Shushant commited on
Commit
5bc9136
Β·
verified Β·
1 Parent(s): 6d2ceab

Training logs (final)

Browse files
Files changed (1) hide show
  1. training_logs/training.log +360 -0
training_logs/training.log CHANGED
@@ -10281,3 +10281,363 @@
10281
  2026-04-04 19:47:54,076 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
10282
  2026-04-04 19:47:54,892 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
10283
  2026-04-04 19:47:55,144 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10281
  2026-04-04 19:47:54,076 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
10282
  2026-04-04 19:47:54,892 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
10283
  2026-04-04 19:47:55,144 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
10284
+ 2026-04-04 19:47:56,716 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
10285
+ 2026-04-04 19:47:56,717 INFO βœ“ Detector pushed β†’ https://huggingface.co/Shushant/ADAL_AI_Detector
10286
+ 2026-04-04 19:47:56,717 INFO Uploading paraphraser β†’ Shushant/ADAL_Paraphrasher …
10287
+ 2026-04-04 19:47:56,826 INFO HTTP Request: POST https://huggingface.co/api/validate-yaml "HTTP/1.1 200 OK"
10288
+ 2026-04-04 19:47:57,458 INFO HTTP Request: POST https://huggingface.co/api/validate-yaml "HTTP/1.1 200 OK"
10289
+ 2026-04-04 19:47:57,605 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_Paraphrasher/preupload/main "HTTP/1.1 200 OK"
10290
+ 2026-04-04 19:47:57,727 INFO HTTP Request: POST https://huggingface.co/Shushant/ADAL_Paraphrasher.git/info/lfs/objects/batch "HTTP/1.1 200 OK"
10291
+ 2026-04-04 19:47:57,839 INFO HTTP Request: GET https://huggingface.co/api/models/Shushant/ADAL_Paraphrasher/xet-write-token/main "HTTP/1.1 200 OK"
10292
+ 2026-04-04 19:48:04,206 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_Paraphrasher/commit/main "HTTP/1.1 200 OK"
10293
+ 2026-04-04 19:48:04,206 INFO βœ“ Paraphraser pushed β†’ https://huggingface.co/Shushant/ADAL_Paraphrasher
10294
+ 2026-04-04 19:48:04,207 INFO ── Hub push complete ──
10295
+
10296
+ 2026-04-04 19:48:17,615 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10297
+ 2026-04-04 19:48:17,615 INFO [Step 156/200] det=skip para=0.0033 raw_rwd=0.1223Β±0.0035 rwd=0.1223
10298
+ 2026-04-04 19:48:17,615 INFO per-gen rewards: chatgpt=0.122 cohere=0.124 cohere-c=0.122 gpt2=0.124 gpt3=0.120 gpt4=0.121 llama-ch=0.123 mistral=0.124 mistral-=0.121 mpt=0.121 mpt-chat=0.121
10299
+ 2026-04-04 19:48:30,947 INFO [Step 157/200] det=skip para=0.0074 raw_rwd=0.1236Β±0.0039 rwd=0.1236
10300
+ 2026-04-04 19:48:30,947 INFO per-gen rewards: chatgpt=0.123 cohere=0.124 cohere-c=0.125 gpt2=0.123 gpt3=0.126 gpt4=0.126 llama-ch=0.122 mistral=0.125 mistral-=0.122 mpt=0.122 mpt-chat=0.125
10301
+ 2026-04-04 19:48:44,317 INFO [Step 158/200] det=skip para=0.0000 raw_rwd=0.1239Β±0.0038 rwd=0.1239
10302
+ 2026-04-04 19:48:44,318 INFO per-gen rewards: chatgpt=0.122 cohere=0.126 cohere-c=0.124 gpt2=0.123 gpt3=0.121 gpt4=0.127 llama-ch=0.124 mistral=0.124 mistral-=0.124 mpt=0.125 mpt-chat=0.122
10303
+ 2026-04-04 19:48:57,735 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10304
+ 2026-04-04 19:48:57,735 INFO [Step 159/200] det=skip para=-0.0000 raw_rwd=0.1222Β±0.0042 rwd=0.1222
10305
+ 2026-04-04 19:48:57,735 INFO per-gen rewards: chatgpt=0.120 cohere=0.122 cohere-c=0.128 gpt2=0.124 gpt3=0.119 gpt4=0.123 llama-ch=0.122 mistral=0.120 mistral-=0.123 mpt=0.122 mpt-chat=0.124
10306
+ 2026-04-04 19:49:11,094 INFO [Step 160/200] det=skip para=0.0000 raw_rwd=0.1230Β±0.0037 rwd=0.1230
10307
+ 2026-04-04 19:49:11,094 INFO per-gen rewards: chatgpt=0.124 cohere=0.120 cohere-c=0.122 gpt2=0.123 gpt3=0.122 gpt4=0.121 llama-ch=0.124 mistral=0.121 mistral-=0.124 mpt=0.124 mpt-chat=0.125
10308
+ 2026-04-04 20:42:35,586 INFO β”Œβ”€ Generator AUROC ────────────────────────────┐
10309
+ 2026-04-04 20:42:35,587 INFO β”‚ MACRO_AVG 0.9951 β—„
10310
+ 2026-04-04 20:42:35,587 INFO β”‚ chatgpt 0.9991
10311
+ 2026-04-04 20:42:35,587 INFO β”‚ cohere 0.9852
10312
+ 2026-04-04 20:42:35,587 INFO β”‚ cohere-chat 0.9934
10313
+ 2026-04-04 20:42:35,587 INFO β”‚ gpt2 0.9954
10314
+ 2026-04-04 20:42:35,587 INFO β”‚ gpt3 0.9982
