Hugging Face's logo

Shushant
/
ADAL_AI_Detector

Text Classification
Safetensors
English
roberta
ai-generated-text-detection
adversarial-training
Model card Files
xet
 Community
Shushant commited on
Commit
42e70c8
·
verified ·
1 Parent(s): 3377148

Training logs (best)

Browse files
Files changed (1) hide show
  1. training_logs/training.log +252 -0
training_logs/training.log CHANGED
@@ -9890,3 +9890,255 @@
9890
  2026-04-04 12:21:51,950 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
9891
  2026-04-04 12:21:52,952 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
9892
  2026-04-04 12:21:53,092 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9890
  2026-04-04 12:21:51,950 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
9891
  2026-04-04 12:21:52,952 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
9892
  2026-04-04 12:21:53,092 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
9893
+ 2026-04-04 12:21:54,965 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
9894
+ 2026-04-04 12:21:54,965 INFO ✓ Detector pushed → https://huggingface.co/Shushant/ADAL_AI_Detector
9895
+ 2026-04-04 12:21:54,965 INFO Uploading paraphraser → Shushant/ADAL_Paraphrasher …
9896
+ 2026-04-04 12:21:55,076 INFO HTTP Request: POST https://huggingface.co/api/validate-yaml "HTTP/1.1 200 OK"
9897
+ 2026-04-04 12:21:55,712 INFO HTTP Request: POST https://huggingface.co/api/validate-yaml "HTTP/1.1 200 OK"
9898
+ 2026-04-04 12:21:55,899 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_Paraphrasher/preupload/main "HTTP/1.1 200 OK"
9899
+ 2026-04-04 12:21:56,033 INFO HTTP Request: POST https://huggingface.co/Shushant/ADAL_Paraphrasher.git/info/lfs/objects/batch "HTTP/1.1 200 OK"
9900
+ 2026-04-04 12:21:56,148 INFO HTTP Request: GET https://huggingface.co/api/models/Shushant/ADAL_Paraphrasher/xet-write-token/main "HTTP/1.1 200 OK"
9901
+ 2026-04-04 12:22:03,055 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_Paraphrasher/commit/main "HTTP/1.1 200 OK"
9902
+ 2026-04-04 12:22:03,056 INFO ✓ Paraphraser pushed → https://huggingface.co/Shushant/ADAL_Paraphrasher
9903
+ 2026-04-04 12:22:03,056 INFO ── Hub push complete ──
9904
+
9905
+ 2026-04-04 12:22:35,997 INFO [Step 116/200] det=3.0626 para=0.0000 raw_rwd=0.1635±0.0105 rwd=0.1635
9906
+ 2026-04-04 12:22:35,998 INFO loss components: L_human=0.433 L_xm_original=0.545 L_t5_para=0.424 L_synonym_replacement=0.430 L_homoglyphs=0.425 L_article_deletion=0.451 L_misspelling=0.493 total=3.203
9907
+ 2026-04-04 12:22:35,998 INFO per-gen rewards: chatgpt=0.162 cohere=0.160 cohere-c=0.166 gpt2=0.161 gpt3=0.161 gpt4=0.162 llama-ch=0.168 mistral=0.167 mistral-=0.159 mpt=0.167 mpt-chat=0.162
9908
+ 2026-04-04 12:22:49,032 INFO [Step 117/200] det=skip para=0.0000 raw_rwd=0.1348±0.0047 rwd=0.1348
9909
+ 2026-04-04 12:22:49,032 INFO per-gen rewards: chatgpt=0.136 cohere=0.137 cohere-c=0.134 gpt2=0.136 gpt3=0.135 gpt4=0.133 llama-ch=0.135 mistral=0.132 mistral-=0.135 mpt=0.135 mpt-chat=0.135
9910
+ 2026-04-04 12:23:22,331 INFO [Step 118/200] det=3.0786 para=-0.0000 raw_rwd=0.1332±0.0046 rwd=0.1332
9911
