": 340,
364
- "generation_rate": 0.0401,
365
- "pass_rate": 0.8479
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2396,
372
- "has_connection_rate": 0.2396,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1912,
376
- "fraction": 0.798
377
- },
378
- "2": {
379
- "count": 469,
380
- "fraction": 0.1957
381
- },
382
- "3": {
383
- "count": 14,
384
- "fraction": 0.0058
385
- },
386
- "4+": {
387
- "count": 1,
388
- "fraction": 0.0004
389
- }
390
- }
391
- },
392
- "per_conn_count_metrics": {
393
- "0": {
394
- "count": 7604,
395
- "validity": 0.9527,
396
- "uniqueness": 0.9948,
397
- "novelty": 0.5371
398
- },
399
- "1": {
400
- "count": 1912,
401
- "validity": 0.6857,
402
- "uniqueness": 0.9992,
403
- "novelty": 0.9038
404
- },
405
- "2": {
406
- "count": 469,
407
- "validity": 0.3753,
408
- "uniqueness": 1.0,
409
- "novelty": 0.892
410
- },
411
- "3+": {
412
- "count": 15,
413
- "validity": 0.1333,
414
- "uniqueness": 1.0,
415
- "novelty": 1.0
416
- }
417
- }
418
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run2/eval_summary.txt DELETED
@@ -1,57 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:30:37.675005
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 8733/10000 (87.3%)
9
- Uniqueness: 8694/8733 (99.6%)
10
- Novelty: 5273/8694 (60.7%)
11
-
12
- Diversity: 0.8849 (avg MAP4C Jaccard distance)
13
- Molecules: 8694
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.3983, median=0.3972, std=0.1822
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8860
23
- Self-NN: mean=0.3752, median=0.3433
24
-
25
- Structural Summary:
26
- Multi-block rate: 12.0%
27
- Avg peptide len: 6.8
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1634 888 0.1634 54.4%
33
- head2tail ring 401 340 0.0401 84.8%
34
- PG_Amide 302 173 0.0302 57.3%
35
- Ester 300 126 0.0300 42.0%
36
- Disulfide 231 128 0.0231 55.4%
37
- Lanthionin 13 5 0.0013 38.5%
38
- Sulfanilamide 7 5 0.0007 71.4%
39
- Bz_Stapled 4 1 0.0004 25.0%
40
- Trp-Cys 4 3 0.0004 75.0%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 24.0% (2396/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1912 (79.8%)
46
- 2 connection(s): 469 (19.6%)
47
- 3 connection(s): 14 (0.6%)
48
- 4+ connection(s): 1 (0.0%)
49
-
50
- Per-Connection-Count Metrics:
51
- Bucket Count Validity Uniqueness Novelty
52
- ---------- -------- ---------- ------------ ----------
53
- 0 conn 7604 95.3% 99.5% 54.3%
54
- 1 conn 1912 68.6% 99.9% 91.4%
55
- 2 conn 469 37.5% 100.0% 90.3%
56
- 3+ conn 15 13.3% 100.0% 100.0%
57
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run2/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.8733,
4
- "valid_count": 8733,
5
- "monomer_accuracy": {
6
- "mean": 0.10169111704471999,
7
- "std": 0.21432767434230826,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7526,
18
- "pct_above_50": 0.058,
19
- "pct_perfect": 0.0098
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run2/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:b5d720c3fb29094ca9ceebbcfd7a965099d028b81660ba29ec34e05839dac422
3
- size 11467694
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run3/eval_report.json DELETED
@@ -1,414 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:55.292114",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 8684,
7
- "num_unique": 8648,
8
- "num_novel": 5149
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 8684,
13
- "rate": 0.8684
14
- },
15
- "uniqueness": {
16
- "num_valid": 8684,
17
- "num_unique": 8648,
18
- "rate": 0.9958544449562413
19
- },
20
- "novelty": {
21
- "num_unique": 8648,
22
- "num_novel": 5149,
23
- "rate": 0.5953977798334875
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.8838,
27
- "num_molecules": 8648,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.3952,
32
- "median": 0.3882,
33
- "std": 0.1826,
34
- "percentiles": {
35
- "5": 0.0,
36
- "25": 0.2578,
37
- "50": 0.3882,
38
- "75": 0.5607,
39
- "95": 0.6548
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 269,
97
- 1,
98
- 9,
99
- 10,
100
- 12,
101
- 24,
102
- 17,
103
- 42,
104
- 62,
105
- 97,
106
- 179,
107
- 293,
108
- 262,
109
- 300,
110
- 234,
111
- 218,
112
- 164,
113
- 115,
114
- 131,
115
- 136,
116
- 116,
117
- 113,
118
- 116,
119
- 137,
120
- 158,
121
- 159,
122
- 177,
123
- 193,
124
- 213,
125
- 229,
126
- 226,
127
- 202,
128
- 172,
129
- 107,
130
- 56,
131
- 27,
132
- 18,
133
- 3,
134
- 2,
135
- 0,
136
- 1,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.8855,
156
- "nn_self_distance": {
157
- "mean": 0.3733,
158
- "median": 0.3398,
159
- "std": 0.1493,
160
- "percentiles": {
161
- "5": 0.1548,
162
- "25": 0.2646,
163
- "50": 0.3398,
164
- "75": 0.5015,
165
- "95": 0.623
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 6,
223
- 19,
224
- 35,
225
- 30,
226
- 38,
227
- 36,
228
- 36,
229
- 70,
230
- 74,
231
- 102,
232
- 159,
233
- 257,
234
- 302,
235
- 366,
236
- 346,
237
- 325,
238
- 304,
239
- 226,
240
- 182,
241
- 193,
242
- 146,
243
- 131,
244
- 124,
245
- 110,
246
- 118,
247
- 163,
248
- 140,
249
- 150,
250
- 172,
251
- 185,
252
- 184,
253
- 115,
254
- 74,
255
- 42,
256
- 21,
257
- 11,
258
- 4,
259
- 4,
260
- 0,
261
- 0,
262
- 0,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.1156,
282
- "peptide_length_mean": 6.77,
283
- "peptide_length_dist": {
284
- "2": 126,
285
- "3": 688,
286
- "4": 4378,
287
- "5": 380,
288
- "6": 405,
289
- "7": 353,
290
- "8": 314,
291
- "9": 411,
292
- "10": 281,
293
- "11": 162,
294
- "12": 141,
295
- "13": 130,
296
- "14": 89,
297
- "15": 118,
298
- "16": 68,
299
- "17": 68,
300
- "18": 61,
301
- "19": 41,
302
- "20": 103,
303
- "21": 62,
304
- "22": 25,
305
- "23": 28,
306
- "24": 31,
307
- "25": 221
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1637,
315
- "valid": 845,
316
- "generation_rate": 0.1637,
317
- "pass_rate": 0.5162
318
- },
319
- "Bz_Stapled": {
320
- "generated": 3,
321
- "valid": 1,
322
- "generation_rate": 0.0003,
323
- "pass_rate": 0.3333
324
- },
325
- "Disulfide": {
326
- "generated": 231,
327
- "valid": 85,
328
- "generation_rate": 0.0231,
329
- "pass_rate": 0.368
330
- },
331
- "Ester": {
332
- "generated": 310,
333
- "valid": 124,
334
- "generation_rate": 0.031,
335
- "pass_rate": 0.4
336
- },
337
- "Lanthionin": {
338
- "generated": 9,
339
- "valid": 1,
340
- "generation_rate": 0.0009,
341
- "pass_rate": 0.1111
342
- },
343
- "PG_Amide": {
344
- "generated": 304,
345
- "valid": 178,
346
- "generation_rate": 0.0304,
347
- "pass_rate": 0.5855
348
- },
349
- "Sulfanilamide": {
350
- "generated": 8,
351
- "valid": 4,
352
- "generation_rate": 0.0008,
353
- "pass_rate": 0.5
354
- },
355
- "Trp-Cys": {
356
- "generated": 7,
357
- "valid": 6,
358
- "generation_rate": 0.0007,
359
- "pass_rate": 0.8571
360
- },
361
- "head2tail ring": {
362
- "generated": 399,
363
- "valid": 349,
364
- "generation_rate": 0.0399,
365
- "pass_rate": 0.8747
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2412,
372
- "has_connection_rate": 0.2412,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1936,
376
- "fraction": 0.8027
377
- },
378
- "2": {
379
- "count": 456,
380
- "fraction": 0.1891
381
- },
382
- "3": {
383
- "count": 20,
384
- "fraction": 0.0083
385
- }
386
- }
387
- },
388
- "per_conn_count_metrics": {
389
- "0": {
390
- "count": 7588,
391
- "validity": 0.951,
392
- "uniqueness": 0.9951,
393
- "novelty": 0.535
394
- },
395
- "1": {
396
- "count": 1936,
397
- "validity": 0.6942,
398
- "uniqueness": 0.9993,
399
- "novelty": 0.8883
400
- },
401
- "2": {
402
- "count": 456,
403
- "validity": 0.2697,
404
- "uniqueness": 1.0,
405
- "novelty": 0.9187
406
- },
407
- "3+": {
408
- "count": 20,
409
- "validity": 0.05,
410
- "uniqueness": 1.0,
411
- "novelty": 1.0
412
- }
413
- }
414
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run3/eval_summary.txt DELETED
@@ -1,56 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:31:58.864888
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 8684/10000 (86.8%)
9
- Uniqueness: 8648/8684 (99.6%)
10
- Novelty: 5208/8648 (60.2%)
11
-
12
- Diversity: 0.8832 (avg MAP4C Jaccard distance)
13
- Molecules: 8648
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.3973, median=0.3918, std=0.1855
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8846
23
- Self-NN: mean=0.3737, median=0.3428
24
-
25
- Structural Summary:
26
- Multi-block rate: 11.6%
27
- Avg peptide len: 6.8
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1637 845 0.1637 51.6%
33
- head2tail ring 399 349 0.0399 87.5%
34
- Ester 310 124 0.0310 40.0%
35
- PG_Amide 304 178 0.0304 58.6%
36
- Disulfide 231 85 0.0231 36.8%
37
- Lanthionin 9 1 0.0009 11.1%
38
- Sulfanilamide 8 4 0.0008 50.0%
39
- Trp-Cys 7 6 0.0007 85.7%
40
- Bz_Stapled 3 1 0.0003 33.3%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 24.1% (2412/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1936 (80.3%)
46
- 2 connection(s): 456 (18.9%)
47
- 3 connection(s): 20 (0.8%)
48
-
49
- Per-Connection-Count Metrics:
50
- Bucket Count Validity Uniqueness Novelty
51
- ---------- -------- ---------- ------------ ----------
52
- 0 conn 7588 95.1% 99.5% 54.1%
53
- 1 conn 1936 69.4% 99.9% 90.2%
54
- 2 conn 456 27.0% 100.0% 91.9%
55
- 3+ conn 20 5.0% 100.0% 100.0%
56
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run3/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.8684,
4
- "valid_count": 8684,
5
- "monomer_accuracy": {
6
- "mean": 0.10061371680606974,
7
- "std": 0.21239770258870322,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7518,
18
- "pct_above_50": 0.054,
19
- "pct_perfect": 0.0116
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run3/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:d08db86baa07e21d3f9e741beecb62e926b04aff0794e4b1ac4ca96d97abd452
3
- size 11468177
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run4/eval_report.json DELETED
@@ -1,418 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:49.907373",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 8778,
7
- "num_unique": 8747,
8
- "num_novel": 5229
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 8778,
13
- "rate": 0.8778