10315
+ 2026-04-04 20:42:35,587 INFO β”‚ gpt4 0.9995
10316
+ 2026-04-04 20:42:35,587 INFO β”‚ llama-chat 0.9994
10317
+ 2026-04-04 20:42:35,587 INFO β”‚ mistral 0.9913
10318
+ 2026-04-04 20:42:35,587 INFO β”‚ mistral-chat 0.9994
10319
+ 2026-04-04 20:42:35,587 INFO β”‚ mpt 0.9865
10320
+ 2026-04-04 20:42:35,587 INFO β”‚ mpt-chat 0.9991
10321
+ 2026-04-04 20:42:35,587 INFO β”‚ (prev best) 0.9951
10322
+ 2026-04-04 20:42:35,587 INFO β”œβ”€ Attack AUROC (robustness) ──────────────────
10323
+ 2026-04-04 20:42:35,587 INFO β”‚ article_deletion 0.9993 ~
10324
+ 2026-04-04 20:42:35,587 INFO β”‚ homoglyphs 0.9994 ~
10325
+ 2026-04-04 20:42:35,587 INFO β”‚ misspelling 0.9996 ~
10326
+ 2026-04-04 20:42:35,587 INFO β”‚ no_attack 0.9994 ~
10327
+ 2026-04-04 20:42:35,587 INFO β”‚ synonym_replacement 0.9990 ~
10328
+ 2026-04-04 20:42:35,587 INFO β”‚ t5_paraphrase 1.0000 ~
10329
+ 2026-04-04 20:42:35,587 INFO β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
10330
+ 2026-04-04 20:42:35,587 INFO No improvement (1/25) best=0.9951
10331
+
10332
+ 2026-04-04 20:42:49,194 INFO [Step 161/200] det=skip para=-0.0000 raw_rwd=0.1229Β±0.0041 rwd=0.1229
10333
+ 2026-04-04 20:42:49,194 INFO per-gen rewards: chatgpt=0.118 cohere=0.123 cohere-c=0.124 gpt2=0.125 gpt3=0.124 gpt4=0.124 llama-ch=0.123 mistral=0.120 mistral-=0.124 mpt=0.124 mpt-chat=0.122
10334
+ 2026-04-04 20:43:03,457 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10335
+ 2026-04-04 20:43:03,457 INFO [Step 162/200] det=skip para=0.0000 raw_rwd=0.1228Β±0.0039 rwd=0.1228
10336
+ 2026-04-04 20:43:03,457 INFO per-gen rewards: chatgpt=0.121 cohere=0.121 cohere-c=0.121 gpt2=0.122 gpt3=0.124 gpt4=0.121 llama-ch=0.125 mistral=0.124 mistral-=0.122 mpt=0.122 mpt-chat=0.123
10337
+ 2026-04-04 20:43:16,821 INFO [Step 163/200] det=skip para=-0.0000 raw_rwd=0.1229Β±0.0043 rwd=0.1229
10338
+ 2026-04-04 20:43:16,821 INFO per-gen rewards: chatgpt=0.121 cohere=0.124 cohere-c=0.122 gpt2=0.121 gpt3=0.123 gpt4=0.120 llama-ch=0.122 mistral=0.124 mistral-=0.125 mpt=0.122 mpt-chat=0.124
10339
+ 2026-04-04 20:43:30,206 INFO [Step 164/200] det=skip para=0.0000 raw_rwd=0.1237Β±0.0032 rwd=0.1237
10340
+ 2026-04-04 20:43:30,206 INFO per-gen rewards: chatgpt=0.126 cohere=0.124 cohere-c=0.124 gpt2=0.122 gpt3=0.121 gpt4=0.122 llama-ch=0.123 mistral=0.125 mistral-=0.123 mpt=0.125 mpt-chat=0.125
10341
+ 2026-04-04 20:43:43,653 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10342
+ 2026-04-04 20:43:43,653 INFO [Step 165/200] det=skip para=0.0000 raw_rwd=0.1230Β±0.0041 rwd=0.1230
10343
+ 2026-04-04 20:43:43,653 INFO per-gen rewards: chatgpt=0.120 cohere=0.126 cohere-c=0.121 gpt2=0.124 gpt3=0.126 gpt4=0.123 llama-ch=0.123 mistral=0.121 mistral-=0.121 mpt=0.126 mpt-chat=0.122
10344
+ 2026-04-04 21:37:07,034 INFO β”Œβ”€ Generator AUROC ────────────────────────────┐
10345
+ 2026-04-04 21:37:07,034 INFO β”‚ MACRO_AVG 0.9951 β—„
10346
+ 2026-04-04 21:37:07,034 INFO β”‚ chatgpt 0.9991
10347
+ 2026-04-04 21:37:07,034 INFO β”‚ cohere 0.9852
10348
+ 2026-04-04 21:37:07,034 INFO β”‚ cohere-chat 0.9934
10349
+ 2026-04-04 21:37:07,034 INFO β”‚ gpt2 0.9954
10350
+ 2026-04-04 21:37:07,034 INFO β”‚ gpt3 0.9982
10351
+ 2026-04-04 21:37:07,034 INFO β”‚ gpt4 0.9995
10352
+ 2026-04-04 21:37:07,034 INFO β”‚ llama-chat 0.9994
10353
+ 2026-04-04 21:37:07,034 INFO β”‚ mistral 0.9913
10354
+ 2026-04-04 21:37:07,034 INFO β”‚ mistral-chat 0.9994
10355
+ 2026-04-04 21:37:07,034 INFO β”‚ mpt 0.9865
10356
+ 2026-04-04 21:37:07,034 INFO β”‚ mpt-chat 0.9991
10357
+ 2026-04-04 21:37:07,035 INFO β”‚ (prev best) 0.9951
10358
+ 2026-04-04 21:37:07,035 INFO β”œβ”€ Attack AUROC (robustness) ──────────────────
10359
+ 2026-04-04 21:37:07,035 INFO β”‚ article_deletion 0.9993 ~
10360
+ 2026-04-04 21:37:07,035 INFO β”‚ homoglyphs 0.9994 ~
10361
+ 2026-04-04 21:37:07,035 INFO β”‚ misspelling 0.9996 ~
10362
+ 2026-04-04 21:37:07,035 INFO β”‚ no_attack 0.9994 ~
10363
+ 2026-04-04 21:37:07,035 INFO β”‚ synonym_replacement 0.9990 ~
10364
+ 2026-04-04 21:37:07,035 INFO β”‚ t5_paraphrase 1.0000 ~
10365
+ 2026-04-04 21:37:07,035 INFO β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
