accepted_for_training bool 1
class | ai_reviewed bool 1
class | category stringclasses 11
values | family_id stringlengths 15 27 | human_reviewed bool 1
class | id stringlengths 13 25 | input dict | label_reason stringclasses 48
values | label_source stringclasses 2
values | mode stringclasses 2
values | pilot_label stringclasses 2
values | pilot_label_reason stringclasses 25
values | pilot_label_source stringclasses 2
values | proposed_route stringclasses 2
values | proposed_split stringclasses 1
value | provenance unknown | recipe_group stringclasses 17
values | review_decision stringclasses 2
values | seed_curation dict | source_input_sha256 stringlengths 64 64 ⌀ | source_proposed_route stringclasses 2
values | split_reservation stringclasses 1
value | status stringclasses 1
value | template_group_id stringlengths 20 31 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
false | true | ordinary_image | fam2_train_p001 | false | fam2_train_p001_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p001",
"generation_request_sha256": "c9b398340de4af6086bac18902f66bdfdbaec76b7d409582e3df3e0f77a54416",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 520fa0e9dcdea72383d6614ed841366b18b217cd7ffc70ba24507e274d79458f | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_d139e80e3ad37e68 |
false | true | new_image_or_mixed | fam2_train_p001 | false | fam2_train_p001_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p001",
"generation_request_sha256": "c9b398340de4af6086bac18902f66bdfdbaec76b7d409582e3df3e0f77a54416",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 7528c2c83504482c3865f3b114b0a75f2734edcdc94eeff620cccb4b9484a2c4 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_d139e80e3ad37e68 |
false | true | ordinary_image | fam2_train_p002 | false | fam2_train_p002_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Read the text in that image for me again.",
"role": "user"
},
{
"content": "I read the tex... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p002",
"generation_request_sha256": "c9b398340de4af6086bac18902f66bdfdbaec76b7d409582e3df3e0f77a54416",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | c06851f96ff8b543820b4a808e4760f1d0a9ff982316f64c4fa8978d8584bac6 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_45e7e81e5f8ef6e9 |
false | true | explicit_reanalysis | fam2_train_p002 | false | fam2_train_p002_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Check the authenticity of this image again, please.",
"role": "user"
},
{
"content": "I ra... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p002",
"generation_request_sha256": "c9b398340de4af6086bac18902f66bdfdbaec76b7d409582e3df3e0f77a54416",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 17b371b4c8647dcc1a5bcc108c70ea8fc0a41121ea16b2ca69073dd9ce152455 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_45e7e81e5f8ef6e9 |
false | true | explain_existing | fam2_train_p003 | false | fam2_train_p003_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Why did the system say this is real?",
"role": "user"
},
{
"content": "I can explain the s... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p003",
"generation_request_sha256": "07d214e1a80140273d7686bc61b501fb0d5b9dd955a2f606eea11f3c6dc976ea",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | b2f6b5f855cd590e663070e3a319256dd199da2c340f01a0a3222ca3489b4967 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_85de71b0b93433fd |
false | true | explicit_reanalysis | fam2_train_p003 | false | fam2_train_p003_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Check it again with a deeper analysis.",
"role": "user"
},
{
"content": "I can run a fresh... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p003",
"generation_request_sha256": "07d214e1a80140273d7686bc61b501fb0d5b9dd955a2f606eea11f3c6dc976ea",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 4c1cf2a63f034caf395c5abfd4b2c42c1e801dbc63d114a74c9bbedc0c78140c | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_85de71b0b93433fd |
false | true | ordinary_image | fam2_train_p004 | false | fam2_train_p004_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p004",
"generation_request_sha256": "07d214e1a80140273d7686bc61b501fb0d5b9dd955a2f606eea11f3c6dc976ea",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 6e35c52a3d4035c73615d157073bf2a0e2834d5402d3ab425562139a62c51e45 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_4edce2670e5d08b1 |
false | true | new_image_or_mixed | fam2_train_p004 | false | fam2_train_p004_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p004",