14
- },
15
- "uniqueness": {
16
- "num_valid": 8778,
17
- "num_unique": 8747,
18
- "rate": 0.9964684438368648
19
- },
20
- "novelty": {
21
- "num_unique": 8747,
22
- "num_novel": 5229,
23
- "rate": 0.5978049617011547
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.886,
27
- "num_molecules": 8747,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.3994,
32
- "median": 0.3994,
33
- "std": 0.183,
34
- "percentiles": {
35
- "5": 0.0,
36
- "25": 0.2583,
37
- "50": 0.3994,
38
- "75": 0.5615,
39
- "95": 0.6582
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 266,
97
- 2,
98
- 4,
99
- 9,
100
- 17,
101
- 18,
102
- 20,
103
- 32,
104
- 62,
105
- 118,
106
- 176,
107
- 254,
108
- 289,
109
- 244,
110
- 271,
111
- 204,
112
- 169,
113
- 124,
114
- 118,
115
- 105,
116
- 120,
117
- 127,
118
- 122,
119
- 119,
120
- 161,
121
- 177,
122
- 190,
123
- 213,
124
- 201,
125
- 239,
126
- 245,
127
- 185,
128
- 159,
129
- 103,
130
- 62,
131
- 42,
132
- 19,
133
- 11,
134
- 1,
135
- 2,
136
- 0,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.8849,
156
- "nn_self_distance": {
157
- "mean": 0.372,
158
- "median": 0.3394,
159
- "std": 0.1495,
160
- "percentiles": {
161
- "5": 0.1558,
162
- "25": 0.2627,
163
- "50": 0.3394,
164
- "75": 0.4971,
165
- "95": 0.624
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 3,
223
- 17,
224
- 28,
225
- 35,
226
- 38,
227
- 35,
228
- 42,
229
- 63,
230
- 92,
231
- 117,
232
- 172,
233
- 260,
234
- 309,
235
- 315,
236
- 358,
237
- 337,
238
- 296,
239
- 230,
240
- 194,
241
- 164,
242
- 176,
243
- 133,
244
- 131,
245
- 115,
246
- 108,
247
- 128,
248
- 134,
249
- 153,
250
- 193,
251
- 163,
252
- 182,
253
- 116,
254
- 76,
255
- 45,
256
- 15,
257
- 12,
258
- 8,
259
- 3,
260
- 3,
261
- 1,
262
- 0,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.1242,
282
- "peptide_length_mean": 6.77,
283
- "peptide_length_dist": {
284
- "2": 138,
285
- "3": 689,
286
- "4": 4406,
287
- "5": 388,
288
- "6": 409,
289
- "7": 360,
290
- "8": 332,
291
- "9": 405,
292
- "10": 278,
293
- "11": 165,
294
- "12": 143,
295
- "13": 130,
296
- "14": 94,
297
- "15": 124,
298
- "16": 73,
299
- "17": 67,
300
- "18": 63,
301
- "19": 42,
302
- "20": 114,
303
- "21": 54,
304
- "22": 22,
305
- "23": 27,
306
- "24": 28,
307
- "25": 227
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1636,
315
- "valid": 890,
316
- "generation_rate": 0.1636,
317
- "pass_rate": 0.544
318
- },
319
- "Bz_Stapled": {
320
- "generated": 5,
321
- "valid": 0,
322
- "generation_rate": 0.0005,
323
- "pass_rate": 0.0
324
- },
325
- "Disulfide": {
326
- "generated": 249,
327
- "valid": 136,
328
- "generation_rate": 0.0249,
329
- "pass_rate": 0.5462
330
- },
331
- "Ester": {
332
- "generated": 320,
333
- "valid": 147,
334
- "generation_rate": 0.032,
335
- "pass_rate": 0.4594
336
- },
337
- "Lanthionin": {
338
- "generated": 14,
339
- "valid": 4,
340
- "generation_rate": 0.0014,
341
- "pass_rate": 0.2857
342
- },
343
- "PG_Amide": {
344
- "generated": 314,
345
- "valid": 216,
346
- "generation_rate": 0.0314,
347
- "pass_rate": 0.6879
348
- },
349
- "Sulfanilamide": {
350
- "generated": 7,
351
- "valid": 2,
352
- "generation_rate": 0.0007,
353
- "pass_rate": 0.2857
354
- },
355
- "Trp-Cys": {
356
- "generated": 9,
357
- "valid": 5,
358
- "generation_rate": 0.0009,
359
- "pass_rate": 0.5556
360
- },
361
- "head2tail ring": {
362
- "generated": 381,
363
- "valid": 325,
364
- "generation_rate": 0.0381,
365
- "pass_rate": 0.853
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2402,
372
- "has_connection_rate": 0.2402,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1896,
376
- "fraction": 0.7893
377
- },
378
- "2": {
379
- "count": 481,
380
- "fraction": 0.2002
381
- },
382
- "3": {
383
- "count": 23,
384
- "fraction": 0.0096
385
- },
386
- "4+": {
387
- "count": 2,
388
- "fraction": 0.0008
389
- }
390
- }
391
- },
392
- "per_conn_count_metrics": {
393
- "0": {
394
- "count": 7598,
395
- "validity": 0.9545,
396
- "uniqueness": 0.9959,
397
- "novelty": 0.5332
398
- },
399
- "1": {
400
- "count": 1896,
401
- "validity": 0.702,
402
- "uniqueness": 0.9992,
403
- "novelty": 0.909
404
- },
405
- "2": {
406
- "count": 481,
407
- "validity": 0.3971,
408
- "uniqueness": 1.0,
409
- "novelty": 0.8691
410
- },
411
- "3+": {
412
- "count": 25,
413
- "validity": 0.16,
414
- "uniqueness": 1.0,
415
- "novelty": 0.75
416
- }
417
- }
418
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run4/eval_summary.txt DELETED
@@ -1,57 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:31:57.091845
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 8778/10000 (87.8%)
9
- Uniqueness: 8747/8778 (99.6%)
10
- Novelty: 5298/8747 (60.6%)
11
-
12
- Diversity: 0.8856 (avg MAP4C Jaccard distance)
13
- Molecules: 8747
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.4014, median=0.3989, std=0.1817
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8846
23
- Self-NN: mean=0.3737, median=0.3428
24
-
25
- Structural Summary:
26
- Multi-block rate: 12.4%
27
- Avg peptide len: 6.8
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1636 890 0.1636 54.4%
33
- head2tail ring 381 325 0.0381 85.3%
34
- Ester 320 147 0.0320 45.9%
35
- PG_Amide 314 216 0.0314 68.8%
36
- Disulfide 249 136 0.0249 54.6%
37
- Lanthionin 14 4 0.0014 28.6%
38
- Trp-Cys 9 5 0.0009 55.6%
39
- Sulfanilamide 7 2 0.0007 28.6%
40
- Bz_Stapled 5 0 0.0005 0.0%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 24.0% (2402/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1896 (78.9%)
46
- 2 connection(s): 481 (20.0%)
47
- 3 connection(s): 23 (1.0%)
48
- 4+ connection(s): 2 (0.1%)
49
-
50
- Per-Connection-Count Metrics:
51
- Bucket Count Validity Uniqueness Novelty
52
- ---------- -------- ---------- ------------ ----------
53
- 0 conn 7598 95.5% 99.6% 54.0%
54
- 1 conn 1896 70.2% 99.9% 92.4%
55
- 2 conn 481 39.7% 100.0% 87.4%
56
- 3+ conn 25 16.0% 100.0% 75.0%
57
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run4/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.8778,
4
- "valid_count": 8778,
5
- "monomer_accuracy": {
6
- "mean": 0.10346335061996827,
7
- "std": 0.21471265232937553,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7453,
18
- "pct_above_50": 0.0551,
19
- "pct_perfect": 0.0121
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run4/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:32a8070528709db26c6e15ed5708ca55838ac161c85b174a5f5a59b89b14d6be
3
- size 11469061
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run5/eval_report.json DELETED
@@ -1,414 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:49.572833",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 8658,
7
- "num_unique": 8623,
8
- "num_novel": 5069
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 8658,
13
- "rate": 0.8658
14
- },
15
- "uniqueness": {
16
- "num_valid": 8658,
17
- "num_unique": 8623,
18
- "rate": 0.9959574959574959
19
- },
20
- "novelty": {
21
- "num_unique": 8623,
22
- "num_novel": 5069,
23
- "rate": 0.5878464571494839
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.8846,
27
- "num_molecules": 8623,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.3981,
32
- "median": 0.3865,
33
- "std": 0.1793,
34
- "percentiles": {
35
- "5": 0.0952,
36
- "25": 0.2588,
37
- "50": 0.3865,
38
- "75": 0.5625,
39
- "95": 0.6548
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 220,
97
- 4,
98
- 5,
99
- 10,
100
- 12,
101
- 17,
102
- 22,
103
- 46,
104
- 88,
105
- 101,
106
- 192,
107
- 258,
108
- 293,
109
- 275,
110
- 264,
111
- 187,
112
- 186,
113
- 157,
114
- 120,
115
- 111,
116
- 111,
117
- 134,
118
- 122,
119
- 111,
120
- 144,
121
- 148,
122
- 185,
123
- 199,
124
- 210,
125
- 251,
126
- 238,
127
- 208,
128
- 154,
129
- 97,
130
- 80,
131
- 32,
132
- 4,
133
- 3,
134
- 1,
135
- 0,
136
- 0,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.8848,
156
- "nn_self_distance": {
157
- "mean": 0.3726,
158
- "median": 0.342,
159
- "std": 0.148,
160
- "percentiles": {
161
- "5": 0.1582,
162
- "25": 0.2661,
163
- "50": 0.342,
164
- "75": 0.4938,
165
- "95": 0.624
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 3,
223
- 15,
224
- 22,
225
- 41,
226
- 45,
227
- 40,
228
- 40,
229
- 47,
230
- 81,
231
- 132,
232
- 174,
233
- 253,
234
- 253,
235
- 347,
236
- 382,
237
- 312,
238
- 291,
239
- 245,
240
- 212,
241
- 169,
242
- 151,
243
- 139,
244
- 139,
245
- 132,
246
- 119,
247
- 128,
248
- 147,
249
- 150,
250
- 173,
251
- 167,
252
- 171,
253
- 134,
254
- 72,
255
- 45,
256
- 16,
257
- 8,
258
- 4,
259
- 1,
260
- 0,
261
- 0,
262
- 0,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.1142,
282
- "peptide_length_mean": 6.73,
283
- "peptide_length_dist": {
284
- "2": 127,
285
- "3": 696,
286
- "4": 4387,
287
- "5": 376,
288
- "6": 393,
289
- "7": 340,
290
- "8": 318,
291
- "9": 412,
292
- "10": 272,
293
- "11": 163,
294
- "12": 141,
295
- "13": 133,
296
- "14": 92,
297
- "15": 118,
298
- "16": 70,
299
- "17": 65,
300
- "18": 58,
301
- "19": 41,
302
- "20": 110,
303
- "21": 56,
304
- "22": 23,
305
- "23": 25,
306
- "24": 28,
307
- "25": 214
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1646,
315
- "valid": 837,
316
- "generation_rate": 0.1646,
317
- "pass_rate": 0.5085
318
- },
319
- "Bz_Stapled": {
320
- "generated": 4,
321
- "valid": 0,
322
- "generation_rate": 0.0004,
323
- "pass_rate": 0.0
324
- },
325
- "Disulfide": {
326
- "generated": 254,
327
- "valid": 105,
328
- "generation_rate": 0.0254,
329
- "pass_rate": 0.4134
330
- },
331
- "Ester": {
332
- "generated": 309,
333
- "valid": 140,
334
- "generation_rate": 0.0309,
335
- "pass_rate": 0.4531
336
- },
337
- "Lanthionin": {
338
- "generated": 10,
339
- "valid": 2,
340
- "generation_rate": 0.001,
341
- "pass_rate": 0.2
342
- },
343
- "PG_Amide": {
344
- "generated": 297,
345
- "valid": 176,
346
- "generation_rate": 0.0297,
347
- "pass_rate": 0.5926
348
- },
349
- "Sulfanilamide": {
350
- "generated": 10,
351
- "valid": 3,
352
- "generation_rate": 0.001,
353
- "pass_rate": 0.3
354
- },
355
- "Trp-Cys": {
356
- "generated": 5,
357
- "valid": 1,
358
- "generation_rate": 0.0005,
359