10366
+ 2026-04-04 21:37:07,035 INFO No improvement (2/25) best=0.9951
10367
+
10368
+ 2026-04-04 21:37:20,516 INFO [Step 166/200] det=skip para=0.0000 raw_rwd=0.1228Β±0.0039 rwd=0.1228
10369
+ 2026-04-04 21:37:20,516 INFO per-gen rewards: chatgpt=0.123 cohere=0.123 cohere-c=0.123 gpt2=0.122 gpt3=0.122 gpt4=0.120 llama-ch=0.124 mistral=0.125 mistral-=0.121 mpt=0.124 mpt-chat=0.122
10370
+ 2026-04-04 21:37:33,705 INFO [Step 167/200] det=skip para=0.0000 raw_rwd=0.1231Β±0.0041 rwd=0.1231
10371
+ 2026-04-04 21:37:33,706 INFO per-gen rewards: chatgpt=0.122 cohere=0.125 cohere-c=0.125 gpt2=0.126 gpt3=0.124 gpt4=0.123 llama-ch=0.120 mistral=0.122 mistral-=0.123 mpt=0.121 mpt-chat=0.124
10372
+ 2026-04-04 21:37:46,985 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10373
+ 2026-04-04 21:37:46,985 INFO [Step 168/200] det=skip para=0.0000 raw_rwd=0.1229Β±0.0036 rwd=0.1229
10374
+ 2026-04-04 21:37:46,985 INFO per-gen rewards: chatgpt=0.124 cohere=0.122 cohere-c=0.122 gpt2=0.124 gpt3=0.120 gpt4=0.125 llama-ch=0.122 mistral=0.123 mistral-=0.123 mpt=0.123 mpt-chat=0.124
10375
+ 2026-04-04 21:38:00,385 INFO [Step 169/200] det=skip para=0.0000 raw_rwd=0.1233Β±0.0040 rwd=0.1233
10376
+ 2026-04-04 21:38:00,385 INFO per-gen rewards: chatgpt=0.123 cohere=0.125 cohere-c=0.125 gpt2=0.122 gpt3=0.126 gpt4=0.123 llama-ch=0.122 mistral=0.125 mistral-=0.123 mpt=0.124 mpt-chat=0.122
10377
+ 2026-04-04 21:38:13,768 INFO [Step 170/200] det=skip para=-0.0000 raw_rwd=0.1223Β±0.0041 rwd=0.1223
10378
+ 2026-04-04 21:38:13,769 INFO per-gen rewards: chatgpt=0.119 cohere=0.122 cohere-c=0.124 gpt2=0.121 gpt3=0.126 gpt4=0.120 llama-ch=0.122 mistral=0.125 mistral-=0.123 mpt=0.123 mpt-chat=0.120
10379
+ 2026-04-04 22:31:35,886 INFO β”Œβ”€ Generator AUROC ────────────────────────────┐
10380
+ 2026-04-04 22:31:35,887 INFO β”‚ MACRO_AVG 0.9951 β—„
10381
+ 2026-04-04 22:31:35,887 INFO β”‚ chatgpt 0.9991
10382
+ 2026-04-04 22:31:35,887 INFO β”‚ cohere 0.9852
10383
+ 2026-04-04 22:31:35,887 INFO β”‚ cohere-chat 0.9934
10384
+ 2026-04-04 22:31:35,887 INFO β”‚ gpt2 0.9954
10385
+ 2026-04-04 22:31:35,887 INFO β”‚ gpt3 0.9982
10386
+ 2026-04-04 22:31:35,887 INFO β”‚ gpt4 0.9995
10387
+ 2026-04-04 22:31:35,887 INFO β”‚ llama-chat 0.9994
10388
+ 2026-04-04 22:31:35,887 INFO β”‚ mistral 0.9913
10389
+ 2026-04-04 22:31:35,887 INFO β”‚ mistral-chat 0.9994
10390
+ 2026-04-04 22:31:35,887 INFO β”‚ mpt 0.9865
10391
+ 2026-04-04 22:31:35,887 INFO β”‚ mpt-chat 0.9991
10392
+ 2026-04-04 22:31:35,887 INFO β”‚ (prev best) 0.9951
10393
+ 2026-04-04 22:31:35,887 INFO β”œβ”€ Attack AUROC (robustness) ──────────────────
10394
+ 2026-04-04 22:31:35,887 INFO β”‚ article_deletion 0.9993 ~
10395
+ 2026-04-04 22:31:35,887 INFO β”‚ homoglyphs 0.9994 ~
10396
+ 2026-04-04 22:31:35,887 INFO β”‚ misspelling 0.9996 ~
10397
+ 2026-04-04 22:31:35,887 INFO β”‚ no_attack 0.9994 ~
10398
+ 2026-04-04 22:31:35,887 INFO β”‚ synonym_replacement 0.9990 ~
10399
+ 2026-04-04 22:31:35,887 INFO β”‚ t5_paraphrase 1.0000 ~
10400
+ 2026-04-04 22:31:35,887 INFO β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
10401
+ 2026-04-04 22:31:35,887 INFO No improvement (3/25) best=0.9951
10402
+
10403
+ 2026-04-04 22:31:49,460 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10404
+ 2026-04-04 22:31:49,461 INFO [Step 171/200] det=skip para=-0.0000 raw_rwd=0.1222Β±0.0040 rwd=0.1222
10405
+ 2026-04-04 22:31:49,461 INFO per-gen rewards: chatgpt=0.120 cohere=0.122 cohere-c=0.122 gpt2=0.124 gpt3=0.123 gpt4=0.119 llama-ch=0.124 mistral=0.123 mistral-=0.121 mpt=0.121 mpt-chat=0.122
10406
+ 2026-04-04 22:32:02,869 INFO [Step 172/200] det=skip para=-0.0000 raw_rwd=0.1228Β±0.0046 rwd=0.1228
10407
+ 2026-04-04 22:32:02,869 INFO per-gen rewards: chatgpt=0.122 cohere=0.126 cohere-c=0.121 gpt2=0.123 gpt3=0.118 gpt4=0.127 llama-ch=0.126 mistral=0.121 mistral-=0.124 mpt=0.122 mpt-chat=0.122
10408
+ 2026-04-04 22:32:16,268 INFO [Step 173/200] det=skip para=-0.0000 raw_rwd=0.1232Β±0.0039 rwd=0.1232
10409
+ 2026-04-04 22:32:16,268 INFO per-gen rewards: chatgpt=0.121 cohere=0.125 cohere-c=0.124 gpt2=0.123 gpt3=0.123 gpt4=0.123 llama-ch=0.121 mistral=0.124 mistral-=0.126 mpt=0.124 mpt-chat=0.122