"generation_request_sha256": "07d214e1a80140273d7686bc61b501fb0d5b9dd955a2f606eea11f3c6dc976ea",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | bf68b1ff561bfad01c950b0d096620056c53d8dc98ee478977ccf7ecf0a1d917 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_4edce2670e5d08b1 |
false | true | ordinary_image | fam2_train_p005 | false | fam2_train_p005_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you read what’s written there?",
"role": "user"
},
{
"content": "I read the text in th... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p005",
"generation_request_sha256": "07d214e1a80140273d7686bc61b501fb0d5b9dd955a2f606eea11f3c6dc976ea",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 40699ff546501afb4a9ad83ce0bcd4b3a654cd592db30f6212db9a8b030ef0fa | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_6f85eb92ce972562 |
false | true | explicit_reanalysis | fam2_train_p005 | false | fam2_train_p005_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Run the check again, just to confirm.",
"role": "user"
},
{
"content": "I ran the authenti... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p005",
"generation_request_sha256": "07d214e1a80140273d7686bc61b501fb0d5b9dd955a2f606eea11f3c6dc976ea",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 594eaa20292d7648e06a2a6eda3c42675a4c23fb88678f54e1001804bdc025be | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_6f85eb92ce972562 |
false | true | explain_existing | fam2_train_p006 | false | fam2_train_p006_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Explain why it was marked as fake.",
"role": "user"
},
{
"content": "Would you like me to ... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p006",
"generation_request_sha256": "07d214e1a80140273d7686bc61b501fb0d5b9dd955a2f606eea11f3c6dc976ea",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 3282f9b8a55b5242393ee911a1d4035fabb9eefacc16c2c852e40aa49a8e5f52 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_8fe0958abdcd07c5 |
false | true | explicit_reanalysis | fam2_train_p006 | false | fam2_train_p006_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Try a new check, I’d like to be sure.",
"role": "user"
},
{
"content": "Would you like me ... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p006",
"generation_request_sha256": "07d214e1a80140273d7686bc61b501fb0d5b9dd955a2f606eea11f3c6dc976ea",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 469d24f05f5cc574d0f61a14002b48431661f32a112435b4372186cb7fe3fc9b | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_8fe0958abdcd07c5 |
false | true | ordinary_image | fam2_train_p007 | false | fam2_train_p007_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p007",
"generation_request_sha256": "b70861bde60284186db0137bc87d3c8e89536082a6fe11c85edd7fea60423ec4",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | c7f652de0548a8840d38c309182d24f153c6bc5bfab2f61ff8d25b7c3dc30edf | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_69c2fe233ca345ff |
false | true | new_image_or_mixed | fam2_train_p007 | false | fam2_train_p007_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p007",
"generation_request_sha256": "b70861bde60284186db0137bc87d3c8e89536082a6fe11c85edd7fea60423ec4",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | b149450f1b9729a5718960de6082ec8c2e2b5865d40d20b27ef849f7d1e183f8 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_69c2fe233ca345ff |
false | true | ordinary_image | fam2_train_p008 | false | fam2_train_p008_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "What does it say in that old photo?",
"role": "user"
},
{
"content": "I read the text in t... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p008",
"generation_request_sha256": "b70861bde60284186db0137bc87d3c8e89536082a6fe11c85edd7fea60423ec4",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 957b2b62a4f4c48824aca4c706cb4654d3cfad6cf0980ecefc4aa214e8791115 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_b2c65f1d35315d96 |
false | true | explicit_reanalysis | fam2_train_p008 | false | fam2_train_p008_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you verify the image again?",
"role": "user"
},
{
"content": "I ran the authenticity c... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p008",
"generation_request_sha256": "b70861bde60284186db0137bc87d3c8e89536082a6fe11c85edd7fea60423ec4",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 5497eff15abdc6e0655a9b3c22b4da4f295e451b87bd266c6f9f9c183130c858 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_b2c65f1d35315d96 |