+ 2026-04-04 12:23:22,331 INFO loss components: L_human=0.431 L_xm_original=0.433 L_t5_para=0.424 L_synonym_replacement=0.425 L_homoglyphs=0.425 L_article_deletion=0.433 L_misspelling=0.426 total=2.996
9912
+ 2026-04-04 12:23:22,331 INFO per-gen rewards: chatgpt=0.131 cohere=0.130 cohere-c=0.131 gpt2=0.134 gpt3=0.131 gpt4=0.137 llama-ch=0.132 mistral=0.136 mistral-=0.135 mpt=0.134 mpt-chat=0.133
9913
+ 2026-04-04 12:23:35,629 INFO [Step 119/200] det=skip para=0.0000 raw_rwd=0.1412±0.0046 rwd=0.1412
9914
+ 2026-04-04 12:23:35,629 INFO per-gen rewards: chatgpt=0.140 cohere=0.138 cohere-c=0.141 gpt2=0.141 gpt3=0.145 gpt4=0.143 llama-ch=0.143 mistral=0.142 mistral-=0.141 mpt=0.141 mpt-chat=0.139
9915
+ 2026-04-04 12:24:08,942 INFO [Step 120/200] det=3.0967 para=-0.0089 raw_rwd=0.1409±0.0049 rwd=0.1409
9916
+ 2026-04-04 12:24:08,942 INFO loss components: L_human=0.430 L_xm_original=0.427 L_t5_para=0.425 L_synonym_replacement=0.427 L_homoglyphs=0.424 L_article_deletion=0.424 L_misspelling=0.429 total=2.986
9917
+ 2026-04-04 12:24:08,942 INFO per-gen rewards: chatgpt=0.140 cohere=0.140 cohere-c=0.142 gpt2=0.142 gpt3=0.142 gpt4=0.143 llama-ch=0.140 mistral=0.139 mistral-=0.143 mpt=0.140 mpt-chat=0.141
9918
+ 2026-04-04 13:17:39,171 INFO ┌─ Generator AUROC ────────────────────────────┐
9919
+ 2026-04-04 13:17:39,172 INFO │ MACRO_AVG 0.9934 ◄
9920
+ 2026-04-04 13:17:39,172 INFO │ chatgpt 0.9987
9921
+ 2026-04-04 13:17:39,172 INFO │ cohere 0.9826
9922
+ 2026-04-04 13:17:39,172 INFO │ cohere-chat 0.9898
9923
+ 2026-04-04 13:17:39,172 INFO │ gpt2 0.9925
9924
+ 2026-04-04 13:17:39,172 INFO │ gpt3 0.9976
9925
+ 2026-04-04 13:17:39,172 INFO │ gpt4 0.9991
9926
+ 2026-04-04 13:17:39,172 INFO │ llama-chat 0.9987
9927
+ 2026-04-04 13:17:39,172 INFO │ mistral 0.9886
9928
+ 2026-04-04 13:17:39,172 INFO │ mistral-chat 0.9988
9929
+ 2026-04-04 13:17:39,172 INFO │ mpt 0.9822
9930
+ 2026-04-04 13:17:39,172 INFO │ mpt-chat 0.9985
9931
+ 2026-04-04 13:17:39,172 INFO │ (prev best) 0.9936
9932
+ 2026-04-04 13:17:39,172 INFO ├─ Attack AUROC (robustness) ─────────────────┤
9933
+ 2026-04-04 13:17:39,172 INFO │ article_deletion 0.9988 ~
9934
+ 2026-04-04 13:17:39,172 INFO │ homoglyphs 0.9991 ~
9935
+ 2026-04-04 13:17:39,172 INFO │ misspelling 0.9982 ~
9936
+ 2026-04-04 13:17:39,172 INFO │ no_attack 0.9986 ~
9937
+ 2026-04-04 13:17:39,172 INFO │ synonym_replacement 0.9987 ~
9938
+ 2026-04-04 13:17:39,172 INFO │ t5_paraphrase 1.0000 ~
9939
+ 2026-04-04 13:17:39,172 INFO └─────────────────────────────────────────────┘
9940
+ 2026-04-04 13:17:39,172 INFO No improvement (1/25) best=0.9936
9941
+
9942
+ 2026-04-04 13:17:52,673 INFO [Step 121/200] det=skip para=0.0000 raw_rwd=0.1491±0.0050 rwd=0.1491
9943
+ 2026-04-04 13:17:52,674 INFO per-gen rewards: chatgpt=0.148 cohere=0.146 cohere-c=0.147 gpt2=0.150 gpt3=0.152 gpt4=0.145 llama-ch=0.150 mistral=0.148 mistral-=0.148 mpt=0.149 mpt-chat=0.153
9944
+ 2026-04-04 13:18:25,430 INFO [Step 122/200] det=3.1111 para=-0.0000 raw_rwd=0.1494±0.0042 rwd=0.1494
9945
+ 2026-04-04 13:18:25,430 INFO loss components: L_human=0.427 L_xm_original=0.430 L_t5_para=0.427 L_synonym_replacement=0.430 L_homoglyphs=0.425 L_article_deletion=0.425 L_misspelling=0.429 total=2.992