10410
+ 2026-04-04 22:32:29,726 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10411
+ 2026-04-04 22:32:29,726 INFO [Step 174/200] det=skip para=0.0000 raw_rwd=0.1234Β±0.0052 rwd=0.1234
10412
+ 2026-04-04 22:32:29,726 INFO per-gen rewards: chatgpt=0.117 cohere=0.125 cohere-c=0.122 gpt2=0.122 gpt3=0.124 gpt4=0.124 llama-ch=0.125 mistral=0.125 mistral-=0.125 mpt=0.120 mpt-chat=0.126
10413
+ 2026-04-04 22:32:43,094 INFO [Step 175/200] det=skip para=0.0000 raw_rwd=0.1226Β±0.0037 rwd=0.1226
10414
+ 2026-04-04 22:32:43,094 INFO per-gen rewards: chatgpt=0.120 cohere=0.122 cohere-c=0.123 gpt2=0.123 gpt3=0.124 gpt4=0.123 llama-ch=0.123 mistral=0.122 mistral-=0.123 mpt=0.119 mpt-chat=0.126
10415
+ 2026-04-04 23:26:04,656 INFO β”Œβ”€ Generator AUROC ────────────────────────────┐
10416
+ 2026-04-04 23:26:04,657 INFO β”‚ MACRO_AVG 0.9951 β—„
10417
+ 2026-04-04 23:26:04,657 INFO β”‚ chatgpt 0.9991
10418
+ 2026-04-04 23:26:04,657 INFO β”‚ cohere 0.9852
10419
+ 2026-04-04 23:26:04,657 INFO β”‚ cohere-chat 0.9934
10420
+ 2026-04-04 23:26:04,657 INFO β”‚ gpt2 0.9954
10421
+ 2026-04-04 23:26:04,657 INFO β”‚ gpt3 0.9982
10422
+ 2026-04-04 23:26:04,657 INFO β”‚ gpt4 0.9995
10423
+ 2026-04-04 23:26:04,657 INFO β”‚ llama-chat 0.9994
10424
+ 2026-04-04 23:26:04,657 INFO β”‚ mistral 0.9913
10425
+ 2026-04-04 23:26:04,657 INFO β”‚ mistral-chat 0.9994
10426
+ 2026-04-04 23:26:04,657 INFO β”‚ mpt 0.9865
10427
+ 2026-04-04 23:26:04,657 INFO β”‚ mpt-chat 0.9991
10428
+ 2026-04-04 23:26:04,657 INFO β”‚ (prev best) 0.9951
10429
+ 2026-04-04 23:26:04,657 INFO β”œβ”€ Attack AUROC (robustness) ──────────────────
10430
+ 2026-04-04 23:26:04,657 INFO β”‚ article_deletion 0.9993 ~
10431
+ 2026-04-04 23:26:04,657 INFO β”‚ homoglyphs 0.9994 ~
10432
+ 2026-04-04 23:26:04,657 INFO β”‚ misspelling 0.9996 ~
10433
+ 2026-04-04 23:26:04,657 INFO β”‚ no_attack 0.9994 ~
10434
+ 2026-04-04 23:26:04,657 INFO β”‚ synonym_replacement 0.9990 ~
10435
+ 2026-04-04 23:26:04,657 INFO β”‚ t5_paraphrase 1.0000 ~
10436
+ 2026-04-04 23:26:04,657 INFO β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
10437
+ 2026-04-04 23:26:04,658 INFO No improvement (4/25) best=0.9951
10438
+
10439
+ 2026-04-04 23:26:18,261 INFO [Step 176/200] det=skip para=0.0000 raw_rwd=0.1234Β±0.0040 rwd=0.1234
10440
+ 2026-04-04 23:26:18,262 INFO per-gen rewards: chatgpt=0.125 cohere=0.124 cohere-c=0.118 gpt2=0.123 gpt3=0.124 gpt4=0.121 llama-ch=0.125 mistral=0.123 mistral-=0.124 mpt=0.123 mpt-chat=0.125
10441
+ 2026-04-04 23:26:31,632 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10442
+ 2026-04-04 23:26:31,632 INFO [Step 177/200] det=skip para=-0.0000 raw_rwd=0.1224Β±0.0037 rwd=0.1224
10443
+ 2026-04-04 23:26:31,632 INFO per-gen rewards: chatgpt=0.121 cohere=0.122 cohere-c=0.125 gpt2=0.124 gpt3=0.123 gpt4=0.121 llama-ch=0.121 mistral=0.123 mistral-=0.122 mpt=0.123 mpt-chat=0.122
10444
+ 2026-04-04 23:26:45,047 INFO [Step 178/200] det=skip para=-0.0000 raw_rwd=0.1226Β±0.0038 rwd=0.1226
10445
+ 2026-04-04 23:26:45,047 INFO per-gen rewards: chatgpt=0.120 cohere=0.122 cohere-c=0.122 gpt2=0.122 gpt3=0.122 gpt4=0.123 llama-ch=0.124 mistral=0.124 mistral-=0.121 mpt=0.125 mpt-chat=0.123
10446
+ 2026-04-04 23:26:58,376 INFO [Step 179/200] det=skip para=0.0000 raw_rwd=0.1234Β±0.0035 rwd=0.1234
10447
+ 2026-04-04 23:26:58,376 INFO per-gen rewards: chatgpt=0.121 cohere=0.128 cohere-c=0.124 gpt2=0.124 gpt3=0.120 gpt4=0.124 llama-ch=0.124 mistral=0.122 mistral-=0.122 mpt=0.124 mpt-chat=0.125
10448
+ 2026-04-04 23:27:11,720 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10449
+ 2026-04-04 23:27:11,720 INFO [Step 180/200] det=skip para=0.0000 raw_rwd=0.1232Β±0.0036 rwd=0.1232
10450
+ 2026-04-04 23:27:11,720 INFO per-gen rewards: chatgpt=0.127 cohere=0.124 cohere-c=0.123 gpt2=0.124 gpt3=0.123 gpt4=0.120 llama-ch=0.123 mistral=0.122 mistral-=0.123 mpt=0.125 mpt-chat=0.121
10451
+ 2026-04-05 00:20:32,147 INFO β”Œβ”€ Generator AUROC ────────────────────────────┐
10452
+ 2026-04-05 00:20:32,148 INFO β”‚ MACRO_AVG 0.9951 β—„
10453
+ 2026-04-05 00:20:32,148 INFO β”‚ chatgpt 0.9991