- "pass_rate": 0.2
360
- },
361
- "head2tail ring": {
362
- "generated": 348,
363
- "valid": 296,
364
- "generation_rate": 0.0348,
365
- "pass_rate": 0.8506
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2391,
372
- "has_connection_rate": 0.2391,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1916,
376
- "fraction": 0.8013
377
- },
378
- "2": {
379
- "count": 458,
380
- "fraction": 0.1916
381
- },
382
- "3": {
383
- "count": 17,
384
- "fraction": 0.0071
385
- }
386
- }
387
- },
388
- "per_conn_count_metrics": {
389
- "0": {
390
- "count": 7609,
391
- "validity": 0.9505,
392
- "uniqueness": 0.9953,
393
- "novelty": 0.5281
394
- },
395
- "1": {
396
- "count": 1916,
397
- "validity": 0.6748,
398
- "uniqueness": 1.0,
399
- "novelty": 0.8879
400
- },
401
- "2": {
402
- "count": 458,
403
- "validity": 0.2882,
404
- "uniqueness": 0.9924,
405
- "novelty": 0.9084
406
- },
407
- "3+": {
408
- "count": 17,
409
- "validity": 0.0588,
410
- "uniqueness": 1.0,
411
- "novelty": 1.0
412
- }
413
- }
414
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run5/eval_summary.txt DELETED
@@ -1,56 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:31:59.484224
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 8658/10000 (86.6%)
9
- Uniqueness: 8623/8658 (99.6%)
10
- Novelty: 5138/8623 (59.6%)
11
-
12
- Diversity: 0.8847 (avg MAP4C Jaccard distance)
13
- Molecules: 8623
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.4001, median=0.3987, std=0.1829
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8846
23
- Self-NN: mean=0.3737, median=0.3428
24
-
25
- Structural Summary:
26
- Multi-block rate: 11.4%
27
- Avg peptide len: 6.7
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1646 837 0.1646 50.8%
33
- head2tail ring 348 296 0.0348 85.1%
34
- Ester 309 140 0.0309 45.3%
35
- PG_Amide 297 176 0.0297 59.3%
36
- Disulfide 254 105 0.0254 41.3%
37
- Lanthionin 10 2 0.0010 20.0%
38
- Sulfanilamide 10 3 0.0010 30.0%
39
- Trp-Cys 5 1 0.0005 20.0%
40
- Bz_Stapled 4 0 0.0004 0.0%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 23.9% (2391/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1916 (80.1%)
46
- 2 connection(s): 458 (19.2%)
47
- 3 connection(s): 17 (0.7%)
48
-
49
- Per-Connection-Count Metrics:
50
- Bucket Count Validity Uniqueness Novelty
51
- ---------- -------- ---------- ------------ ----------
52
- 0 conn 7609 95.0% 99.5% 53.5%
53
- 1 conn 1916 67.5% 100.0% 90.2%
54
- 2 conn 458 28.8% 99.2% 90.8%
55
- 3+ conn 17 5.9% 100.0% 100.0%
56
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run5/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.8658,
4
- "valid_count": 8658,
5
- "monomer_accuracy": {
6
- "mean": 0.10191266963428727,
7
- "std": 0.21260300891684258,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7471,
18
- "pct_above_50": 0.054,
19
- "pct_perfect": 0.0114
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/GPT/run5/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:41758de0edd2e7041f4a62284b642274834f4a8af3f936a57df6a83b45a77f81
3
- size 11464209
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/comparison_summary_run1.json DELETED
@@ -1,67 +0,0 @@
1
- {
2
- "num_constraints": 10000,
3
- "paired_wins": {
4
- "GPT_wins": 1138,
5
- "BERT_wins": 2408,
6
- "ties": 6454,
7
- "ties_both_zero": 5893
8
- },
9
- "infilling_metrics": {
10
- "GPT": {
11
- "num_samples": 10000,
12
- "validity_rate": 0.8687,
13
- "valid_count": 8687,
14
- "monomer_accuracy": {
15
- "mean": 0.102699750870045,
16
- "std": 0.21467870024616584,
17
- "median": 0.0,
18
- "n": 10000
19
- },
20
- "constraint_satisfaction": {
21
- "mean": 1.0,
22
- "std": 0.0,
23
- "n": 10000
24
- },
25
- "accuracy_distribution": {
26
- "pct_zero": 0.7473,
27
- "pct_above_50": 0.057,
28
- "pct_perfect": 0.0112
29
- }
30
- },
31
- "BERT": {
32
- "num_samples": 10000,
33
- "validity_rate": 0.8939,
34
- "valid_count": 8939,
35
- "monomer_accuracy": {
36
- "mean": 0.166446053619583,
37
- "std": 0.28598628151878797,
38
- "median": 0.0,
39
- "n": 10000
40
- },
41
- "constraint_satisfaction": {
42
- "mean": 1.0,
43
- "std": 0.0,
44
- "n": 10000
45
- },
46
- "accuracy_distribution": {
47
- "pct_zero": 0.6734,
48
- "pct_above_50": 0.1257,
49
- "pct_perfect": 0.0342
50
- }
51
- }
52
- },
53
- "generation_metrics": {
54
- "GPT": {
55
- "validity_rate": 0.8687,
56
- "uniqueness_rate": 0.9980430528375733,
57
- "novelty_rate": 0.596078431372549,
58
- "internal_diversity": 0.728
59
- },
60
- "BERT": {
61
- "validity_rate": 0.8939,
62
- "uniqueness_rate": 0.9928403624566506,
63
- "novelty_rate": 0.5077183098591549,
64
- "internal_diversity": 0.7221
65
- }
66
- }
67
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/comparison_summary_run2.json DELETED
@@ -1,67 +0,0 @@
1
- {
2
- "num_constraints": 10000,
3
- "paired_wins": {
4
- "GPT_wins": 1164,
5
- "BERT_wins": 2503,
6
- "ties": 6333,
7
- "ties_both_zero": 5851
8
- },
9
- "infilling_metrics": {
10
- "GPT": {
11
- "num_samples": 10000,
12
- "validity_rate": 0.8733,
13
- "valid_count": 8733,
14
- "monomer_accuracy": {
15
- "mean": 0.10169111704471999,
16
- "std": 0.21432767434230826,
17
- "median": 0.0,
18
- "n": 10000
19
- },
20
- "constraint_satisfaction": {
21
- "mean": 1.0,
22
- "std": 0.0,
23
- "n": 10000
24
- },
25
- "accuracy_distribution": {
26
- "pct_zero": 0.7526,
27
- "pct_above_50": 0.058,
28
- "pct_perfect": 0.0098
29
- }
30
- },
31
- "BERT": {
32
- "num_samples": 10000,
33
- "validity_rate": 0.8943,
34
- "valid_count": 8943,
35
- "monomer_accuracy": {
36
- "mean": 0.1690475914689885,
37
- "std": 0.2877969656999556,
38
- "median": 0.0,
39
- "n": 10000
40
- },
41
- "constraint_satisfaction": {
42
- "mean": 1.0,
43
- "std": 0.0,
44
- "n": 10000
45
- },
46
- "accuracy_distribution": {
47
- "pct_zero": 0.6701,
48
- "pct_above_50": 0.1268,
49
- "pct_perfect": 0.0352
50
- }
51
- }
52
- },
53
- "generation_metrics": {
54
- "GPT": {
55
- "validity_rate": 0.8733,
56
- "uniqueness_rate": 0.9955341806939196,
57
- "novelty_rate": 0.6065102369450196,
58
- "internal_diversity": 0.7282
59
- },
60
- "BERT": {
61
- "validity_rate": 0.8943,
62
- "uniqueness_rate": 0.9930672033993068,
63
- "novelty_rate": 0.5039972976016215,
64
- "internal_diversity": 0.7221
65
- }
66
- }
67
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/comparison_summary_run3.json DELETED
@@ -1,67 +0,0 @@
1
- {
2
- "num_constraints": 10000,
3
- "paired_wins": {
4
- "GPT_wins": 1135,
5
- "BERT_wins": 2475,
6
- "ties": 6390,
7
- "ties_both_zero": 5875
8
- },
9
- "infilling_metrics": {
10
- "GPT": {
11
- "num_samples": 10000,
12
- "validity_rate": 0.8684,
13
- "valid_count": 8684,
14
- "monomer_accuracy": {
15
- "mean": 0.10061371680606974,
16
- "std": 0.21239770258870322,
17
- "median": 0.0,
18
- "n": 10000
19
- },
20
- "constraint_satisfaction": {
21
- "mean": 1.0,
22
- "std": 0.0,
23
- "n": 10000
24
- },
25
- "accuracy_distribution": {
26
- "pct_zero": 0.7518,
27
- "pct_above_50": 0.054,
28
- "pct_perfect": 0.0116
29
- }
30
- },
31
- "BERT": {
32
- "num_samples": 10000,
33
- "validity_rate": 0.892,
34
- "valid_count": 8920,
35
- "monomer_accuracy": {
36
- "mean": 0.16679071352666944,
37
- "std": 0.2879362012786756,
38
- "median": 0.0,
39
- "n": 10000
40
- },
41
- "constraint_satisfaction": {
42
- "mean": 1.0,
43
- "std": 0.0,
44
- "n": 10000
45
- },
46
- "accuracy_distribution": {
47
- "pct_zero": 0.673,
48
- "pct_above_50": 0.1286,
49
- "pct_perfect": 0.0357
50
- }
51
- }
52
- },
53
- "generation_metrics": {
54
- "GPT": {
55
- "validity_rate": 0.8684,
56
- "uniqueness_rate": 0.9958544449562413,
57
- "novelty_rate": 0.6022201665124884,
58
- "internal_diversity": 0.7276
59
- },
60
- "BERT": {
61
- "validity_rate": 0.892,
62
- "uniqueness_rate": 0.992152466367713,
63
- "novelty_rate": 0.49977401129943505,
64
- "internal_diversity": 0.7204
65
- }
66
- }
67
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/comparison_summary_run4.json DELETED
@@ -1,67 +0,0 @@
1
- {
2
- "num_constraints": 10000,
3
- "paired_wins": {
4
- "GPT_wins": 1179,
5
- "BERT_wins": 2478,
6
- "ties": 6343,
7
- "ties_both_zero": 5860
8
- },
9
- "infilling_metrics": {
10
- "GPT": {
11
- "num_samples": 10000,
12
- "validity_rate": 0.8778,
13
- "valid_count": 8778,
14
- "monomer_accuracy": {
15
- "mean": 0.10346335061996827,
16
- "std": 0.21471265232937553,
17
- "median": 0.0,
18
- "n": 10000
19
- },
20
- "constraint_satisfaction": {
21
- "mean": 1.0,
22
- "std": 0.0,
23
- "n": 10000
24
- },
25
- "accuracy_distribution": {
26
- "pct_zero": 0.7453,
27
- "pct_above_50": 0.0551,
28
- "pct_perfect": 0.0121
29
- }
30
- },
31
- "BERT": {
32
- "num_samples": 10000,
33
- "validity_rate": 0.8948,
34
- "valid_count": 8948,
35
- "monomer_accuracy": {
36
- "mean": 0.16618561964995787,
37
- "std": 0.28760508727053113,
38
- "median": 0.0,
39
- "n": 10000
40
- },
41
- "constraint_satisfaction": {
42
- "mean": 1.0,
43
- "std": 0.0,
44
- "n": 10000
45
- },
46
- "accuracy_distribution": {
47
- "pct_zero": 0.6745,
48
- "pct_above_50": 0.1269,
49
- "pct_perfect": 0.0354
50
- }
51
- }
52
- },
53
- "generation_metrics": {
54
- "GPT": {
55
- "validity_rate": 0.8778,
56
- "uniqueness_rate": 0.9964684438368648,
57
- "novelty_rate": 0.60569338058763,
58
- "internal_diversity": 0.7289
59
- },
60
- "BERT": {
61
- "validity_rate": 0.8948,
62
- "uniqueness_rate": 0.9927358068842199,
63
- "novelty_rate": 0.5078239333558483,
64
- "internal_diversity": 0.7218
65
- }
66
- }
67
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.5/comparison_summary_run5.json DELETED
@@ -1,67 +0,0 @@
1
- {
2
- "num_constraints": 10000,
3
- "paired_wins": {
4
- "GPT_wins": 1177,
5
- "BERT_wins": 2470,
6
- "ties": 6353,
7
- "ties_both_zero": 5860
8
- },
9
- "infilling_metrics": {
10
- "GPT": {
11
- "num_samples": 10000,
12
- "validity_rate": 0.8658,
13
- "valid_count": 8658,
14
- "monomer_accuracy": {
15
- "mean": 0.10191266963428727,
16
- "std": 0.21260300891684258,
17
- "median": 0.0,
18
- "n": 10000
19
- },
20
- "constraint_satisfaction": {
21
- "mean": 1.0,
22
- "std": 0.0,
23
- "n": 10000
24
- },
25
- "accuracy_distribution": {
26
- "pct_zero": 0.7471,