9946
+ 2026-04-04 13:18:25,430 INFO per-gen rewards: chatgpt=0.149 cohere=0.149 cohere-c=0.153 gpt2=0.150 gpt3=0.147 gpt4=0.149 llama-ch=0.148 mistral=0.151 mistral-=0.150 mpt=0.147 mpt-chat=0.150
9947
+ 2026-04-04 13:18:38,727 INFO [Step 123/200] det=skip para=0.0060 raw_rwd=0.1429±0.0042 rwd=0.1429
9948
+ 2026-04-04 13:18:38,727 INFO per-gen rewards: chatgpt=0.138 cohere=0.145 cohere-c=0.143 gpt2=0.143 gpt3=0.144 gpt4=0.143 llama-ch=0.143 mistral=0.146 mistral-=0.141 mpt=0.142 mpt-chat=0.142
9949
+ 2026-04-04 13:19:12,863 INFO [Step 124/200] det=3.0339 para=-0.0019 raw_rwd=0.1427±0.0045 rwd=0.1427
9950
+ 2026-04-04 13:19:12,863 INFO loss components: L_human=0.481 L_xm_original=0.465 L_t5_para=0.427 L_synonym_replacement=0.431 L_homoglyphs=0.426 L_article_deletion=0.425 L_misspelling=0.438 total=3.092
9951
+ 2026-04-04 13:19:12,863 INFO per-gen rewards: chatgpt=0.146 cohere=0.144 cohere-c=0.143 gpt2=0.140 gpt3=0.144 gpt4=0.135 llama-ch=0.145 mistral=0.142 mistral-=0.142 mpt=0.143 mpt-chat=0.144
9952
+ 2026-04-04 13:19:26,151 INFO [Step 125/200] det=skip para=-0.0029 raw_rwd=0.1550±0.0049 rwd=0.1550
9953
+ 2026-04-04 13:19:26,151 INFO per-gen rewards: chatgpt=0.154 cohere=0.157 cohere-c=0.153 gpt2=0.156 gpt3=0.154 gpt4=0.153 llama-ch=0.156 mistral=0.155 mistral-=0.155 mpt=0.152 mpt-chat=0.156
9954
+ 2026-04-04 14:12:46,309 INFO ┌─ Generator AUROC ────────────────────────────┐
9955
+ 2026-04-04 14:12:46,310 INFO │ MACRO_AVG 0.9926 ◄
9956
+ 2026-04-04 14:12:46,310 INFO │ chatgpt 0.9974
9957
+ 2026-04-04 14:12:46,310 INFO │ cohere 0.9808
9958
+ 2026-04-04 14:12:46,310 INFO │ cohere-chat 0.9890
9959
+ 2026-04-04 14:12:46,310 INFO │ gpt2 0.9925
9960
+ 2026-04-04 14:12:46,310 INFO │ gpt3 0.9965
9961
+ 2026-04-04 14:12:46,310 INFO │ gpt4 0.9981
9962
+ 2026-04-04 14:12:46,310 INFO │ llama-chat 0.9970
9963
+ 2026-04-04 14:12:46,310 INFO │ mistral 0.9881
9964
+ 2026-04-04 14:12:46,310 INFO │ mistral-chat 0.9976
9965
+ 2026-04-04 14:12:46,310 INFO │ mpt 0.9841
9966
+ 2026-04-04 14:12:46,310 INFO │ mpt-chat 0.9976
9967
+ 2026-04-04 14:12:46,310 INFO │ (prev best) 0.9936
9968
+ 2026-04-04 14:12:46,310 INFO ├─ Attack AUROC (robustness) ─────────────────┤
9969
+ 2026-04-04 14:12:46,310 INFO │ article_deletion 0.9979 ~
9970
+ 2026-04-04 14:12:46,310 INFO │ homoglyphs 0.9983 ~
9971
+ 2026-04-04 14:12:46,310 INFO │ misspelling 0.9978 ~
9972
+ 2026-04-04 14:12:46,310 INFO │ no_attack 0.9969 ~
9973
+ 2026-04-04 14:12:46,310 INFO │ synonym_replacement 0.9976 ~
9974
+ 2026-04-04 14:12:46,310 INFO │ t5_paraphrase 1.0000 ~
9975
+ 2026-04-04 14:12:46,310 INFO └─────────────────────────────────────────────┘
9976
+ 2026-04-04 14:12:46,310 INFO No improvement (2/25) best=0.9936
9977
+
9978
+ 2026-04-04 14:13:20,000 INFO [Step 126/200] det=3.0607 para=0.0000 raw_rwd=0.1547±0.0049 rwd=0.1547
9979
+ 2026-04-04 14:13:20,000 INFO loss components: L_human=0.484 L_xm_original=0.425 L_t5_para=0.423 L_synonym_replacement=0.428 L_homoglyphs=0.427 L_article_deletion=0.426 L_misspelling=0.425 total=3.037
9980