10454
+ 2026-04-05 00:20:32,148 INFO β”‚ cohere 0.9852
10455
+ 2026-04-05 00:20:32,148 INFO β”‚ cohere-chat 0.9934
10456
+ 2026-04-05 00:20:32,148 INFO β”‚ gpt2 0.9954
10457
+ 2026-04-05 00:20:32,148 INFO β”‚ gpt3 0.9982
10458
+ 2026-04-05 00:20:32,148 INFO β”‚ gpt4 0.9995
10459
+ 2026-04-05 00:20:32,148 INFO β”‚ llama-chat 0.9994
10460
+ 2026-04-05 00:20:32,148 INFO β”‚ mistral 0.9913
10461
+ 2026-04-05 00:20:32,148 INFO β”‚ mistral-chat 0.9994
10462
+ 2026-04-05 00:20:32,148 INFO β”‚ mpt 0.9865
10463
+ 2026-04-05 00:20:32,148 INFO β”‚ mpt-chat 0.9991
10464
+ 2026-04-05 00:20:32,148 INFO β”‚ (prev best) 0.9951
10465
+ 2026-04-05 00:20:32,148 INFO β”œβ”€ Attack AUROC (robustness) ──────────────────
10466
+ 2026-04-05 00:20:32,148 INFO β”‚ article_deletion 0.9992 ~
10467
+ 2026-04-05 00:20:32,148 INFO β”‚ homoglyphs 0.9994 ~
10468
+ 2026-04-05 00:20:32,148 INFO β”‚ misspelling 0.9996 ~
10469
+ 2026-04-05 00:20:32,148 INFO β”‚ no_attack 0.9994 ~
10470
+ 2026-04-05 00:20:32,148 INFO β”‚ synonym_replacement 0.9990 ~
10471
+ 2026-04-05 00:20:32,148 INFO β”‚ t5_paraphrase 1.0000 ~
10472
+ 2026-04-05 00:20:32,148 INFO β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
10473
+ 2026-04-05 00:20:32,148 INFO No improvement (5/25) best=0.9951
10474
+
10475
+ 2026-04-05 00:20:45,552 INFO [Step 181/200] det=skip para=0.0000 raw_rwd=0.1229Β±0.0043 rwd=0.1229
10476
+ 2026-04-05 00:20:45,552 INFO per-gen rewards: chatgpt=0.121 cohere=0.122 cohere-c=0.124 gpt2=0.123 gpt3=0.124 gpt4=0.126 llama-ch=0.122 mistral=0.124 mistral-=0.124 mpt=0.122 mpt-chat=0.121
10477
+ 2026-04-05 00:20:58,718 INFO [Step 182/200] det=skip para=-0.0000 raw_rwd=0.1224Β±0.0036 rwd=0.1224
10478
+ 2026-04-05 00:20:58,718 INFO per-gen rewards: chatgpt=0.122 cohere=0.125 cohere-c=0.124 gpt2=0.122 gpt3=0.121 gpt4=0.120 llama-ch=0.119 mistral=0.125 mistral-=0.123 mpt=0.122 mpt-chat=0.123
10479
+ 2026-04-05 00:21:11,906 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10480
+ 2026-04-05 00:21:11,906 INFO [Step 183/200] det=skip para=-0.0000 raw_rwd=0.1229Β±0.0044 rwd=0.1229
10481
+ 2026-04-05 00:21:11,906 INFO per-gen rewards: chatgpt=0.121 cohere=0.121 cohere-c=0.120 gpt2=0.121 gpt3=0.123 gpt4=0.121 llama-ch=0.124 mistral=0.124 mistral-=0.123 mpt=0.125 mpt-chat=0.124
10482
+ 2026-04-05 00:21:25,176 INFO [Step 184/200] det=skip para=-0.0000 raw_rwd=0.1226Β±0.0042 rwd=0.1226
10483
+ 2026-04-05 00:21:25,176 INFO per-gen rewards: chatgpt=0.125 cohere=0.124 cohere-c=0.121 gpt2=0.121 gpt3=0.122 gpt4=0.124 llama-ch=0.123 mistral=0.122 mistral-=0.123 mpt=0.122 mpt-chat=0.122
10484
+ 2026-04-05 00:21:38,467 INFO [Step 185/200] det=skip para=0.0000 raw_rwd=0.1228Β±0.0034 rwd=0.1228
10485
+ 2026-04-05 00:21:38,467 INFO per-gen rewards: chatgpt=0.124 cohere=0.122 cohere-c=0.121 gpt2=0.122 gpt3=0.125 gpt4=0.120 llama-ch=0.125 mistral=0.124 mistral-=0.123 mpt=0.123 mpt-chat=0.122
10486
+ 2026-04-05 01:14:59,637 INFO β”Œβ”€ Generator AUROC ────────────────────────────┐
10487
+ 2026-04-05 01:14:59,638 INFO β”‚ MACRO_AVG 0.9951 β—„
10488
+ 2026-04-05 01:14:59,638 INFO β”‚ chatgpt 0.9991
10489
+ 2026-04-05 01:14:59,638 INFO β”‚ cohere 0.9852
10490
+ 2026-04-05 01:14:59,638 INFO β”‚ cohere-chat 0.9934
10491
+ 2026-04-05 01:14:59,638 INFO β”‚ gpt2 0.9954
10492
+ 2026-04-05 01:14:59,638 INFO β”‚ gpt3 0.9982
10493
+ 2026-04-05 01:14:59,638 INFO β”‚ gpt4 0.9995
10494
+ 2026-04-05 01:14:59,638 INFO β”‚ llama-chat 0.9994
10495
+ 2026-04-05 01:14:59,638 INFO β”‚ mistral 0.9913
10496
+ 2026-04-05 01:14:59,638 INFO β”‚ mistral-chat 0.9994
10497
+ 2026-04-05 01:14:59,638 INFO β”‚ mpt 0.9865
10498
+ 2026-04-05 01:14:59,638 INFO β”‚ mpt-chat 0.9991
10499
+ 2026-04-05 01:14:59,638 INFO β”‚ (prev best) 0.9951
10500
+ 2026-04-05 01:14:59,638 INFO β”œβ”€ Attack AUROC (robustness) ──────────────────
10501
+ 2026-04-05 01:14:59,638 INFO β”‚ article_deletion 0.9993 ~
10502
+ 2026-04-05 01:14:59,638 INFO β”‚ homoglyphs 0.9994 ~
10503
+ 2026-04-05 01:14:59,638 INFO β”‚ misspelling 0.9996 ~
10504
+ 2026-04-05 01:14:59,638 INFO β”‚ no_attack 0.9994 ~
10505
+ 2026-04-05 01:14:59,638 INFO β”‚ synonym_replacement 0.9990 ~