27
- "pct_above_50": 0.054,
28
- "pct_perfect": 0.0114
29
- }
30
- },
31
- "BERT": {
32
- "num_samples": 10000,
33
- "validity_rate": 0.8937,
34
- "valid_count": 8937,
35
- "monomer_accuracy": {
36
- "mean": 0.1656202258199317,
37
- "std": 0.2858019613818268,
38
- "median": 0.0,
39
- "n": 10000
40
- },
41
- "constraint_satisfaction": {
42
- "mean": 1.0,
43
- "std": 0.0,
44
- "n": 10000
45
- },
46
- "accuracy_distribution": {
47
- "pct_zero": 0.6761,
48
- "pct_above_50": 0.1258,
49
- "pct_perfect": 0.0328
50
- }
51
- }
52
- },
53
- "generation_metrics": {
54
- "GPT": {
55
- "validity_rate": 0.8658,
56
- "uniqueness_rate": 0.9959574959574959,
57
- "novelty_rate": 0.5958483126522092,
58
- "internal_diversity": 0.7274
59
- },
60
- "BERT": {
61
- "validity_rate": 0.8937,
62
- "uniqueness_rate": 0.9927268658386483,
63
- "novelty_rate": 0.5099188458070334,
64
- "internal_diversity": 0.7203
65
- }
66
- }
67
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run1/eval_report.json DELETED
@@ -1,414 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:50.091845",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 8988,
7
- "num_unique": 8912,
8
- "num_novel": 3980
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 8988,
13
- "rate": 0.8988
14
- },
15
- "uniqueness": {
16
- "num_valid": 8988,
17
- "num_unique": 8912,
18
- "rate": 0.9915442812639075
19
- },
20
- "novelty": {
21
- "num_unique": 8912,
22
- "num_novel": 3980,
23
- "rate": 0.44658886894075406
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.8827,
27
- "num_molecules": 8912,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.3416,
32
- "median": 0.3164,
33
- "std": 0.1737,
34
- "percentiles": {
35
- "5": 0.0,
36
- "25": 0.2329,
37
- "50": 0.3164,
38
- "75": 0.486,
39
- "95": 0.6157
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 373,
97
- 10,
98
- 19,
99
- 17,
100
- 24,
101
- 32,
102
- 40,
103
- 71,
104
- 85,
105
- 158,
106
- 236,
107
- 290,
108
- 341,
109
- 324,
110
- 290,
111
- 245,
112
- 203,
113
- 155,
114
- 145,
115
- 123,
116
- 145,
117
- 121,
118
- 132,
119
- 132,
120
- 129,
121
- 136,
122
- 141,
123
- 166,
124
- 178,
125
- 166,
126
- 141,
127
- 106,
128
- 57,
129
- 42,
130
- 13,
131
- 11,
132
- 1,
133
- 0,
134
- 2,
135
- 0,
136
- 0,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.8851,
156
- "nn_self_distance": {
157
- "mean": 0.3742,
158
- "median": 0.3418,
159
- "std": 0.1507,
160
- "percentiles": {
161
- "5": 0.1582,
162
- "25": 0.2622,
163
- "50": 0.3418,
164
- "75": 0.5068,
165
- "95": 0.625
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 4,
223
- 15,
224
- 39,
225
- 23,
226
- 38,
227
- 41,
228
- 40,
229
- 58,
230
- 84,
231
- 119,
232
- 174,
233
- 273,
234
- 306,
235
- 334,
236
- 350,
237
- 308,
238
- 267,
239
- 237,
240
- 196,
241
- 174,
242
- 155,
243
- 138,
244
- 105,
245
- 115,
246
- 116,
247
- 126,
248
- 145,
249
- 187,
250
- 171,
251
- 189,
252
- 179,
253
- 147,
254
- 78,
255
- 36,
256
- 15,
257
- 9,
258
- 3,
259
- 4,
260
- 2,
261
- 0,
262
- 0,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.116,
282
- "peptide_length_mean": 6.9,
283
- "peptide_length_dist": {
284
- "2": 139,
285
- "3": 691,
286
- "4": 4478,
287
- "5": 391,
288
- "6": 410,
289
- "7": 347,
290
- "8": 331,
291
- "9": 439,
292
- "10": 288,
293
- "11": 173,
294
- "12": 151,
295
- "13": 137,
296
- "14": 95,
297
- "15": 121,
298
- "16": 80,
299
- "17": 73,
300
- "18": 66,
301
- "19": 45,
302
- "20": 124,
303
- "21": 63,
304
- "22": 28,
305
- "23": 32,
306
- "24": 34,
307
- "25": 252
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1593,
315
- "valid": 855,
316
- "generation_rate": 0.1593,
317
- "pass_rate": 0.5367
318
- },
319
- "Bz_Stapled": {
320
- "generated": 3,
321
- "valid": 0,
322
- "generation_rate": 0.0003,
323
- "pass_rate": 0.0
324
- },
325
- "Disulfide": {
326
- "generated": 261,
327
- "valid": 76,
328
- "generation_rate": 0.0261,
329
- "pass_rate": 0.2912
330
- },
331
- "Ester": {
332
- "generated": 354,
333
- "valid": 178,
334
- "generation_rate": 0.0354,
335
- "pass_rate": 0.5028
336
- },
337
- "Lanthionin": {
338
- "generated": 7,
339
- "valid": 2,
340
- "generation_rate": 0.0007,
341
- "pass_rate": 0.2857
342
- },
343
- "PG_Amide": {
344
- "generated": 304,
345
- "valid": 163,
346
- "generation_rate": 0.0304,
347
- "pass_rate": 0.5362
348
- },
349
- "Sulfanilamide": {
350
- "generated": 8,
351
- "valid": 1,
352
- "generation_rate": 0.0008,
353
- "pass_rate": 0.125
354
- },
355
- "Trp-Cys": {
356
- "generated": 4,
357
- "valid": 2,
358
- "generation_rate": 0.0004,
359
- "pass_rate": 0.5
360
- },
361
- "head2tail ring": {
362
- "generated": 315,
363
- "valid": 287,
364
- "generation_rate": 0.0315,
365
- "pass_rate": 0.9111
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2313,
372
- "has_connection_rate": 0.2313,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1796,
376
- "fraction": 0.7765
377
- },
378
- "2": {
379
- "count": 498,
380
- "fraction": 0.2153
381
- },
382
- "3": {
383
- "count": 19,
384
- "fraction": 0.0082
385
- }
386
- }
387
- },
388
- "per_conn_count_metrics": {
389
- "0": {
390
- "count": 7687,
391
- "validity": 0.9844,
392
- "uniqueness": 0.9906,
393
- "novelty": 0.4162
394
- },
395
- "1": {
396
- "count": 1796,
397
- "validity": 0.7121,
398
- "uniqueness": 0.9961,
399
- "novelty": 0.6209
400
- },
401
- "2": {
402
- "count": 498,
403
- "validity": 0.2831,
404
- "uniqueness": 1.0,
405
- "novelty": 0.4823
406
- },
407
- "3+": {
408
- "count": 19,
409
- "validity": 0.0526,
410
- "uniqueness": 1.0,
411
- "novelty": 1.0
412
- }
413
- }
414
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run1/eval_summary.txt DELETED
@@ -1,56 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:31:08.371303
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 8988/10000 (89.9%)
9
- Uniqueness: 8912/8988 (99.2%)
10
- Novelty: 4246/8912 (47.6%)
11
-
12
- Diversity: 0.8817 (avg MAP4C Jaccard distance)
13
- Molecules: 8912
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.3455, median=0.3184, std=0.1732
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8855
23
- Self-NN: mean=0.3726, median=0.3442
24
-
25
- Structural Summary:
26
- Multi-block rate: 11.6%
27
- Avg peptide len: 6.9
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1593 855 0.1593 53.7%
33
- Ester 354 178 0.0354 50.3%
34
- head2tail ring 315 287 0.0315 91.1%
35
- PG_Amide 304 163 0.0304 53.6%
36
- Disulfide 261 76 0.0261 29.1%
37
- Sulfanilamide 8 1 0.0008 12.5%
38
- Lanthionin 7 2 0.0007 28.6%
39
- Trp-Cys 4 2 0.0004 50.0%
40
- Bz_Stapled 3 0 0.0003 0.0%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 23.1% (2313/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1796 (77.6%)
46
- 2 connection(s): 498 (21.5%)
47
- 3 connection(s): 19 (0.8%)
48
-
49
- Per-Connection-Count Metrics:
50
- Bucket Count Validity Uniqueness Novelty
51
- ---------- -------- ---------- ------------ ----------
52
- 0 conn 7687 98.4% 99.1% 43.4%
53
- 1 conn 1796 71.2% 99.6% 72.1%
54
- 2 conn 498 28.3% 100.0% 50.3%
55
- 3+ conn 19 5.3% 100.0% 100.0%
56
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run1/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.8988,
4
- "valid_count": 8988,
5
- "monomer_accuracy": {
6
- "mean": 0.17945897186147186,
7
- "std": 0.3346611685127494,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7316,
18
- "pct_above_50": 0.1509,
19
- "pct_perfect": 0.0991
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run1/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:b10688b61bdde112a8457ed66cdfe09bdf85864b669ac7afaa06d4934f2d967c
3
- size 11632997
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run2/eval_report.json DELETED
@@ -1,418 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:49.927590",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 9040,
7
- "num_unique": 8986,
8
- "num_novel": 4035
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 9040,
13
- "rate": 0.904
14
- },
15
- "uniqueness": {
16
- "num_valid": 9040,
17
- "num_unique": 8986,
18
- "rate": 0.9940265486725663
19
- },
20
- "novelty": {
21
- "num_unique": 8986,
22
- "num_novel": 4035,
23
- "rate": 0.44903182728689073
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.8817,
27
- "num_molecules": 8986,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.3448,
32
- "median": 0.3179,
33
- "std": 0.172,
34
- "percentiles": {
35
- "5": 0.0,
36
- "25": 0.2349,
37
- "50": 0.3179,
38
- "75": 0.4902,
39
- "95": 0.6167
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 353,
97
- 6,
98
- 13,
99
- 17,
100
- 30,
101
- 30,
102
- 28,
103
- 52,
104
- 103,
105
- 141,
106
- 208,
107
- 353,
108
- 349,
109
- 335,
110
- 290,
111
- 226,
112
- 198,
113
- 150,
114
- 154,
115
- 129,
116
- 128,
117
- 129,
118
- 139,
119
- 124,
120
- 117,
121
- 138,
122
- 170,
123
- 163,
124
- 174,
125
- 177,
126
- 148,
127
- 114,
128
- 65,
129
- 28,
130
- 11,
131
- 8,
132
- 2,
133
- 0,
134
- 0,
135
- 0,
136
- 0,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.8856,
156
- "nn_self_distance": {
157
- "mean": 0.375,
158
- "median": 0.3438,
159
- "std": 0.1497,
160
- "percentiles": {
161
- "5": 0.1509,
162
- "25": 0.2666,
163
- "50": 0.3438,
164
- "75": 0.5049,
165
- "95": 0.624
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 3,
223
- 19,
224
- 20,
225
- 30,
226
- 40,
227
- 53,
228
- 44,
229
- 61,
230
- 92,
231
- 106,
232
- 170,
233
- 227,
234
- 291,
235
- 319,
236
- 348,
237
- 344,
238
- 278,
239
- 248,
240
- 216,
241
- 170,
242
- 156,
243
- 146,
244
- 118,
245
- 110,
246
- 111,
247
- 153,
248
- 137,
249
- 163,
250
- 172,
251
- 194,
252
- 168,
253
- 150,
254
- 72,
255
- 30,
256
- 17,
257
- 11,
258
- 3,
259
- 4,
260
- 5,
261
- 1,
262
- 0,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.1235,
282
- "peptide_length_mean": 6.95,
283
- "peptide_length_dist": {
284
- "2": 136,
285
- "3": 705,
286
- "4": 4477,
287
- "5": 388,
288
- "6": 421,
289
- "7": 358,
290
- "8": 338,
291
- "9": 416,
292
- "10": 292,
293
- "11": 174,
294
- "12": 148,
295
- "13": 143,
296
- "14": 99,
297
- "15": 126,
298
- "16": 83,
299