+ 2026-04-04 14:13:20,000 INFO per-gen rewards: chatgpt=0.152 cohere=0.157 cohere-c=0.154 gpt2=0.154 gpt3=0.155 gpt4=0.156 llama-ch=0.157 mistral=0.153 mistral-=0.155 mpt=0.155 mpt-chat=0.154
9981
+ 2026-04-04 14:13:33,261 INFO [Step 127/200] det=skip para=0.0000 raw_rwd=0.1592±0.0047 rwd=0.1592
9982
+ 2026-04-04 14:13:33,261 INFO per-gen rewards: chatgpt=0.159 cohere=0.156 cohere-c=0.157 gpt2=0.161 gpt3=0.159 gpt4=0.158 llama-ch=0.161 mistral=0.158 mistral-=0.159 mpt=0.158 mpt-chat=0.162
9983
+ 2026-04-04 14:14:06,627 INFO [Step 128/200] det=3.0722 para=-0.0025 raw_rwd=0.1598±0.0049 rwd=0.1598
9984
+ 2026-04-04 14:14:06,627 INFO loss components: L_human=0.425 L_xm_original=0.434 L_t5_para=0.426 L_synonym_replacement=0.425 L_homoglyphs=0.427 L_article_deletion=0.433 L_misspelling=0.427 total=2.997
9985
+ 2026-04-04 14:14:06,627 INFO per-gen rewards: chatgpt=0.160 cohere=0.154 cohere-c=0.160 gpt2=0.159 gpt3=0.160 gpt4=0.163 llama-ch=0.163 mistral=0.160 mistral-=0.158 mpt=0.159 mpt-chat=0.160
9986
+ 2026-04-04 14:14:19,952 INFO [Step 129/200] det=skip para=-0.0000 raw_rwd=0.1485±0.0059 rwd=0.1485
9987
+ 2026-04-04 14:14:19,952 INFO per-gen rewards: chatgpt=0.151 cohere=0.145 cohere-c=0.149 gpt2=0.149 gpt3=0.150 gpt4=0.147 llama-ch=0.149 mistral=0.150 mistral-=0.150 mpt=0.146 mpt-chat=0.148
9988
+ 2026-04-04 14:14:52,995 INFO [Step 130/200] det=3.0254 para=-0.0000 raw_rwd=0.1479±0.0049 rwd=0.1479
9989
+ 2026-04-04 14:14:52,995 INFO loss components: L_human=0.427 L_xm_original=0.426 L_t5_para=0.424 L_synonym_replacement=0.426 L_homoglyphs=0.427 L_article_deletion=0.447 L_misspelling=0.442 total=3.018
9990
+ 2026-04-04 14:14:52,995 INFO per-gen rewards: chatgpt=0.149 cohere=0.145 cohere-c=0.144 gpt2=0.151 gpt3=0.148 gpt4=0.148 llama-ch=0.145 mistral=0.147 mistral-=0.150 mpt=0.148 mpt-chat=0.150
9991
+ 2026-04-04 15:08:17,469 INFO ┌─ Generator AUROC ────────────────────────────┐
9992
+ 2026-04-04 15:08:17,470 INFO │ MACRO_AVG 0.9894 ◄
9993
+ 2026-04-04 15:08:17,470 INFO │ chatgpt 0.9931
9994
+ 2026-04-04 15:08:17,470 INFO │ cohere 0.9824
9995
+ 2026-04-04 15:08:17,470 INFO │ cohere-chat 0.9892
9996
+ 2026-04-04 15:08:17,470 INFO │ gpt2 0.9886
9997
+ 2026-04-04 15:08:17,470 INFO │ gpt3 0.9932
9998
+ 2026-04-04 15:08:17,470 INFO │ gpt4 0.9944
9999
+ 2026-04-04 15:08:17,470 INFO │ llama-chat 0.9924
10000
+ 2026-04-04 15:08:17,470 INFO │ mistral 0.9852
10001
+ 2026-04-04 15:08:17,470 INFO │ mistral-chat 0.9935
10002
+ 2026-04-04 15:08:17,470 INFO │ mpt 0.9768
10003
+ 2026-04-04 15:08:17,470 INFO │ mpt-chat 0.9944
10004
+ 2026-04-04 15:08:17,470 INFO │ (prev best) 0.9936
10005
+ 2026-04-04 15:08:17,470 INFO ├─ Attack AUROC (robustness) ─────────────────┤
10006
+ 2026-04-04 15:08:17,470 INFO │ article_deletion 0.9954 ~
10007
+ 2026-04-04 15:08:17,470 INFO │ homoglyphs 0.9988 ~
10008
+ 2026-04-04 15:08:17,470 INFO │ misspelling 0.9965 ~
10009
+ 2026-04-04 15:08:17,470 INFO │ no_attack 0.9923 ~
10010
+ 2026-04-04 15:08:17,470 INFO │ synonym_replacement 0.9946 ~
10011
+ 2026-04-04 15:08:17,470 INFO │ t5_paraphrase 1.0000 ~
10012
+ 2026-04-04 15:08:17,470 INFO └─────────────────────────────────────────────┘