false | true | explain_existing | fam2_train_p009 | false | fam2_train_p009_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Tell me how you reached that conclusion.",
"role": "user"
},
{
"content": "I can explain t... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p009",
"generation_request_sha256": "b70861bde60284186db0137bc87d3c8e89536082a6fe11c85edd7fea60423ec4",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 6aef77e91de936161419450882ab1b24c8339209e1568f6eb6034ffc3575ea2f | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_05d9bf6cbd595635 |
false | true | explicit_reanalysis | fam2_train_p009 | false | fam2_train_p009_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Do a new analysis of the image.",
"role": "user"
},
{
"content": "I can run a fresh authen... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p009",
"generation_request_sha256": "b70861bde60284186db0137bc87d3c8e89536082a6fe11c85edd7fea60423ec4",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 4bc4a8fd8492fc4745ecbd0c58b7b2c8fa03ddfd693cb782ba2c2552a54f0ba0 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_05d9bf6cbd595635 |
false | true | ordinary_image | fam2_train_p010 | false | fam2_train_p010_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p010",
"generation_request_sha256": "b70861bde60284186db0137bc87d3c8e89536082a6fe11c85edd7fea60423ec4",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 2ef0916d7d3e495692e8c42f6857deedb28476144f160e1a4049ac3fa5108dac | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_9d69be7caddaf19b |
false | true | new_image_or_mixed | fam2_train_p010 | false | fam2_train_p010_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p010",
"generation_request_sha256": "b70861bde60284186db0137bc87d3c8e89536082a6fe11c85edd7fea60423ec4",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 564cdc84a3a93f525bb03aee90dc3b0a2c68ae722e3056e2a09925f2b5b9851c | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_9d69be7caddaf19b |
false | true | ordinary_image | fam2_train_p011 | false | fam2_train_p011_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you read the text in this photo again?",
"role": "user"
},
{
"content": "I read the te... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p011",
"generation_request_sha256": "1f39e59a6b2f0019ba455d1c6e24bd0802f642f6bae4792ac6cc1a1c3956df4a",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 5f3033d679ba0e8f5252f466534882f314876a5b145c078f71334039a0c9aa17 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_569b4aaf743188ec |
false | true | explicit_reanalysis | fam2_train_p011 | false | fam2_train_p011_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Could you check if it was altered?",
"role": "user"
},
{
"content": "I ran the authenticit... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p011",
"generation_request_sha256": "1f39e59a6b2f0019ba455d1c6e24bd0802f642f6bae4792ac6cc1a1c3956df4a",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 5fe205f00f2fcebe6331f0772360a0fd3d06fb3ffdad27618cca4dd399b460e4 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_569b4aaf743188ec |
false | true | explain_existing | fam2_train_p012 | false | fam2_train_p012_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you tell me more about the fake result?",
"role": "user"
},
{
"content": "Would you li... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p012",
"generation_request_sha256": "1f39e59a6b2f0019ba455d1c6e24bd0802f642f6bae4792ac6cc1a1c3956df4a",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | b674f54eb7e1b2a3cdadb08b558a2df0f4f750453cd356435210577681b9ec6e | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_6975080df89ab48e |
false | true | explicit_reanalysis | fam2_train_p012 | false | fam2_train_p012_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I want a new check for authenticity, please.",
"role": "user"
},
{
"content": "Would you l... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p012",
"generation_request_sha256": "1f39e59a6b2f0019ba455d1c6e24bd0802f642f6bae4792ac6cc1a1c3956df4a",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | f7a4df01b48591fc1bed370a85dc3d7f470b3e9a7cda8e0b8f00b01b1c1dceb2 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_6975080df89ab48e |
false | true | ordinary_image | fam2_train_p013 | false | fam2_train_p013_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p013",
"generation_request_sha256": "1f39e59a6b2f0019ba455d1c6e24bd0802f642f6bae4792ac6cc1a1c3956df4a",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | c012cf783d4bb7248e6805d9289fa0f7e601675113e00382ca0c014824d1ee5b | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_267d6ddf44c2a8dd |