10506
+ 2026-04-05 01:14:59,638 INFO β”‚ t5_paraphrase 1.0000 ~
10507
+ 2026-04-05 01:14:59,638 INFO β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
10508
+ 2026-04-05 01:14:59,638 INFO No improvement (6/25) best=0.9951
10509
+
10510
+ 2026-04-05 01:15:13,242 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10511
+ 2026-04-05 01:15:13,243 INFO [Step 186/200] det=skip para=-0.0000 raw_rwd=0.1240Β±0.0048 rwd=0.1240
10512
+ 2026-04-05 01:15:13,243 INFO per-gen rewards: chatgpt=0.124 cohere=0.120 cohere-c=0.127 gpt2=0.125 gpt3=0.124 gpt4=0.122 llama-ch=0.122 mistral=0.124 mistral-=0.124 mpt=0.124 mpt-chat=0.126
10513
+ 2026-04-05 01:15:26,614 INFO [Step 187/200] det=skip para=0.0000 raw_rwd=0.1220Β±0.0045 rwd=0.1220
10514
+ 2026-04-05 01:15:26,614 INFO per-gen rewards: chatgpt=0.119 cohere=0.122 cohere-c=0.122 gpt2=0.122 gpt3=0.125 gpt4=0.121 llama-ch=0.121 mistral=0.120 mistral-=0.125 mpt=0.122 mpt-chat=0.122
10515
+ 2026-04-05 01:15:40,039 INFO [Step 188/200] det=skip para=-0.0000 raw_rwd=0.1233Β±0.0034 rwd=0.1233
10516
+ 2026-04-05 01:15:40,040 INFO per-gen rewards: chatgpt=0.126 cohere=0.125 cohere-c=0.124 gpt2=0.122 gpt3=0.121 gpt4=0.122 llama-ch=0.124 mistral=0.124 mistral-=0.123 mpt=0.123 mpt-chat=0.123
10517
+ 2026-04-05 01:15:53,405 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10518
+ 2026-04-05 01:15:53,405 INFO [Step 189/200] det=skip para=-0.0000 raw_rwd=0.1224Β±0.0040 rwd=0.1224
10519
+ 2026-04-05 01:15:53,405 INFO per-gen rewards: chatgpt=0.127 cohere=0.125 cohere-c=0.120 gpt2=0.122 gpt3=0.122 gpt4=0.122 llama-ch=0.119 mistral=0.123 mistral-=0.124 mpt=0.125 mpt-chat=0.120
10520
+ 2026-04-05 01:16:06,621 INFO [Step 190/200] det=skip para=-0.0000 raw_rwd=0.1233Β±0.0042 rwd=0.1233
10521
+ 2026-04-05 01:16:06,621 INFO per-gen rewards: chatgpt=0.126 cohere=0.124 cohere-c=0.125 gpt2=0.124 gpt3=0.122 gpt4=0.123 llama-ch=0.124 mistral=0.121 mistral-=0.125 mpt=0.123 mpt-chat=0.121
10522
+ 2026-04-05 02:09:28,551 INFO β”Œβ”€ Generator AUROC ────────────────────────────┐
10523
+ 2026-04-05 02:09:28,552 INFO β”‚ MACRO_AVG 0.9951 β—„
10524
+ 2026-04-05 02:09:28,552 INFO β”‚ chatgpt 0.9991
10525
+ 2026-04-05 02:09:28,552 INFO β”‚ cohere 0.9852
10526
+ 2026-04-05 02:09:28,552 INFO β”‚ cohere-chat 0.9934
10527
+ 2026-04-05 02:09:28,552 INFO β”‚ gpt2 0.9954
10528
+ 2026-04-05 02:09:28,552 INFO β”‚ gpt3 0.9982
10529
+ 2026-04-05 02:09:28,552 INFO β”‚ gpt4 0.9995
10530
+ 2026-04-05 02:09:28,552 INFO β”‚ llama-chat 0.9994
10531
+ 2026-04-05 02:09:28,552 INFO β”‚ mistral 0.9913
10532
+ 2026-04-05 02:09:28,552 INFO β”‚ mistral-chat 0.9994
10533
+ 2026-04-05 02:09:28,552 INFO β”‚ mpt 0.9865
10534
+ 2026-04-05 02:09:28,552 INFO β”‚ mpt-chat 0.9991
10535
+ 2026-04-05 02:09:28,552 INFO β”‚ (prev best) 0.9951
10536
+ 2026-04-05 02:09:28,552 INFO β”œβ”€ Attack AUROC (robustness) ──────────────────
10537
+ 2026-04-05 02:09:28,552 INFO β”‚ article_deletion 0.9993 ~
10538
+ 2026-04-05 02:09:28,552 INFO β”‚ homoglyphs 0.9994 ~
10539
+ 2026-04-05 02:09:28,553 INFO β”‚ misspelling 0.9996 ~
10540
+ 2026-04-05 02:09:28,553 INFO β”‚ no_attack 0.9994 ~
10541
+ 2026-04-05 02:09:28,553 INFO β”‚ synonym_replacement 0.9990 ~
10542
+ 2026-04-05 02:09:28,553 INFO β”‚ t5_paraphrase 1.0000 ~
10543
+ 2026-04-05 02:09:28,553 INFO β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
10544
+ 2026-04-05 02:09:28,553 INFO No improvement (7/25) best=0.9951
10545
+
10546
+ 2026-04-05 02:09:42,086 INFO [Step 191/200] det=skip para=-0.0000 raw_rwd=0.1221Β±0.0044 rwd=0.1221
10547
+ 2026-04-05 02:09:42,087 INFO per-gen rewards: chatgpt=0.119 cohere=0.120 cohere-c=0.122 gpt2=0.124 gpt3=0.124 gpt4=0.123 llama-ch=0.123 mistral=0.121 mistral-=0.124 mpt=0.121 mpt-chat=0.122
10548
+ 2026-04-05 02:09:55,330 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10549
+ 2026-04-05 02:09:55,330 INFO [Step 192/200] det=skip para=-0.0000 raw_rwd=0.1231Β±0.0038 rwd=0.1231
10550
+ 2026-04-05 02:09:55,330 INFO per-gen rewards: chatgpt=0.122 cohere=0.123 cohere-c=0.121 gpt2=0.124 gpt3=0.122 gpt4=0.124 llama-ch=0.123 mistral=0.123 mistral-=0.124 mpt=0.123 mpt-chat=0.123
10551