10013
+ 2026-04-04 15:08:17,470 INFO No improvement (3/25) best=0.9936
10014
+
10015
+ 2026-04-04 15:08:31,016 INFO [Step 131/200] det=skip para=0.0000 raw_rwd=0.1436±0.0049 rwd=0.1436
10016
+ 2026-04-04 15:08:31,016 INFO per-gen rewards: chatgpt=0.148 cohere=0.147 cohere-c=0.143 gpt2=0.142 gpt3=0.141 gpt4=0.141 llama-ch=0.143 mistral=0.143 mistral-=0.146 mpt=0.140 mpt-chat=0.145
10017
+ 2026-04-04 15:09:04,695 INFO [Step 132/200] det=3.0473 para=0.0000 raw_rwd=0.1431±0.0053 rwd=0.1431
10018
+ 2026-04-04 15:09:04,695 INFO loss components: L_human=0.425 L_xm_original=0.471 L_t5_para=0.427 L_synonym_replacement=0.438 L_homoglyphs=0.426 L_article_deletion=0.430 L_misspelling=0.427 total=3.044
10019
+ 2026-04-04 15:09:04,695 INFO per-gen rewards: chatgpt=0.148 cohere=0.145 cohere-c=0.141 gpt2=0.141 gpt3=0.144 gpt4=0.148 llama-ch=0.140 mistral=0.144 mistral-=0.144 mpt=0.141 mpt-chat=0.143
10020
+ 2026-04-04 15:09:17,963 INFO [Step 133/200] det=skip para=0.0301 raw_rwd=0.1400±0.0058 rwd=0.1400
10021
+ 2026-04-04 15:09:17,963 INFO per-gen rewards: chatgpt=0.138 cohere=0.134 cohere-c=0.141 gpt2=0.142 gpt3=0.141 gpt4=0.141 llama-ch=0.138 mistral=0.144 mistral-=0.138 mpt=0.142 mpt-chat=0.139
10022
+ 2026-04-04 15:09:51,222 INFO [Step 134/200] det=3.0896 para=-0.0000 raw_rwd=0.1399±0.0041 rwd=0.1399
10023
+ 2026-04-04 15:09:51,222 INFO loss components: L_human=0.469 L_xm_original=0.430 L_t5_para=0.427 L_synonym_replacement=0.434 L_homoglyphs=0.428 L_article_deletion=0.437 L_misspelling=0.433 total=3.058
10024
+ 2026-04-04 15:09:51,222 INFO per-gen rewards: chatgpt=0.142 cohere=0.137 cohere-c=0.138 gpt2=0.140 gpt3=0.143 gpt4=0.142 llama-ch=0.140 mistral=0.140 mistral-=0.141 mpt=0.141 mpt-chat=0.137
10025
+ 2026-04-04 15:10:04,502 INFO [Step 135/200] det=skip para=0.0000 raw_rwd=0.1389±0.0034 rwd=0.1389
10026
+ 2026-04-04 15:10:04,503 INFO per-gen rewards: chatgpt=0.141 cohere=0.142 cohere-c=0.138 gpt2=0.138 gpt3=0.140 gpt4=0.139 llama-ch=0.139 mistral=0.139 mistral-=0.138 mpt=0.138 mpt-chat=0.138
10027
+ 2026-04-04 16:03:23,680 INFO ┌─ Generator AUROC ────────────────────────────┐
10028
+ 2026-04-04 16:03:23,680 INFO │ MACRO_AVG 0.9906 ◄
10029
+ 2026-04-04 16:03:23,680 INFO │ chatgpt 0.9945
10030
+ 2026-04-04 16:03:23,680 INFO │ cohere 0.9849
10031
+ 2026-04-04 16:03:23,680 INFO │ cohere-chat 0.9899
10032
+ 2026-04-04 16:03:23,680 INFO │ gpt2 0.9896
10033
+ 2026-04-04 16:03:23,680 INFO │ gpt3 0.9941
10034
+ 2026-04-04 16:03:23,680 INFO │ gpt4 0.9955
10035
+ 2026-04-04 16:03:23,681 INFO │ llama-chat 0.9937
10036
+ 2026-04-04 16:03:23,681 INFO │ mistral 0.9865
10037
+ 2026-04-04 16:03:23,681 INFO │ mistral-chat 0.9946
10038
+ 2026-04-04 16:03:23,681 INFO │ mpt 0.9776
10039
+ 2026-04-04 16:03:23,681 INFO │ mpt-chat 0.9955
10040
+ 2026-04-04 16:03:23,681 INFO │ (prev best) 0.9936
10041
+ 2026-04-04 16:03:23,681 INFO ├─ Attack AUROC (robustness) ─────────────────┤
10042
+ 2026-04-04 16:03:23,681 INFO │ article_deletion 0.9964 ~
10043
+ 2026-04-04 16:03:23,681 INFO │ homoglyphs 0.9995 ~
10044
+ 2026-04-04 16:03:23,681 INFO │ misspelling 0.9963 ~
10045
+ 2026-04-04 16:03:23,681 INFO │ no_attack 0.9936 ~