false | true | new_image_or_mixed | fam2_train_p013 | false | fam2_train_p013_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p013",
"generation_request_sha256": "1f39e59a6b2f0019ba455d1c6e24bd0802f642f6bae4792ac6cc1a1c3956df4a",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 0008abae1bff40b122435cddabc762211aa33750a5a226e5909736012e783a6f | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_267d6ddf44c2a8dd |
false | true | ordinary_image | fam2_train_p014 | false | fam2_train_p014_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Read the text again from this picture.",
"role": "user"
},
{
"content": "I read the text i... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p014",
"generation_request_sha256": "1f39e59a6b2f0019ba455d1c6e24bd0802f642f6bae4792ac6cc1a1c3956df4a",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 2f411bf2e47a0d73b20f95f40b4b036ecc03d4f3f7cc223d4ba49230d5b94d9c | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_b9dffabf6629e19b |
false | true | explicit_reanalysis | fam2_train_p014 | false | fam2_train_p014_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Check again to see if it's real or fake.",
"role": "user"
},
{
"content": "I ran the authe... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p014",
"generation_request_sha256": "1f39e59a6b2f0019ba455d1c6e24bd0802f642f6bae4792ac6cc1a1c3956df4a",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | c7a3782a2f9d58be9b9d2822e5fdfa611e9163c2e2cdc91a01e2bc6fd6f481bd | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_b9dffabf6629e19b |
false | true | explain_existing | fam2_train_p015 | false | fam2_train_p015_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you explain what the result means?",
"role": "user"
},
{
"content": "I can explain the... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p015",
"generation_request_sha256": "ee1283b50fba0b7cd1882187c327c84959457a31e3ccbed732b2b6312225b119",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | e0f041ffede9b786bcfd2f44ba0d32d5bcf70e46604559a3bbf86be6f0cff547 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_8ec0e90fa217f0c4 |
false | true | explicit_reanalysis | fam2_train_p015 | false | fam2_train_p015_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Do another authenticity check on it.",
"role": "user"
},
{
"content": "I can run a fresh a... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p015",
"generation_request_sha256": "ee1283b50fba0b7cd1882187c327c84959457a31e3ccbed732b2b6312225b119",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | f2d64b718751c9c5474ea46510279cbdadbccedd73117d2309346273e648bb28 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_8ec0e90fa217f0c4 |
false | true | ordinary_image | fam2_train_p016 | false | fam2_train_p016_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p016",
"generation_request_sha256": "ee1283b50fba0b7cd1882187c327c84959457a31e3ccbed732b2b6312225b119",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 2c920a4eea68ab2f0918aa261854907c81ce4d4bec42aee1fd2a03460343d86e | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_ff20d9249e538a57 |
false | true | new_image_or_mixed | fam2_train_p016 | false | fam2_train_p016_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p016",
"generation_request_sha256": "ee1283b50fba0b7cd1882187c327c84959457a31e3ccbed732b2b6312225b119",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 65d9f30cdd0bda54170b56f28f40d1996289fd448c691eb8a8cd32b81ed1bb9e | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_ff20d9249e538a57 |
false | true | ordinary_image | fam2_train_p017 | false | fam2_train_p017_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Read and confirm the text in this picture.",
"role": "user"
},
{
"content": "I read the te... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p017",
"generation_request_sha256": "ee1283b50fba0b7cd1882187c327c84959457a31e3ccbed732b2b6312225b119",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 609f690c2de7ffd46ee17b6d37e004d8e21667013e18881051224e95b954f70c | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_a4a4d99f9231cc77 |
false | true | explicit_reanalysis | fam2_train_p017 | false | fam2_train_p017_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Run a new authenticity check on it again.",
"role": "user"
},
{
"content": "I ran the auth... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p017",
"generation_request_sha256": "ee1283b50fba0b7cd1882187c327c84959457a31e3ccbed732b2b6312225b119",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | d0662b8c3eedb8ba5d37387a8e111eb5d0a0a75b310be8405c560ac638a79da4 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_a4a4d99f9231cc77 |