- "17": 75,
300
- "18": 66,
301
- "19": 47,
302
- "20": 121,
303
- "21": 63,
304
- "22": 29,
305
- "23": 30,
306
- "24": 33,
307
- "25": 272
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1598,
315
- "valid": 907,
316
- "generation_rate": 0.1598,
317
- "pass_rate": 0.5676
318
- },
319
- "Bz_Stapled": {
320
- "generated": 3,
321
- "valid": 1,
322
- "generation_rate": 0.0003,
323
- "pass_rate": 0.3333
324
- },
325
- "Disulfide": {
326
- "generated": 257,
327
- "valid": 135,
328
- "generation_rate": 0.0257,
329
- "pass_rate": 0.5253
330
- },
331
- "Ester": {
332
- "generated": 334,
333
- "valid": 173,
334
- "generation_rate": 0.0334,
335
- "pass_rate": 0.518
336
- },
337
- "Lanthionin": {
338
- "generated": 7,
339
- "valid": 4,
340
- "generation_rate": 0.0007,
341
- "pass_rate": 0.5714
342
- },
343
- "PG_Amide": {
344
- "generated": 309,
345
- "valid": 206,
346
- "generation_rate": 0.0309,
347
- "pass_rate": 0.6667
348
- },
349
- "Sulfanilamide": {
350
- "generated": 7,
351
- "valid": 3,
352
- "generation_rate": 0.0007,
353
- "pass_rate": 0.4286
354
- },
355
- "Trp-Cys": {
356
- "generated": 1,
357
- "valid": 1,
358
- "generation_rate": 0.0001,
359
- "pass_rate": 1.0
360
- },
361
- "head2tail ring": {
362
- "generated": 322,
363
- "valid": 285,
364
- "generation_rate": 0.0322,
365
- "pass_rate": 0.8851
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2312,
372
- "has_connection_rate": 0.2312,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1809,
376
- "fraction": 0.7824
377
- },
378
- "2": {
379
- "count": 481,
380
- "fraction": 0.208
381
- },
382
- "3": {
383
- "count": 21,
384
- "fraction": 0.0091
385
- },
386
- "4+": {
387
- "count": 1,
388
- "fraction": 0.0004
389
- }
390
- }
391
- },
392
- "per_conn_count_metrics": {
393
- "0": {
394
- "count": 7688,
395
- "validity": 0.981,
396
- "uniqueness": 0.9931,
397
- "novelty": 0.4158
398
- },
399
- "1": {
400
- "count": 1809,
401
- "validity": 0.712,
402
- "uniqueness": 0.9984,
403
- "novelty": 0.6306
404
- },
405
- "2": {
406
- "count": 481,
407
- "validity": 0.422,
408
- "uniqueness": 1.0,
409
- "novelty": 0.5271
410
- },
411
- "3+": {
412
- "count": 22,
413
- "validity": 0.3182,
414
- "uniqueness": 1.0,
415
- "novelty": 0.4286
416
- }
417
- }
418
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run2/eval_summary.txt DELETED
@@ -1,57 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:31:10.911737
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 9040/10000 (90.4%)
9
- Uniqueness: 8986/9040 (99.4%)
10
- Novelty: 4301/8986 (47.9%)
11
-
12
- Diversity: 0.8825 (avg MAP4C Jaccard distance)
13
- Molecules: 8986
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.3490, median=0.3223, std=0.1716
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8867
23
- Self-NN: mean=0.3738, median=0.3433
24
-
25
- Structural Summary:
26
- Multi-block rate: 12.3%
27
- Avg peptide len: 7.0
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1598 907 0.1598 56.8%
33
- Ester 334 173 0.0334 51.8%
34
- head2tail ring 322 285 0.0322 88.5%
35
- PG_Amide 309 206 0.0309 66.7%
36
- Disulfide 257 135 0.0257 52.5%
37
- Lanthionin 7 4 0.0007 57.1%
38
- Sulfanilamide 7 3 0.0007 42.9%
39
- Bz_Stapled 3 1 0.0003 33.3%
40
- Trp-Cys 1 1 0.0001 100.0%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 23.1% (2312/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1809 (78.2%)
46
- 2 connection(s): 481 (20.8%)
47
- 3 connection(s): 21 (0.9%)
48
- 4+ connection(s): 1 (0.0%)
49
-
50
- Per-Connection-Count Metrics:
51
- Bucket Count Validity Uniqueness Novelty
52
- ---------- -------- ---------- ------------ ----------
53
- 0 conn 7688 98.1% 99.3% 43.5%
54
- 1 conn 1809 71.2% 99.8% 72.4%
55
- 2 conn 481 42.2% 100.0% 54.7%
56
- 3+ conn 22 31.8% 100.0% 42.9%
57
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run2/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.904,
4
- "valid_count": 9040,
5
- "monomer_accuracy": {
6
- "mean": 0.1859096031746032,
7
- "std": 0.3394147178238279,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7228,
18
- "pct_above_50": 0.1577,
19
- "pct_perfect": 0.1032
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run2/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:88a3a92755386555de9cec180895853c20bd97ffed41c4d24f1233e13384da81
3
- size 11625335
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run3/eval_report.json DELETED
@@ -1,418 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:49.229777",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 8975,
7
- "num_unique": 8917,
8
- "num_novel": 3979
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 8975,
13
- "rate": 0.8975
14
- },
15
- "uniqueness": {
16
- "num_valid": 8975,
17
- "num_unique": 8917,
18
- "rate": 0.9935376044568245
19
- },
20
- "novelty": {
21
- "num_unique": 8917,
22
- "num_novel": 3979,
23
- "rate": 0.4462263092968487
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.8823,
27
- "num_molecules": 8917,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.3453,
32
- "median": 0.3154,
33
- "std": 0.1702,
34
- "percentiles": {
35
- "5": 0.0,
36
- "25": 0.2373,
37
- "50": 0.3154,
38
- "75": 0.4912,
39
- "95": 0.6167
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 311,
97
- 8,
98
- 13,
99
- 32,
100
- 32,
101
- 33,
102
- 41,
103
- 73,
104
- 82,
105
- 143,
106
- 211,
107
- 318,
108
- 350,
109
- 335,
110
- 317,
111
- 241,
112
- 215,
113
- 177,
114
- 137,
115
- 126,
116
- 105,
117
- 120,
118
- 128,
119
- 137,
120
- 129,
121
- 148,
122
- 167,
123
- 145,
124
- 172,
125
- 189,
126
- 129,
127
- 113,
128
- 69,
129
- 27,
130
- 15,
131
- 4,
132
- 2,
133
- 2,
134
- 4,
135
- 0,
136
- 0,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.885,
156
- "nn_self_distance": {
157
- "mean": 0.3739,
158
- "median": 0.3457,
159
- "std": 0.1493,
160
- "percentiles": {
161
- "5": 0.1543,
162
- "25": 0.2622,
163
- "50": 0.3457,
164
- "75": 0.5015,
165
- "95": 0.6216
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 1,
223
- 27,
224
- 22,
225
- 30,
226
- 24,
227
- 42,
228
- 62,
229
- 64,
230
- 82,
231
- 120,
232
- 163,
233
- 253,
234
- 324,
235
- 352,
236
- 313,
237
- 299,
238
- 259,
239
- 227,
240
- 232,
241
- 171,
242
- 143,
243
- 156,
244
- 124,
245
- 138,
246
- 106,
247
- 130,
248
- 150,
249
- 181,
250
- 187,
251
- 214,
252
- 143,
253
- 105,
254
- 75,
255
- 41,
256
- 17,
257
- 11,
258
- 8,
259
- 2,
260
- 2,
261
- 0,
262
- 0,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.1163,
282
- "peptide_length_mean": 6.91,
283
- "peptide_length_dist": {
284
- "2": 133,
285
- "3": 690,
286
- "4": 4475,
287
- "5": 392,
288
- "6": 414,
289
- "7": 346,
290
- "8": 327,
291
- "9": 436,
292
- "10": 286,
293
- "11": 178,
294
- "12": 147,
295
- "13": 133,
296
- "14": 91,
297
- "15": 126,
298
- "16": 79,
299
- "17": 76,
300
- "18": 67,
301
- "19": 48,
302
- "20": 121,
303
- "21": 64,
304
- "22": 28,
305
- "23": 33,
306
- "24": 35,
307
- "25": 250
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1598,
315
- "valid": 862,
316
- "generation_rate": 0.1598,
317
- "pass_rate": 0.5394
318
- },
319
- "Bz_Stapled": {
320
- "generated": 3,
321
- "valid": 0,
322
- "generation_rate": 0.0003,
323
- "pass_rate": 0.0
324
- },
325
- "Disulfide": {
326
- "generated": 265,
327
- "valid": 75,
328
- "generation_rate": 0.0265,
329
- "pass_rate": 0.283
330
- },
331
- "Ester": {
332
- "generated": 353,
333
- "valid": 175,
334
- "generation_rate": 0.0353,
335
- "pass_rate": 0.4958
336
- },
337
- "Lanthionin": {
338
- "generated": 8,
339
- "valid": 4,
340
- "generation_rate": 0.0008,
341
- "pass_rate": 0.5
342
- },
343
- "PG_Amide": {
344
- "generated": 307,
345
- "valid": 162,
346
- "generation_rate": 0.0307,
347
- "pass_rate": 0.5277
348
- },
349
- "Sulfanilamide": {
350
- "generated": 9,
351
- "valid": 1,
352
- "generation_rate": 0.0009,
353
- "pass_rate": 0.1111
354
- },
355
- "Trp-Cys": {
356
- "generated": 1,
357
- "valid": 0,
358
- "generation_rate": 0.0001,
359
- "pass_rate": 0.0
360
- },
361
- "head2tail ring": {
362
- "generated": 310,
363
- "valid": 280,
364
- "generation_rate": 0.031,
365
- "pass_rate": 0.9032
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2307,
372
- "has_connection_rate": 0.2307,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1779,
376
- "fraction": 0.7711
377
- },
378
- "2": {
379
- "count": 510,
380
- "fraction": 0.2211
381
- },
382
- "3": {
383
- "count": 17,
384
- "fraction": 0.0074
385
- },
386
- "4+": {
387
- "count": 1,
388
- "fraction": 0.0004
389
- }
390
- }
391
- },
392
- "per_conn_count_metrics": {
393
- "0": {
394
- "count": 7693,
395
- "validity": 0.983,
396
- "uniqueness": 0.993,
397
- "novelty": 0.4138
398
- },
399
- "1": {
400
- "count": 1779,
401
- "validity": 0.7133,
402
- "uniqueness": 0.9961,
403
- "novelty": 0.6266
404
- },
405
- "2": {
406
- "count": 510,
407
- "validity": 0.2784,
408
- "uniqueness": 1.0,
409
- "novelty": 0.5563
410
- },
411
- "3+": {
412
- "count": 18,
413
- "validity": 0.1111,
414
- "uniqueness": 1.0,
415
- "novelty": 0.5
416
- }
417
- }
418
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run3/eval_summary.txt DELETED
@@ -1,57 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:31:34.721342
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 8975/10000 (89.8%)
9
- Uniqueness: 8917/8975 (99.4%)
10
- Novelty: 4227/8917 (47.4%)
11
-
12
- Diversity: 0.8827 (avg MAP4C Jaccard distance)
13
- Molecules: 8917
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.3429, median=0.3130, std=0.1718
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8846
23
- Self-NN: mean=0.3737, median=0.3428
24
-
25
- Structural Summary:
26
- Multi-block rate: 11.6%
27
- Avg peptide len: 6.9
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1598 862 0.1598 53.9%
33
- Ester 353 175 0.0353 49.6%
34
- head2tail ring 310 280 0.0310 90.3%
35
- PG_Amide 307 162 0.0307 52.8%
36
- Disulfide 265 75 0.0265 28.3%
37
- Sulfanilamide 9 1 0.0009 11.1%
38
- Lanthionin 8 4 0.0008 50.0%
39
- Bz_Stapled 3 0 0.0003 0.0%
40
- Trp-Cys 1 0 0.0001 0.0%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 23.1% (2307/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1779 (77.1%)