10046
+ 2026-04-04 16:03:23,681 INFO │ synonym_replacement 0.9952 ~
10047
+ 2026-04-04 16:03:23,681 INFO │ t5_paraphrase 1.0000 ~
10048
+ 2026-04-04 16:03:23,681 INFO └─────────────────────────────────────────────┘
10049
+ 2026-04-04 16:03:23,681 INFO No improvement (4/25) best=0.9936
10050
+
10051
+ 2026-04-04 16:03:57,654 INFO [Step 136/200] det=3.0342 para=-0.0000 raw_rwd=0.1392±0.0047 rwd=0.1392
10052
+ 2026-04-04 16:03:57,654 INFO loss components: L_human=0.428 L_xm_original=0.426 L_t5_para=0.426 L_synonym_replacement=0.436 L_homoglyphs=0.424 L_article_deletion=0.426 L_misspelling=0.425 total=2.991
10053
+ 2026-04-04 16:03:57,654 INFO per-gen rewards: chatgpt=0.139 cohere=0.138 cohere-c=0.135 gpt2=0.139 gpt3=0.142 gpt4=0.139 llama-ch=0.136 mistral=0.139 mistral-=0.141 mpt=0.145 mpt-chat=0.139
10054
+ 2026-04-04 16:04:10,951 INFO [Step 137/200] det=skip para=-0.0000 raw_rwd=0.1479±0.0060 rwd=0.1479
10055
+ 2026-04-04 16:04:10,951 INFO per-gen rewards: chatgpt=0.151 cohere=0.142 cohere-c=0.148 gpt2=0.153 gpt3=0.147 gpt4=0.144 llama-ch=0.150 mistral=0.148 mistral-=0.148 mpt=0.144 mpt-chat=0.149
10056
+ 2026-04-04 16:04:43,767 INFO [Step 138/200] det=3.0527 para=-0.0000 raw_rwd=0.1470±0.0056 rwd=0.1470
10057
+ 2026-04-04 16:04:43,767 INFO loss components: L_human=0.435 L_xm_original=0.431 L_t5_para=0.428 L_synonym_replacement=0.426 L_homoglyphs=0.425 L_article_deletion=0.443 L_misspelling=0.427 total=3.015
10058
+ 2026-04-04 16:04:43,767 INFO per-gen rewards: chatgpt=0.144 cohere=0.156 cohere-c=0.145 gpt2=0.146 gpt3=0.149 gpt4=0.149 llama-ch=0.148 mistral=0.147 mistral-=0.144 mpt=0.145 mpt-chat=0.147
10059
+ 2026-04-04 16:04:56,983 INFO [Step 139/200] det=skip para=0.0000 raw_rwd=0.1493±0.0060 rwd=0.1493
10060
+ 2026-04-04 16:04:56,983 INFO per-gen rewards: chatgpt=0.147 cohere=0.147 cohere-c=0.146 gpt2=0.150 gpt3=0.145 gpt4=0.151 llama-ch=0.151 mistral=0.148 mistral-=0.149 mpt=0.153 mpt-chat=0.150
10061
+ 2026-04-04 16:05:30,530 INFO [Step 140/200] det=3.1151 para=-0.0000 raw_rwd=0.1509±0.0063 rwd=0.1509
10062
+ 2026-04-04 16:05:30,530 INFO loss components: L_human=0.523 L_xm_original=0.533 L_t5_para=0.425 L_synonym_replacement=0.425 L_homoglyphs=0.427 L_article_deletion=0.434 L_misspelling=0.458 total=3.225
10063
+ 2026-04-04 16:05:30,530 INFO per-gen rewards: chatgpt=0.154 cohere=0.154 cohere-c=0.156 gpt2=0.148 gpt3=0.144 gpt4=0.152 llama-ch=0.152 mistral=0.151 mistral-=0.151 mpt=0.148 mpt-chat=0.150
10064
+ 2026-04-04 16:58:53,565 INFO ┌─ Generator AUROC ────────────────────────────┐
10065
+ 2026-04-04 16:58:53,566 INFO │ MACRO_AVG 0.9922 ◄
10066
+ 2026-04-04 16:58:53,566 INFO │ chatgpt 0.9957
10067
+ 2026-04-04 16:58:53,566 INFO │ cohere 0.9824
10068
+ 2026-04-04 16:58:53,566 INFO │ cohere-chat 0.9901
10069
+ 2026-04-04 16:58:53,566 INFO │ gpt2 0.9932
10070
+ 2026-04-04 16:58:53,566 INFO │ gpt3 0.9954
10071
+ 2026-04-04 16:58:53,566 INFO │ gpt4 0.9959
10072
+ 2026-04-04 16:58:53,566 INFO │ llama-chat 0.9954
10073
+ 2026-04-04 16:58:53,567 INFO │ mistral 0.9889
10074
+ 2026-04-04 16:58:53,567 INFO │ mistral-chat 0.9962
10075