+ 2026-04-05 02:10:08,654 INFO [Step 193/200] det=skip para=-0.0000 raw_rwd=0.1236Β±0.0046 rwd=0.1236
10552
+ 2026-04-05 02:10:08,654 INFO per-gen rewards: chatgpt=0.119 cohere=0.124 cohere-c=0.122 gpt2=0.125 gpt3=0.124 gpt4=0.123 llama-ch=0.125 mistral=0.123 mistral-=0.125 mpt=0.126 mpt-chat=0.122
10553
+ 2026-04-05 02:10:21,813 INFO [Step 194/200] det=skip para=0.0000 raw_rwd=0.1223Β±0.0036 rwd=0.1223
10554
+ 2026-04-05 02:10:21,814 INFO per-gen rewards: chatgpt=0.122 cohere=0.121 cohere-c=0.121 gpt2=0.126 gpt3=0.123 gpt4=0.121 llama-ch=0.124 mistral=0.120 mistral-=0.123 mpt=0.123 mpt-chat=0.120
10555
+ 2026-04-05 02:10:34,953 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10556
+ 2026-04-05 02:10:34,953 INFO [Step 195/200] det=skip para=0.0079 raw_rwd=0.1229Β±0.0044 rwd=0.1229
10557
+ 2026-04-05 02:10:34,954 INFO per-gen rewards: chatgpt=0.124 cohere=0.122 cohere-c=0.123 gpt2=0.123 gpt3=0.122 gpt4=0.119 llama-ch=0.122 mistral=0.124 mistral-=0.124 mpt=0.123 mpt-chat=0.125
10558
+ 2026-04-05 03:03:56,824 INFO β”Œβ”€ Generator AUROC ────────────────────────────┐
10559
+ 2026-04-05 03:03:56,825 INFO β”‚ MACRO_AVG 0.9951 β—„
10560
+ 2026-04-05 03:03:56,825 INFO β”‚ chatgpt 0.9991
10561
+ 2026-04-05 03:03:56,825 INFO β”‚ cohere 0.9852
10562
+ 2026-04-05 03:03:56,825 INFO β”‚ cohere-chat 0.9934
10563
+ 2026-04-05 03:03:56,825 INFO β”‚ gpt2 0.9954
10564
+ 2026-04-05 03:03:56,825 INFO β”‚ gpt3 0.9982
10565
+ 2026-04-05 03:03:56,825 INFO β”‚ gpt4 0.9995
10566
+ 2026-04-05 03:03:56,825 INFO β”‚ llama-chat 0.9994
10567
+ 2026-04-05 03:03:56,825 INFO β”‚ mistral 0.9913
10568
+ 2026-04-05 03:03:56,825 INFO β”‚ mistral-chat 0.9994
10569
+ 2026-04-05 03:03:56,825 INFO β”‚ mpt 0.9865
10570
+ 2026-04-05 03:03:56,825 INFO β”‚ mpt-chat 0.9991
10571
+ 2026-04-05 03:03:56,825 INFO β”‚ (prev best) 0.9951
10572
+ 2026-04-05 03:03:56,825 INFO β”œβ”€ Attack AUROC (robustness) ──────────────────
10573
+ 2026-04-05 03:03:56,825 INFO β”‚ article_deletion 0.9993 ~
10574
+ 2026-04-05 03:03:56,825 INFO β”‚ homoglyphs 0.9994 ~
10575
+ 2026-04-05 03:03:56,826 INFO β”‚ misspelling 0.9996 ~
10576
+ 2026-04-05 03:03:56,826 INFO β”‚ no_attack 0.9994 ~
10577
+ 2026-04-05 03:03:56,826 INFO β”‚ synonym_replacement 0.9990 ~
10578
+ 2026-04-05 03:03:56,826 INFO β”‚ t5_paraphrase 1.0000 ~
10579
+ 2026-04-05 03:03:56,826 INFO β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
10580
+ 2026-04-05 03:03:56,826 INFO No improvement (8/25) best=0.9951
10581
+
10582
+ 2026-04-05 03:04:10,451 INFO [Step 196/200] det=skip para=-0.0026 raw_rwd=0.1232Β±0.0036 rwd=0.1232
10583
+ 2026-04-05 03:04:10,451 INFO per-gen rewards: chatgpt=0.126 cohere=0.123 cohere-c=0.125 gpt2=0.124 gpt3=0.117 gpt4=0.122 llama-ch=0.124 mistral=0.125 mistral-=0.124 mpt=0.123 mpt-chat=0.121
10584
+ 2026-04-05 03:04:23,829 INFO [Step 197/200] det=skip para=-0.0000 raw_rwd=0.1228Β±0.0042 rwd=0.1228
10585
+ 2026-04-05 03:04:23,829 INFO per-gen rewards: chatgpt=0.125 cohere=0.122 cohere-c=0.123 gpt2=0.121 gpt3=0.123 gpt4=0.124 llama-ch=0.123 mistral=0.123 mistral-=0.126 mpt=0.121 mpt-chat=0.121
10586
+ 2026-04-05 03:04:37,218 INFO ⚠ AUROC=0.9951 > threshold=0.995 β†’ freezing detector for 3 steps
10587
+ 2026-04-05 03:04:37,219 INFO [Step 198/200] det=skip para=0.0058 raw_rwd=0.1219Β±0.0041 rwd=0.1219
10588
+ 2026-04-05 03:04:37,219 INFO per-gen rewards: chatgpt=0.123 cohere=0.122 cohere-c=0.122 gpt2=0.120 gpt3=0.124 gpt4=0.116 llama-ch=0.123 mistral=0.121 mistral-=0.122 mpt=0.123 mpt-chat=0.123
10589
+ 2026-04-05 03:04:50,589 INFO [Step 199/200] det=skip para=0.0000 raw_rwd=0.1220Β±0.0041 rwd=0.1220
10590
+ 2026-04-05 03:04:50,589 INFO per-gen rewards: chatgpt=0.122 cohere=0.122 cohere-c=0.117 gpt2=0.124 gpt3=0.122 gpt4=0.122 llama-ch=0.124 mistral=0.122 mistral-=0.124 mpt=0.121 mpt-chat=0.120
10591
+ 2026-04-05 03:05:03,937 INFO [Step 200/200] det=skip para=-0.0000 raw_rwd=0.1229Β±0.0037 rwd=0.1229
10592
+ 2026-04-05 03:05:03,937 INFO per-gen rewards: chatgpt=0.124 cohere=0.122 cohere-c=0.122 gpt2=0.121 gpt3=0.123 gpt4=0.121 llama-ch=0.124 mistral=0.123 mistral-=0.125 mpt=0.124 mpt-chat=0.121
10593