46
- 2 connection(s): 510 (22.1%)
47
- 3 connection(s): 17 (0.7%)
48
- 4+ connection(s): 1 (0.0%)
49
-
50
- Per-Connection-Count Metrics:
51
- Bucket Count Validity Uniqueness Novelty
52
- ---------- -------- ---------- ------------ ----------
53
- 0 conn 7693 98.3% 99.3% 43.0%
54
- 1 conn 1779 71.3% 99.6% 72.2%
55
- 2 conn 510 27.8% 100.0% 57.8%
56
- 3+ conn 18 11.1% 100.0% 50.0%
57
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run3/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.8975,
4
- "valid_count": 8975,
5
- "monomer_accuracy": {
6
- "mean": 0.17857642496392495,
7
- "std": 0.33342857016809685,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7326,
18
- "pct_above_50": 0.1509,
19
- "pct_perfect": 0.0968
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run3/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:a517f21165e2cb9ea7c53288b0abfea9b4c60a349d99794b6e9af99ab6956635
3
- size 11630991
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run4/eval_report.json DELETED
@@ -1,417 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:49.219358",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 9016,
7
- "num_unique": 8960,
8
- "num_novel": 4057
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 9016,
13
- "rate": 0.9016
14
- },
15
- "uniqueness": {
16
- "num_valid": 9016,
17
- "num_unique": 8960,
18
- "rate": 0.9937888198757764
19
- },
20
- "novelty": {
21
- "num_unique": 8960,
22
- "num_novel": 4057,
23
- "rate": 0.45279017857142856
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.8835,
27
- "num_molecules": 8960,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.3435,
32
- "median": 0.3145,
33
- "std": 0.1714,
34
- "percentiles": {
35
- "5": 0.0,
36
- "25": 0.2352,
37
- "50": 0.3145,
38
- "75": 0.4873,
39
- "95": 0.6177
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 332,
97
- 14,
98
- 15,
99
- 16,
100
- 29,
101
- 26,
102
- 47,
103
- 77,
104
- 102,
105
- 146,
106
- 219,
107
- 306,
108
- 372,
109
- 333,
110
- 283,
111
- 245,
112
- 190,
113
- 157,
114
- 137,
115
- 126,
116
- 118,
117
- 137,
118
- 140,
119
- 130,
120
- 149,
121
- 148,
122
- 153,
123
- 151,
124
- 164,
125
- 153,
126
- 143,
127
- 123,
128
- 59,
129
- 22,
130
- 23,
131
- 11,
132
- 3,
133
- 1,
134
- 0,
135
- 0,
136
- 0,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.885,
156
- "nn_self_distance": {
157
- "mean": 0.375,
158
- "median": 0.3447,
159
- "std": 0.1507,
160
- "percentiles": {
161
- "5": 0.1523,
162
- "25": 0.2656,
163
- "50": 0.3447,
164
- "75": 0.5034,
165
- "95": 0.6274
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 1,
223
- 18,
224
- 33,
225
- 33,
226
- 53,
227
- 29,
228
- 42,
229
- 63,
230
- 80,
231
- 125,
232
- 160,
233
- 255,
234
- 272,
235
- 330,
236
- 316,
237
- 332,
238
- 283,
239
- 301,
240
- 203,
241
- 171,
242
- 141,
243
- 128,
244
- 107,
245
- 121,
246
- 123,
247
- 119,
248
- 145,
249
- 175,
250
- 191,
251
- 177,
252
- 161,
253
- 152,
254
- 80,
255
- 39,
256
- 21,
257
- 12,
258
- 1,
259
- 4,
260
- 1,
261
- 1,
262
- 1,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.1199,
282
- "peptide_length_mean": 6.95,
283
- "peptide_length_dist": {
284
- "2": 134,
285
- "3": 686,
286
- "4": 4462,
287
- "5": 393,
288
- "6": 410,
289
- "7": 348,
290
- "8": 340,
291
- "9": 451,
292
- "10": 290,
293
- "11": 175,
294
- "12": 154,
295
- "13": 140,
296
- "14": 101,
297
- "15": 129,
298
- "16": 76,
299
- "17": 74,
300
- "18": 65,
301
- "19": 48,
302
- "20": 123,
303
- "21": 60,
304
- "22": 28,
305
- "23": 32,
306
- "24": 34,
307
- "25": 263
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1608,
315
- "valid": 898,
316
- "generation_rate": 0.1608,
317
- "pass_rate": 0.5585
318
- },
319
- "Bz_Stapled": {
320
- "generated": 3,
321
- "valid": 0,
322
- "generation_rate": 0.0003,
323
- "pass_rate": 0.0
324
- },
325
- "Disulfide": {
326
- "generated": 253,
327
- "valid": 130,
328
- "generation_rate": 0.0253,
329
- "pass_rate": 0.5138
330
- },
331
- "Ester": {
332
- "generated": 347,
333
- "valid": 178,
334
- "generation_rate": 0.0347,
335
- "pass_rate": 0.513
336
- },
337
- "Lanthionin": {
338
- "generated": 5,
339
- "valid": 1,
340
- "generation_rate": 0.0005,
341
- "pass_rate": 0.2
342
- },
343
- "PG_Amide": {
344
- "generated": 290,
345
- "valid": 136,
346
- "generation_rate": 0.029,
347
- "pass_rate": 0.469
348
- },
349
- "Sulfanilamide": {
350
- "generated": 6,
351
- "valid": 2,
352
- "generation_rate": 0.0006,
353
- "pass_rate": 0.3333
354
- },
355
- "Trp-Cys": {
356
- "generated": 2,
357
- "valid": 1,
358
- "generation_rate": 0.0002,
359
- "pass_rate": 0.5
360
- },
361
- "head2tail ring": {
362
- "generated": 306,
363
- "valid": 279,
364
- "generation_rate": 0.0306,
365
- "pass_rate": 0.9118
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2297,
372
- "has_connection_rate": 0.2297,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1793,
376
- "fraction": 0.7806
377
- },
378
- "2": {
379
- "count": 486,
380
- "fraction": 0.2116
381
- },
382
- "3": {
383
- "count": 17,
384
- "fraction": 0.0074
385
- },
386
- "4+": {
387
- "count": 1,
388
- "fraction": 0.0004
389
- }
390
- }
391
- },
392
- "per_conn_count_metrics": {
393
- "0": {
394
- "count": 7703,
395
- "validity": 0.9826,
396
- "uniqueness": 0.993,
397
- "novelty": 0.419
398
- },
399
- "1": {
400
- "count": 1793,
401
- "validity": 0.7078,
402
- "uniqueness": 0.9976,
403
- "novelty": 0.6335
404
- },
405
- "2": {
406
- "count": 486,
407
- "validity": 0.3663,
408
- "uniqueness": 1.0,
409
- "novelty": 0.5955
410
- },
411
- "3+": {
412
- "count": 18,
413
- "validity": 0.0,
414
- "uniqueness": 0.0
415
- }
416
- }
417
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run4/eval_summary.txt DELETED
@@ -1,57 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:31:33.749020
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 9016/10000 (90.2%)
9
- Uniqueness: 8960/9016 (99.4%)
10
- Novelty: 4314/8960 (48.1%)
11
-
12
- Diversity: 0.8810 (avg MAP4C Jaccard distance)
13
- Molecules: 8960
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.3447, median=0.3154, std=0.1731
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8846
23
- Self-NN: mean=0.3737, median=0.3428
24
-
25
- Structural Summary:
26
- Multi-block rate: 12.0%
27
- Avg peptide len: 7.0
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1608 898 0.1608 55.9%
33
- Ester 347 178 0.0347 51.3%
34
- head2tail ring 306 279 0.0306 91.2%
35
- PG_Amide 290 136 0.0290 46.9%
36
- Disulfide 253 130 0.0253 51.4%
37
- Sulfanilamide 6 2 0.0006 33.3%
38
- Lanthionin 5 1 0.0005 20.0%
39
- Bz_Stapled 3 0 0.0003 0.0%
40
- Trp-Cys 2 1 0.0002 50.0%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 23.0% (2297/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1793 (78.1%)
46
- 2 connection(s): 486 (21.2%)
47
- 3 connection(s): 17 (0.7%)
48
- 4+ connection(s): 1 (0.0%)
49
-
50
- Per-Connection-Count Metrics:
51
- Bucket Count Validity Uniqueness Novelty
52
- ---------- -------- ---------- ------------ ----------
53
- 0 conn 7703 98.3% 99.3% 43.7%
54
- 1 conn 1793 70.8% 99.8% 72.8%
55
- 2 conn 486 36.6% 100.0% 61.8%
56
- 3+ conn 18 0.0% 0.0% N/A
57
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run4/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.9016,
4
- "valid_count": 9016,
5
- "monomer_accuracy": {
6
- "mean": 0.18175662337662338,
7
- "std": 0.3363774793341434,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.728,
18
- "pct_above_50": 0.1542,
19
- "pct_perfect": 0.0995
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run4/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:82fa9675dd4b545e1a3e7d7f708b9e0ec2355b37263bdcd3f9e1661c05d26218
3
- size 11628964
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run5/eval_report.json DELETED
@@ -1,418 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:49.792785",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 9013,
7
- "num_unique": 8955,
8
- "num_novel": 4036
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 9013,
13
- "rate": 0.9013
14
- },
15
- "uniqueness": {
16
- "num_valid": 9013,
17
- "num_unique": 8955,
18
- "rate": 0.9935648507711085
19
- },
20
- "novelty": {
21
- "num_unique": 8955,
22
- "num_novel": 4036,
23
- "rate": 0.4506979341150195
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.8825,
27
- "num_molecules": 8955,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.346,
32
- "median": 0.3164,
33
- "std": 0.1711,
34
- "percentiles": {
35
- "5": 0.0,
36
- "25": 0.2378,
37
- "50": 0.3164,
38
- "75": 0.4932,
39
- "95": 0.6167
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 324,
97
- 14,
98
- 18,
99
- 21,
100
- 23,
101
- 31,
102
- 31,
103
- 64,
104
- 101,
105
- 146,
106
- 218,
107
- 305,
108
- 362,
109
- 342,
110
- 294,
111
- 236,
112
- 187,
113
- 162,
114
- 164,
115
- 113,
116
- 139,
117
- 111,
118
- 140,
119
- 121,
120
- 133,
121
- 160,
122
- 138,
123
- 193,
124
- 175,
125
- 179,
126
- 128,
127
- 97,
128
- 68,
129
- 30,
130
- 14,
131
- 8,
132
- 6,
133
- 2,
134
- 0,
135
- 1,
136
- 1,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.8848,
156
- "nn_self_distance": {
157
- "mean": 0.373,
158
- "median": 0.3398,
159
- "std": 0.15,
160
- "percentiles": {
161
- "5": 0.1543,
162
- "25": 0.2627,
163
- "50": 0.3398,
164
- "75": 0.5029,
165
- "95": 0.6226
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 2,
223
- 15,
224
- 24,
225
- 35,
226
- 42,
227
- 46,
228
- 49,
229
- 65,
230
- 91,
231
- 117,
232
- 147,
233
- 278,
234
- 294,
235
- 339,
236
- 317,
237
- 365,
238
- 277,
239
- 241,
240
- 201,
241
- 145,
242
- 156,
243
- 126,
244
- 122,
245
- 116,
246
- 114,
247
- 131,
248
- 157,
249
- 165,
250
- 182,
251
- 190,
252
- 177,
253
- 126,
254
- 68,
255
- 46,
256
- 16,
257
- 12,
258
- 3,
259
- 2,
260
- 1,
261
- 0,
262
- 0,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.1202,
282
- "peptide_length_mean": 6.97,
283
- "peptide_length_dist": {
284
- "2": 133,
285
- "3": 686,
286
- "4": 4468,
287
- "5": 387,
288
- "6": 405,
289
- "7": 362,
290
- "8": 343,