+ 2026-04-04 16:58:53,567 INFO │ mpt 0.9845
10076
+ 2026-04-04 16:58:53,567 INFO │ mpt-chat 0.9970
10077
+ 2026-04-04 16:58:53,567 INFO │ (prev best) 0.9936
10078
+ 2026-04-04 16:58:53,567 INFO ├─ Attack AUROC (robustness) ─────────────────┤
10079
+ 2026-04-04 16:58:53,567 INFO │ article_deletion 0.9989 ~
10080
+ 2026-04-04 16:58:53,567 INFO │ homoglyphs 0.9992 ~
10081
+ 2026-04-04 16:58:53,567 INFO │ misspelling 0.9994 ~
10082
+ 2026-04-04 16:58:53,567 INFO │ no_attack 0.9952 ~
10083
+ 2026-04-04 16:58:53,567 INFO │ synonym_replacement 0.9989 ~
10084
+ 2026-04-04 16:58:53,567 INFO │ t5_paraphrase 1.0000 ~
10085
+ 2026-04-04 16:58:53,567 INFO └─────────────────────────────────────────────┘
10086
+ 2026-04-04 16:58:53,567 INFO No improvement (5/25) best=0.9936
10087
+
10088
+ 2026-04-04 16:59:07,110 INFO [Step 141/200] det=skip para=0.0632 raw_rwd=0.1474±0.0117 rwd=0.1474
10089
+ 2026-04-04 16:59:07,111 INFO per-gen rewards: chatgpt=0.146 cohere=0.146 cohere-c=0.147 gpt2=0.147 gpt3=0.143 gpt4=0.149 llama-ch=0.143 mistral=0.157 mistral-=0.146 mpt=0.148 mpt-chat=0.147
10090
+ 2026-04-04 16:59:40,049 INFO [Step 142/200] det=3.0274 para=-0.0000 raw_rwd=0.1465±0.0056 rwd=0.1465
10091
+ 2026-04-04 16:59:40,050 INFO loss components: L_human=0.429 L_xm_original=0.439 L_t5_para=0.426 L_synonym_replacement=0.427 L_homoglyphs=0.425 L_article_deletion=0.433 L_misspelling=0.426 total=3.005
10092
+ 2026-04-04 16:59:40,050 INFO per-gen rewards: chatgpt=0.142 cohere=0.144 cohere-c=0.142 gpt2=0.149 gpt3=0.148 gpt4=0.145 llama-ch=0.147 mistral=0.148 mistral-=0.147 mpt=0.148 mpt-chat=0.145
10093
+ 2026-04-04 16:59:53,249 INFO [Step 143/200] det=skip para=0.0000 raw_rwd=0.1426±0.0048 rwd=0.1426
10094
+ 2026-04-04 16:59:53,250 INFO per-gen rewards: chatgpt=0.142 cohere=0.142 cohere-c=0.144 gpt2=0.144 gpt3=0.141 gpt4=0.143 llama-ch=0.143 mistral=0.145 mistral-=0.140 mpt=0.144 mpt-chat=0.142
10095
+ 2026-04-04 17:00:26,679 INFO [Step 144/200] det=3.0791 para=-0.0041 raw_rwd=0.1431±0.0046 rwd=0.1431
10096
+ 2026-04-04 17:00:26,679 INFO loss components: L_human=0.436 L_xm_original=0.744 L_t5_para=0.425 L_synonym_replacement=0.426 L_homoglyphs=0.429 L_article_deletion=0.439 L_misspelling=0.427 total=3.325
10097
+ 2026-04-04 17:00:26,679 INFO per-gen rewards: chatgpt=0.140 cohere=0.143 cohere-c=0.139 gpt2=0.145 gpt3=0.140 gpt4=0.145 llama-ch=0.146 mistral=0.143 mistral-=0.142 mpt=0.144 mpt-chat=0.143
10098
+ 2026-04-04 17:00:40,047 INFO [Step 145/200] det=skip para=-0.0000 raw_rwd=0.1596±0.0047 rwd=0.1596
10099
+ 2026-04-04 17:00:40,047 INFO per-gen rewards: chatgpt=0.162 cohere=0.159 cohere-c=0.162 gpt2=0.158 gpt3=0.158 gpt4=0.157 llama-ch=0.158 mistral=0.158 mistral-=0.160 mpt=0.161 mpt-chat=0.162
10100
+ 2026-04-04 17:54:01,116 INFO ┌─ Generator AUROC ────────────────────────────┐
10101
+ 2026-04-04 17:54:01,116 INFO │ MACRO_AVG 0.9941 ◄
10102
+ 2026-04-04 17:54:01,116 INFO │ chatgpt 0.9989
10103
+ 2026-04-04 17:54:01,116 INFO │ cohere 0.9798
10104
+ 2026-04-04 17:54:01,117 INFO │ cohere-chat 0.9906
10105
+ 2026-04-04 17:54:01,117 INFO │ gpt2 0.9947
10106
+ 2026-04-04 17:54:01,117 INFO │ gpt3 0.9976
10107