+ 2026-04-05 03:58:24,227 INFO β”Œβ”€ Generator AUROC ────────────────────────────┐
10594
+ 2026-04-05 03:58:24,228 INFO β”‚ MACRO_AVG 0.9951 β—„
10595
+ 2026-04-05 03:58:24,228 INFO β”‚ chatgpt 0.9991
10596
+ 2026-04-05 03:58:24,228 INFO β”‚ cohere 0.9852
10597
+ 2026-04-05 03:58:24,228 INFO β”‚ cohere-chat 0.9934
10598
+ 2026-04-05 03:58:24,228 INFO β”‚ gpt2 0.9954
10599
+ 2026-04-05 03:58:24,228 INFO β”‚ gpt3 0.9982
10600
+ 2026-04-05 03:58:24,228 INFO β”‚ gpt4 0.9995
10601
+ 2026-04-05 03:58:24,228 INFO β”‚ llama-chat 0.9994
10602
+ 2026-04-05 03:58:24,228 INFO β”‚ mistral 0.9913
10603
+ 2026-04-05 03:58:24,228 INFO β”‚ mistral-chat 0.9994
10604
+ 2026-04-05 03:58:24,228 INFO β”‚ mpt 0.9865
10605
+ 2026-04-05 03:58:24,228 INFO β”‚ mpt-chat 0.9991
10606
+ 2026-04-05 03:58:24,228 INFO β”‚ (prev best) 0.9951
10607
+ 2026-04-05 03:58:24,228 INFO β”œβ”€ Attack AUROC (robustness) ──────────────────
10608
+ 2026-04-05 03:58:24,228 INFO β”‚ article_deletion 0.9993 ~
10609
+ 2026-04-05 03:58:24,228 INFO β”‚ homoglyphs 0.9994 ~
10610
+ 2026-04-05 03:58:24,228 INFO β”‚ misspelling 0.9996 ~
10611
+ 2026-04-05 03:58:24,228 INFO β”‚ no_attack 0.9994 ~
10612
+ 2026-04-05 03:58:24,228 INFO β”‚ synonym_replacement 0.9990 ~
10613
+ 2026-04-05 03:58:24,228 INFO β”‚ t5_paraphrase 1.0000 ~
10614
+ 2026-04-05 03:58:24,228 INFO β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
10615
+ 2026-04-05 03:58:24,228 INFO No improvement (9/25) best=0.9951
10616
+
10617
+ 2026-04-05 03:58:24,228 INFO ════════════════════════════════════════════════════════════════════════
10618
+ 2026-04-05 03:58:24,228 INFO Done. Best macro AUROC = 0.9951
10619
+ 2026-04-05 03:58:24,228 INFO Detector β†’ ./radar_multievasion/best_detector
10620
+ 2026-04-05 03:58:24,228 INFO Paraphraser β†’ ./radar_multievasion/best_paraphraser
10621
+ 2026-04-05 03:58:24,228 INFO Gen AUROC β†’ ./radar_multievasion/per_generator_auroc.tsv
10622
+ 2026-04-05 03:58:24,228 INFO Atk AUROC β†’ ./radar_multievasion/per_attack_auroc.tsv
10623
+ 2026-04-05 03:58:24,228 INFO ════════════════════════════════════════════════════════════════════════
10624
+ 2026-04-05 03:58:24,228 INFO
10625
+ ── Pushing to HuggingFace Hub (trigger=final) ──
10626
+ 2026-04-05 03:58:24,444 INFO HTTP Request: GET https://huggingface.co/api/whoami-v2 "HTTP/1.1 200 OK"
10627
+ 2026-04-05 03:58:24,572 INFO HTTP Request: GET https://huggingface.co/api/models/Shushant/ADAL_AI_Detector "HTTP/1.1 200 OK"
10628
+ 2026-04-05 03:58:24,573 INFO Repo exists: Shushant/ADAL_AI_Detector
10629
+ 2026-04-05 03:58:24,689 INFO HTTP Request: GET https://huggingface.co/api/models/Shushant/ADAL_Paraphrasher "HTTP/1.1 200 OK"
10630
+ 2026-04-05 03:58:24,690 INFO Repo exists: Shushant/ADAL_Paraphrasher
10631
+ 2026-04-05 03:58:24,690 INFO Uploading detector β†’ Shushant/ADAL_AI_Detector …
10632
+ 2026-04-05 03:58:24,801 INFO HTTP Request: POST https://huggingface.co/api/validate-yaml "HTTP/1.1 200 OK"
10633
+ 2026-04-05 03:58:25,932 INFO HTTP Request: POST https://huggingface.co/api/validate-yaml "HTTP/1.1 200 OK"
10634
+ 2026-04-05 03:58:26,096 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
10635
+ 2026-04-05 03:58:26,213 INFO HTTP Request: POST https://huggingface.co/Shushant/ADAL_AI_Detector.git/info/lfs/objects/batch "HTTP/1.1 200 OK"
10636
+ 2026-04-05 03:58:26,337 INFO HTTP Request: GET https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/xet-write-token/main "HTTP/1.1 200 OK"
10637
+ 2026-04-05 03:58:31,349 WARNING No files have been modified since last commit. Skipping to prevent empty commit.
10638
+ 2026-04-05 03:58:31,519 INFO HTTP Request: GET https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/revision/main "HTTP/1.1 200 OK"
10639
+ 2026-04-05 03:58:31,659 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
10640
+ 2026-04-05 03:58:32,777 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
10641
+ 2026-04-05 03:58:32,958 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
10642
+ 2026-04-05 03:58:33,961 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
10643
+ 2026-04-05 03:58:34,168 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"