291
- "9": 417,
292
- "10": 295,
293
- "11": 172,
294
- "12": 153,
295
- "13": 147,
296
- "14": 102,
297
- "15": 125,
298
- "16": 79,
299
- "17": 76,
300
- "18": 67,
301
- "19": 48,
302
- "20": 125,
303
- "21": 63,
304
- "22": 29,
305
- "23": 33,
306
- "24": 36,
307
- "25": 262
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1607,
315
- "valid": 903,
316
- "generation_rate": 0.1607,
317
- "pass_rate": 0.5619
318
- },
319
- "Bz_Stapled": {
320
- "generated": 4,
321
- "valid": 0,
322
- "generation_rate": 0.0004,
323
- "pass_rate": 0.0
324
- },
325
- "Disulfide": {
326
- "generated": 255,
327
- "valid": 131,
328
- "generation_rate": 0.0255,
329
- "pass_rate": 0.5137
330
- },
331
- "Ester": {
332
- "generated": 333,
333
- "valid": 142,
334
- "generation_rate": 0.0333,
335
- "pass_rate": 0.4264
336
- },
337
- "Lanthionin": {
338
- "generated": 5,
339
- "valid": 3,
340
- "generation_rate": 0.0005,
341
- "pass_rate": 0.6
342
- },
343
- "PG_Amide": {
344
- "generated": 292,
345
- "valid": 159,
346
- "generation_rate": 0.0292,
347
- "pass_rate": 0.5445
348
- },
349
- "Sulfanilamide": {
350
- "generated": 5,
351
- "valid": 4,
352
- "generation_rate": 0.0005,
353
- "pass_rate": 0.8
354
- },
355
- "Trp-Cys": {
356
- "generated": 2,
357
- "valid": 1,
358
- "generation_rate": 0.0002,
359
- "pass_rate": 0.5
360
- },
361
- "head2tail ring": {
362
- "generated": 320,
363
- "valid": 290,
364
- "generation_rate": 0.032,
365
- "pass_rate": 0.9062
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2305,
372
- "has_connection_rate": 0.2305,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1804,
376
- "fraction": 0.7826
377
- },
378
- "2": {
379
- "count": 485,
380
- "fraction": 0.2104
381
- },
382
- "3": {
383
- "count": 15,
384
- "fraction": 0.0065
385
- },
386
- "4+": {
387
- "count": 1,
388
- "fraction": 0.0004
389
- }
390
- }
391
- },
392
- "per_conn_count_metrics": {
393
- "0": {
394
- "count": 7695,
395
- "validity": 0.9825,
396
- "uniqueness": 0.993,
397
- "novelty": 0.4196
398
- },
399
- "1": {
400
- "count": 1804,
401
- "validity": 0.7068,
402
- "uniqueness": 0.9969,
403
- "novelty": 0.6168
404
- },
405
- "2": {
406
- "count": 485,
407
- "validity": 0.3629,
408
- "uniqueness": 0.9943,
409
- "novelty": 0.5829
410
- },
411
- "3+": {
412
- "count": 16,
413
- "validity": 0.125,
414
- "uniqueness": 1.0,
415
- "novelty": 0.0
416
- }
417
- }
418
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run5/eval_summary.txt DELETED
@@ -1,57 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:31:34.843785
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 9013/10000 (90.1%)
9
- Uniqueness: 8955/9013 (99.4%)
10
- Novelty: 4300/8955 (48.0%)
11
-
12
- Diversity: 0.8827 (avg MAP4C Jaccard distance)
13
- Molecules: 8955
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.3503, median=0.3193, std=0.1733
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8846
23
- Self-NN: mean=0.3737, median=0.3428
24
-
25
- Structural Summary:
26
- Multi-block rate: 12.0%
27
- Avg peptide len: 7.0
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1607 903 0.1607 56.2%
33
- Ester 333 142 0.0333 42.6%
34
- head2tail ring 320 290 0.0320 90.6%
35
- PG_Amide 292 159 0.0292 54.4%
36
- Disulfide 255 131 0.0255 51.4%
37
- Lanthionin 5 3 0.0005 60.0%
38
- Sulfanilamide 5 4 0.0005 80.0%
39
- Bz_Stapled 4 0 0.0004 0.0%
40
- Trp-Cys 2 1 0.0002 50.0%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 23.1% (2305/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1804 (78.3%)
46
- 2 connection(s): 485 (21.0%)
47
- 3 connection(s): 15 (0.7%)
48
- 4+ connection(s): 1 (0.0%)
49
-
50
- Per-Connection-Count Metrics:
51
- Bucket Count Validity Uniqueness Novelty
52
- ---------- -------- ---------- ------------ ----------
53
- 0 conn 7695 98.2% 99.3% 43.8%
54
- 1 conn 1804 70.7% 99.7% 71.5%
55
- 2 conn 485 36.3% 99.4% 60.6%
56
- 3+ conn 16 12.5% 100.0% 0.0%
57
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run5/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.9013,
4
- "valid_count": 9013,
5
- "monomer_accuracy": {
6
- "mean": 0.17705492424242425,
7
- "std": 0.3321442657795057,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7339,
18
- "pct_above_50": 0.15,
19
- "pct_perfect": 0.0953
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/BERT/run5/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:a1c6f54910f9f97173ce4d94fa5db20f7b7cf2dd8e6940f9fc562de01c903020
3
- size 11630107
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run1/eval_report.json DELETED
@@ -1,413 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:50.265418",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 8891,
7
- "num_unique": 8846,
8
- "num_novel": 4583
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 8891,
13
- "rate": 0.8891
14
- },
15
- "uniqueness": {
16
- "num_valid": 8891,
17
- "num_unique": 8846,
18
- "rate": 0.9949387020582612
19
- },
20
- "novelty": {
21
- "num_unique": 8846,
22
- "num_novel": 4583,
23
- "rate": 0.5180872710829754
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.883,
27
- "num_molecules": 8846,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.3641,
32
- "median": 0.3384,
33
- "std": 0.1758,
34
- "percentiles": {
35
- "5": 0.0,
36
- "25": 0.2456,
37
- "50": 0.3384,
38
- "75": 0.5181,
39
- "95": 0.6299
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 314,
97
- 8,
98
- 11,
99
- 11,
100
- 23,
101
- 22,
102
- 26,
103
- 61,
104
- 87,
105
- 132,
106
- 199,
107
- 279,
108
- 290,
109
- 298,
110
- 314,
111
- 244,
112
- 196,
113
- 148,
114
- 138,
115
- 135,
116
- 132,
117
- 122,
118
- 130,
119
- 137,
120
- 159,
121
- 150,
122
- 160,
123
- 169,
124
- 183,
125
- 187,
126
- 201,
127
- 156,
128
- 93,
129
- 36,
130
- 24,
131
- 15,
132
- 5,
133
- 3,
134
- 2,
135
- 0,
136
- 0,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.8847,
156
- "nn_self_distance": {
157
- "mean": 0.3718,
158
- "median": 0.3403,
159
- "std": 0.149,
160
- "percentiles": {
161
- "5": 0.1582,
162
- "25": 0.2632,
163
- "50": 0.3403,
164
- "75": 0.4966,
165
- "95": 0.6255
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 4,
223
- 23,
224
- 30,
225
- 30,
226
- 34,
227
- 39,
228
- 38,
229
- 58,
230
- 90,
231
- 127,
232
- 171,
233
- 257,
234
- 300,
235
- 315,
236
- 370,
237
- 294,
238
- 309,
239
- 252,
240
- 217,
241
- 161,
242
- 158,
243
- 153,
244
- 125,
245
- 110,
246
- 110,
247
- 129,
248
- 165,
249
- 149,
250
- 161,
251
- 166,
252
- 171,
253
- 124,
254
- 79,
255
- 38,
256
- 20,
257
- 11,
258
- 7,
259
- 2,
260
- 3,
261
- 0,
262
- 0,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.1139,
282
- "peptide_length_mean": 6.84,
283
- "peptide_length_dist": {
284
- "2": 136,
285
- "3": 696,
286
- "4": 4454,
287
- "5": 395,
288
- "6": 403,
289
- "7": 352,
290
- "8": 324,
291
- "9": 426,
292
- "10": 283,
293
- "11": 173,
294
- "12": 142,
295
- "13": 128,
296
- "14": 96,
297
- "15": 121,
298
- "16": 73,
299
- "17": 71,
300
- "18": 65,
301
- "19": 45,
302
- "20": 119,
303
- "21": 61,
304
- "22": 25,
305
- "23": 26,
306
- "24": 31,
307
- "25": 246
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1614,
315
- "valid": 843,
316
- "generation_rate": 0.1614,
317
- "pass_rate": 0.5223
318
- },
319
- "Bz_Stapled": {
320
- "generated": 3,
321
- "valid": 0,
322
- "generation_rate": 0.0003,
323
- "pass_rate": 0.0
324
- },
325
- "Disulfide": {
326
- "generated": 247,
327
- "valid": 83,
328
- "generation_rate": 0.0247,
329
- "pass_rate": 0.336
330
- },
331
- "Ester": {
332
- "generated": 320,
333
- "valid": 148,
334
- "generation_rate": 0.032,
335
- "pass_rate": 0.4625
336
- },
337
- "Lanthionin": {
338
- "generated": 7,
339
- "valid": 0,
340
- "generation_rate": 0.0007,
341
- "pass_rate": 0.0
342
- },
343
- "PG_Amide": {
344
- "generated": 305,
345
- "valid": 164,
346
- "generation_rate": 0.0305,
347
- "pass_rate": 0.5377
348
- },
349
- "Sulfanilamide": {
350
- "generated": 4,
351
- "valid": 2,
352
- "generation_rate": 0.0004,
353
- "pass_rate": 0.5
354
- },
355
- "Trp-Cys": {
356
- "generated": 4,
357
- "valid": 2,
358
- "generation_rate": 0.0004,
359
- "pass_rate": 0.5
360
- },
361
- "head2tail ring": {
362
- "generated": 346,
363
- "valid": 313,
364
- "generation_rate": 0.0346,
365
- "pass_rate": 0.9046
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2340,
372
- "has_connection_rate": 0.234,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1848,
376
- "fraction": 0.7897
377
- },
378
- "2": {
379
- "count": 474,
380
- "fraction": 0.2026
381
- },
382
- "3": {
383
- "count": 18,
384
- "fraction": 0.0077
385
- }
386
- }
387
- },
388
- "per_conn_count_metrics": {
389
- "0": {
390
- "count": 7660,
391
- "validity": 0.9752,
392
- "uniqueness": 0.9941,
393
- "novelty": 0.4605
394
- },
395
- "1": {
396
- "count": 1848,
397
- "validity": 0.6964,
398
- "uniqueness": 0.9992,
399
- "novelty": 0.8235
400
- },
401
- "2": {
402
- "count": 474,
403
- "validity": 0.2827,
404
- "uniqueness": 1.0,
405
- "novelty": 0.7761
406
- },
407
- "3+": {
408
- "count": 18,
409
- "validity": 0.0,
410
- "uniqueness": 0.0
411
- }
412
- }
413
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run1/eval_summary.txt DELETED
@@ -1,56 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:31:58.335464
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 8891/10000 (88.9%)
9
- Uniqueness: 8846/8891 (99.5%)
10
- Novelty: 4740/8846 (53.6%)
11
-
12
- Diversity: 0.8822 (avg MAP4C Jaccard distance)
13
- Molecules: 8846
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.3627, median=0.3337, std=0.1752
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8846
23
- Self-NN: mean=0.3737, median=0.3428
24
-
25
- Structural Summary:
26
- Multi-block rate: 11.4%
27
- Avg peptide len: 6.8
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1614 843 0.1614 52.2%
33
- head2tail ring 346 313 0.0346 90.5%
34
- Ester 320 148 0.0320 46.2%
35
- PG_Amide 305 164 0.0305 53.8%
36
- Disulfide 247 83 0.0247 33.6%
37
- Lanthionin 7 0 0.0007 0.0%
38
- Sulfanilamide 4 2 0.0004 50.0%
39
- Trp-Cys 4 2 0.0004 50.0%
40