+ 2026-04-04 17:54:01,117 INFO │ gpt4 0.9993
10108
+ 2026-04-04 17:54:01,117 INFO │ llama-chat 0.9994
10109
+ 2026-04-04 17:54:01,117 INFO │ mistral 0.9905
10110
+ 2026-04-04 17:54:01,117 INFO │ mistral-chat 0.9994
10111
+ 2026-04-04 17:54:01,117 INFO │ mpt 0.9860
10112
+ 2026-04-04 17:54:01,117 INFO │ mpt-chat 0.9992
10113
+ 2026-04-04 17:54:01,117 INFO │ (prev best) 0.9936
10114
+ 2026-04-04 17:54:01,117 INFO ├─ Attack AUROC (robustness) ─────────────────┤
10115
+ 2026-04-04 17:54:01,117 INFO │ article_deletion 0.9998 ~
10116
+ 2026-04-04 17:54:01,117 INFO │ homoglyphs 1.0000 ~
10117
+ 2026-04-04 17:54:01,117 INFO │ misspelling 0.9999 ~
10118
+ 2026-04-04 17:54:01,117 INFO │ no_attack 0.9994 ~
10119
+ 2026-04-04 17:54:01,117 INFO │ synonym_replacement 0.9999 ~
10120
+ 2026-04-04 17:54:01,117 INFO │ t5_paraphrase 1.0000 ~
10121
+ 2026-04-04 17:54:01,117 INFO └─────────────────────────────────────────────┘
10122
+ 2026-04-04 17:54:05,902 INFO Detector saved → ./radar_multievasion/best_detector
10123
+ 2026-04-04 17:54:08,046 INFO Paraphraser saved → ./radar_multievasion/best_paraphraser
10124
+ 2026-04-04 17:54:08,046 INFO ✓ New best macro AUROC = 0.9941 — saved.
10125
+
10126
+ 2026-04-04 17:54:08,046 INFO
10127
+ ── Pushing to HuggingFace Hub (trigger=best) ──
10128
+ 2026-04-04 17:54:08,265 INFO HTTP Request: GET https://huggingface.co/api/whoami-v2 "HTTP/1.1 200 OK"
10129
+ 2026-04-04 17:54:08,384 INFO HTTP Request: GET https://huggingface.co/api/models/Shushant/ADAL_AI_Detector "HTTP/1.1 200 OK"
10130
+ 2026-04-04 17:54:08,386 INFO Repo exists: Shushant/ADAL_AI_Detector
10131
+ 2026-04-04 17:54:08,503 INFO HTTP Request: GET https://huggingface.co/api/models/Shushant/ADAL_Paraphrasher "HTTP/1.1 200 OK"
10132
+ 2026-04-04 17:54:08,504 INFO Repo exists: Shushant/ADAL_Paraphrasher
10133
+ 2026-04-04 17:54:08,505 INFO Uploading detector → Shushant/ADAL_AI_Detector …
10134
+ 2026-04-04 17:54:08,614 INFO HTTP Request: POST https://huggingface.co/api/validate-yaml "HTTP/1.1 200 OK"
10135
+ 2026-04-04 17:54:09,744 INFO HTTP Request: POST https://huggingface.co/api/validate-yaml "HTTP/1.1 200 OK"
10136
+ 2026-04-04 17:54:09,926 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
10137
+ 2026-04-04 17:54:10,062 INFO HTTP Request: POST https://huggingface.co/Shushant/ADAL_AI_Detector.git/info/lfs/objects/batch "HTTP/1.1 200 OK"
10138
+ 2026-04-04 17:54:10,180 INFO HTTP Request: GET https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/xet-write-token/main "HTTP/1.1 200 OK"
10139
+ 2026-04-04 17:54:57,255 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
10140
+ 2026-04-04 17:54:57,397 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
10141
+ 2026-04-04 17:54:58,324 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
10142
+ 2026-04-04 17:54:58,467 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"
10143
+ 2026-04-04 17:54:59,320 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/commit/main "HTTP/1.1 200 OK"
10144
+ 2026-04-04 17:54:59,460 INFO HTTP Request: POST https://huggingface.co/api/models/Shushant/ADAL_AI_Detector/preupload/main "HTTP/1.1 200 OK"