- Bz_Stapled 3 0 0.0003 0.0%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 23.4% (2340/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1848 (79.0%)
46
- 2 connection(s): 474 (20.3%)
47
- 3 connection(s): 18 (0.8%)
48
-
49
- Per-Connection-Count Metrics:
50
- Bucket Count Validity Uniqueness Novelty
51
- ---------- -------- ---------- ------------ ----------
52
- 0 conn 7660 97.5% 99.4% 47.3%
53
- 1 conn 1848 69.6% 99.9% 87.1%
54
- 2 conn 474 28.3% 100.0% 79.1%
55
- 3+ conn 18 0.0% 0.0% N/A
56
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run1/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.8891,
4
- "valid_count": 8891,
5
- "monomer_accuracy": {
6
- "mean": 0.145532012987013,
7
- "std": 0.29323806298909383,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7582,
18
- "pct_above_50": 0.1165,
19
- "pct_perfect": 0.0589
20
- }
21
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run1/results.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:4f361e30733cd4fbe1b96d0e6e83fd613248317424f051cecad29937df4c0741
3
- size 11621128
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run2/eval_report.json DELETED
@@ -1,413 +0,0 @@
1
- {
2
- "metadata": {
3
- "timestamp": "2026-03-11T14:53:49.726484",
4
- "num_candidates": 10000,
5
- "dedup_method": "InChIKey",
6
- "num_valid": 8960,
7
- "num_unique": 8924,
8
- "num_novel": 4552
9
- },
10
- "validity": {
11
- "num_raw": 10000,
12
- "num_valid": 8960,
13
- "rate": 0.896
14
- },
15
- "uniqueness": {
16
- "num_valid": 8960,
17
- "num_unique": 8924,
18
- "rate": 0.9959821428571428
19
- },
20
- "novelty": {
21
- "num_unique": 8924,
22
- "num_novel": 4552,
23
- "rate": 0.5100851636037651
24
- },
25
- "diversity": {
26
- "internal_diversity": 0.8834,
27
- "num_molecules": 8924,
28
- "num_pairs": 10000
29
- },
30
- "nn_distance": {
31
- "mean": 0.3635,
32
- "median": 0.332,
33
- "std": 0.1756,
34
- "percentiles": {
35
- "5": 0.0,
36
- "25": 0.2437,
37
- "50": 0.332,
38
- "75": 0.521,
39
- "95": 0.6309
40
- },
41
- "histogram": {
42
- "bin_edges": [
43
- 0.0,
44
- 0.02,
45
- 0.04,
46
- 0.06,
47
- 0.08,
48
- 0.1,
49
- 0.12,
50
- 0.14,
51
- 0.16,
52
- 0.18,
53
- 0.2,
54
- 0.22,
55
- 0.24,
56
- 0.26,
57
- 0.28,
58
- 0.3,
59
- 0.32,
60
- 0.34,
61
- 0.36,
62
- 0.38,
63
- 0.4,
64
- 0.42,
65
- 0.44,
66
- 0.46,
67
- 0.48,
68
- 0.5,
69
- 0.52,
70
- 0.54,
71
- 0.56,
72
- 0.58,
73
- 0.6,
74
- 0.62,
75
- 0.64,
76
- 0.66,
77
- 0.68,
78
- 0.7,
79
- 0.72,
80
- 0.74,
81
- 0.76,
82
- 0.78,
83
- 0.8,
84
- 0.82,
85
- 0.84,
86
- 0.86,
87
- 0.88,
88
- 0.9,
89
- 0.92,
90
- 0.94,
91
- 0.96,
92
- 0.98,
93
- 1.0
94
- ],
95
- "counts": [
96
- 316,
97
- 4,
98
- 11,
99
- 10,
100
- 20,
101
- 11,
102
- 30,
103
- 42,
104
- 95,
105
- 131,
106
- 222,
107
- 297,
108
- 313,
109
- 329,
110
- 295,
111
- 227,
112
- 217,
113
- 143,
114
- 107,
115
- 113,
116
- 126,
117
- 128,
118
- 132,
119
- 116,
120
- 170,
121
- 133,
122
- 174,
123
- 172,
124
- 217,
125
- 179,
126
- 186,
127
- 153,
128
- 89,
129
- 50,
130
- 26,
131
- 11,
132
- 3,
133
- 0,
134
- 2,
135
- 0,
136
- 0,
137
- 0,
138
- 0,
139
- 0,
140
- 0,
141
- 0,
142
- 0,
143
- 0,
144
- 0,
145
- 0
146
- ]
147
- },
148
- "num_query": 5000,
149
- "num_reference": 50000
150
- },
151
- "training_baseline": {
152
- "validity": 1.0,
153
- "uniqueness": 1.0,
154
- "novelty": 0.0,
155
- "internal_diversity": 0.885,
156
- "nn_self_distance": {
157
- "mean": 0.3742,
158
- "median": 0.3442,
159
- "std": 0.1497,
160
- "percentiles": {
161
- "5": 0.1582,
162
- "25": 0.2632,
163
- "50": 0.3442,
164
- "75": 0.501,
165
- "95": 0.6245
166
- },
167
- "histogram": {
168
- "bin_edges": [
169
- 0.0,
170
- 0.02,
171
- 0.04,
172
- 0.06,
173
- 0.08,
174
- 0.1,
175
- 0.12,
176
- 0.14,
177
- 0.16,
178
- 0.18,
179
- 0.2,
180
- 0.22,
181
- 0.24,
182
- 0.26,
183
- 0.28,
184
- 0.3,
185
- 0.32,
186
- 0.34,
187
- 0.36,
188
- 0.38,
189
- 0.4,
190
- 0.42,
191
- 0.44,
192
- 0.46,
193
- 0.48,
194
- 0.5,
195
- 0.52,
196
- 0.54,
197
- 0.56,
198
- 0.58,
199
- 0.6,
200
- 0.62,
201
- 0.64,
202
- 0.66,
203
- 0.68,
204
- 0.7,
205
- 0.72,
206
- 0.74,
207
- 0.76,
208
- 0.78,
209
- 0.8,
210
- 0.82,
211
- 0.84,
212
- 0.86,
213
- 0.88,
214
- 0.9,
215
- 0.92,
216
- 0.94,
217
- 0.96,
218
- 0.98,
219
- 1.0
220
- ],
221
- "counts": [
222
- 5,
223
- 14,
224
- 21,
225
- 26,
226
- 41,
227
- 37,
228
- 49,
229
- 62,
230
- 91,
231
- 138,
232
- 183,
233
- 216,
234
- 319,
235
- 364,
236
- 321,
237
- 308,
238
- 261,
239
- 253,
240
- 179,
241
- 172,
242
- 153,
243
- 141,
244
- 138,
245
- 121,
246
- 131,
247
- 129,
248
- 150,
249
- 149,
250
- 188,
251
- 185,
252
- 180,
253
- 119,
254
- 72,
255
- 45,
256
- 22,
257
- 13,
258
- 1,
259
- 2,
260
- 0,
261
- 0,
262
- 1,
263
- 0,
264
- 0,
265
- 0,
266
- 0,
267
- 0,
268
- 0,
269
- 0,
270
- 0,
271
- 0
272
- ]
273
- },
274
- "num_query": 5000,
275
- "num_reference": 25000
276
- },
277
- "num_total": 383817,
278
- "num_unique": 383817
279
- },
280
- "structural": {
281
- "multi_block_rate": 0.1234,
282
- "peptide_length_mean": 6.83,
283
- "peptide_length_dist": {
284
- "2": 142,
285
- "3": 706,
286
- "4": 4487,
287
- "5": 391,
288
- "6": 408,
289
- "7": 353,
290
- "8": 336,
291
- "9": 417,
292
- "10": 288,
293
- "11": 174,
294
- "12": 146,
295
- "13": 141,
296
- "14": 92,
297
- "15": 118,
298
- "16": 78,
299
- "17": 68,
300
- "18": 66,
301
- "19": 44,
302
- "20": 116,
303
- "21": 62,
304
- "22": 28,
305
- "23": 29,
306
- "24": 33,
307
- "25": 237
308
- }
309
- },
310
- "per_connection_type": {
311
- "num_total_candidates": 10000,
312
- "per_type": {
313
- "Amide": {
314
- "generated": 1598,
315
- "valid": 878,
316
- "generation_rate": 0.1598,
317
- "pass_rate": 0.5494
318
- },
319
- "Bz_Stapled": {
320
- "generated": 5,
321
- "valid": 0,
322
- "generation_rate": 0.0005,
323
- "pass_rate": 0.0
324
- },
325
- "Disulfide": {
326
- "generated": 250,
327
- "valid": 125,
328
- "generation_rate": 0.025,
329
- "pass_rate": 0.5
330
- },
331
- "Ester": {
332
- "generated": 343,
333
- "valid": 180,
334
- "generation_rate": 0.0343,
335
- "pass_rate": 0.5248
336
- },
337
- "Lanthionin": {
338
- "generated": 13,
339
- "valid": 6,
340
- "generation_rate": 0.0013,
341
- "pass_rate": 0.4615
342
- },
343
- "PG_Amide": {
344
- "generated": 310,
345
- "valid": 221,
346
- "generation_rate": 0.031,
347
- "pass_rate": 0.7129
348
- },
349
- "Sulfanilamide": {
350
- "generated": 6,
351
- "valid": 2,
352
- "generation_rate": 0.0006,
353
- "pass_rate": 0.3333
354
- },
355
- "Trp-Cys": {
356
- "generated": 6,
357
- "valid": 5,
358
- "generation_rate": 0.0006,
359
- "pass_rate": 0.8333
360
- },
361
- "head2tail ring": {
362
- "generated": 356,
363
- "valid": 309,
364
- "generation_rate": 0.0356,
365
- "pass_rate": 0.868
366
- }
367
- }
368
- },
369
- "connection_count_dist": {
370
- "num_total": 10000,
371
- "num_with_connection": 2350,
372
- "has_connection_rate": 0.235,
373
- "conditional_dist": {
374
- "1": {
375
- "count": 1834,
376
- "fraction": 0.7804
377
- },
378
- "2": {
379
- "count": 495,
380
- "fraction": 0.2106
381
- },
382
- "3": {
383
- "count": 21,
384
- "fraction": 0.0089
385
- }
386
- }
387
- },
388
- "per_conn_count_metrics": {
389
- "0": {
390
- "count": 7650,
391
- "validity": 0.9744,
392
- "uniqueness": 0.9952,
393
- "novelty": 0.4482
394
- },
395
- "1": {
396
- "count": 1834,
397
- "validity": 0.7012,
398
- "uniqueness": 1.0,
399
- "novelty": 0.8188
400
- },
401
- "2": {
402
- "count": 495,
403
- "validity": 0.4444,
404
- "uniqueness": 1.0,
405
- "novelty": 0.7909
406
- },
407
- "3+": {
408
- "count": 21,
409
- "validity": 0.0,
410
- "uniqueness": 0.0
411
- }
412
- }
413
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run2/eval_summary.txt DELETED
@@ -1,56 +0,0 @@
1
- ============================================================
2
- Generation Evaluation Report
3
- ============================================================
4
- Timestamp: 2026-03-06T21:31:59.901512
5
- Total candidates: 10000
6
-
7
- Core Metrics:
8
- Validity: 8960/10000 (89.6%)
9
- Uniqueness: 8924/8960 (99.6%)
10
- Novelty: 4710/8924 (52.8%)
11
-
12
- Diversity: 0.8823 (avg MAP4C Jaccard distance)
13
- Molecules: 8924
14
- Pairs sampled: 10000
15
-
16
- NN Distance (MAP4C): mean=0.3637, median=0.3340, std=0.1785
17
- Query: 5000, Reference: 50000
18
-
19
- Training Baseline:
20
- Total: 383,817, Unique: 383,817
21
- Uniqueness: 1.0000
22
- IntDiv: 0.8846
23
- Self-NN: mean=0.3737, median=0.3428
24
-
25
- Structural Summary:
26
- Multi-block rate: 12.3%
27
- Avg peptide len: 6.8
28
-
29
- Per-Connection-Type Metrics:
30
- Type Generated Valid Gen Rate Pass Rate
31
- -------------------- ---------- -------- ---------- ----------
32
- Amide 1598 878 0.1598 54.9%
33
- head2tail ring 356 309 0.0356 86.8%
34
- Ester 343 180 0.0343 52.5%
35
- PG_Amide 310 221 0.0310 71.3%
36
- Disulfide 250 125 0.0250 50.0%
37
- Lanthionin 13 6 0.0013 46.2%
38
- Sulfanilamide 6 2 0.0006 33.3%
39
- Trp-Cys 6 5 0.0006 83.3%
40
- Bz_Stapled 5 0 0.0005 0.0%
41
-
42
- Connection Count Distribution:
43
- Has ≥1 connection: 23.5% (2350/10000)
44
- Among connected molecules:
45
- 1 connection(s): 1834 (78.0%)
46
- 2 connection(s): 495 (21.1%)
47
- 3 connection(s): 21 (0.9%)
48
-
49
- Per-Connection-Count Metrics:
50
- Bucket Count Validity Uniqueness Novelty
51
- ---------- -------- ---------- ------------ ----------
52
- 0 conn 7650 97.4% 99.5% 46.0%
53
- 1 conn 1834 70.1% 100.0% 87.0%
54
- 2 conn 495 44.4% 100.0% 80.5%
55
- 3+ conn 21 0.0% 0.0% N/A
56
- ============================================================
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Generation_Model/End2end/GPT_vs_BERT_Infilling/ratio_0.7/GPT/run2/infilling_report.json DELETED
@@ -1,21 +0,0 @@
1
- {
2
- "num_samples": 10000,
3
- "validity_rate": 0.896,
4
- "valid_count": 8960,
5
- "monomer_accuracy": {
6
- "mean": 0.14593076118326118,
7
- "std": 0.2933440663981488,
8
- "median": 0.0,
9
- "n": 10000
10
- },
11
- "constraint_satisfaction": {
12
- "mean": 1.0,
13
- "std": 0.0,
14
- "n": 10000
15
- },
16
- "accuracy_distribution": {
17
- "pct_zero": 0.7572,
18
- "pct_above_50": 0.1162,
19
- "pct_perfect": 0.0598
20
- }
21
- }