false | true | explain_existing | fam2_train_p018 | false | fam2_train_p018_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Could you explain how the system decided that?",
"role": "user"
},
{
"content": "Would you... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p018",
"generation_request_sha256": "ee1283b50fba0b7cd1882187c327c84959457a31e3ccbed732b2b6312225b119",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | e1e946b76436c1095a8dc98277a3b7765ce9ad3bc47bc015f568ba8316d26f9c | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_57c45ed439714c9b |
false | true | explicit_reanalysis | fam2_train_p018 | false | fam2_train_p018_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Would you do a new analysis of this picture?",
"role": "user"
},
{
"content": "Would you l... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p018",
"generation_request_sha256": "ee1283b50fba0b7cd1882187c327c84959457a31e3ccbed732b2b6312225b119",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 22f2e8a7cce53dee3a36755fa89f437bb067a302684cafac3c6d9200e3cab25b | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_57c45ed439714c9b |
false | true | ordinary_image | fam2_train_p019 | false | fam2_train_p019_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p019",
"generation_request_sha256": "4fc236f5b8dab819eedeb1c74ee36069f5ae21ea5250d9c3aac8f37d1466fe50",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | a627d3e505da796e783a055f2670637554fec69389a8f2f84d126faf2de28b5b | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_7f83224606edbaaf |
false | true | new_image_or_mixed | fam2_train_p019 | false | fam2_train_p019_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p019",
"generation_request_sha256": "4fc236f5b8dab819eedeb1c74ee36069f5ae21ea5250d9c3aac8f37d1466fe50",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | d8a24c4a414cb6a707731255e9594fc736461a7fdd2fc2a4d7e5c0c1fb86e46f | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_7f83224606edbaaf |
false | true | ordinary_image | fam2_train_p020 | false | fam2_train_p020_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I need to check the text on that picture once more.",
"role": "user"
},
{
"content": "I re... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p020",
"generation_request_sha256": "4fc236f5b8dab819eedeb1c74ee36069f5ae21ea5250d9c3aac8f37d1466fe50",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | b30d18fd0333abf41c5782ac7a5599e531113efdffa0e064ee1e3d1db6728358 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_0d20bf1b48b1d111 |
false | true | explicit_reanalysis | fam2_train_p020 | false | fam2_train_p020_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Could you re-check the authenticity of this photo?",
"role": "user"
},
{
"content": "I ran... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p020",
"generation_request_sha256": "4fc236f5b8dab819eedeb1c74ee36069f5ae21ea5250d9c3aac8f37d1466fe50",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 8ee87d423d1dd35b4cfdfaf85a50c6ad517b6e64b6f4edd8e44b51dfe7f26aa1 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_0d20bf1b48b1d111 |
false | true | explain_existing | fam2_train_p021 | false | fam2_train_p021_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Would you explain how you reached that result?",
"role": "user"
},
{
"content": "I can exp... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p021",
"generation_request_sha256": "4fc236f5b8dab819eedeb1c74ee36069f5ae21ea5250d9c3aac8f37d1466fe50",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 0ec0ea2d7d70d382220576603f599cf15ca0678e3d47e30016ccf94ef922496f | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_8d9fd50c08b7042c |
false | true | explicit_reanalysis | fam2_train_p021 | false | fam2_train_p021_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I’d prefer a new analysis for the authenticity check.",
"role": "user"
},
{
"content": "I ... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p021",
"generation_request_sha256": "4fc236f5b8dab819eedeb1c74ee36069f5ae21ea5250d9c3aac8f37d1466fe50",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 3240e6afae050bdeda5bc57f5124cc2d4910388f131c4b6133d541728c232f8f | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_8d9fd50c08b7042c |
false | true | ordinary_image | fam2_train_p022 | false | fam2_train_p022_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p022",
"generation_request_sha256": "4fc236f5b8dab819eedeb1c74ee36069f5ae21ea5250d9c3aac8f37d1466fe50",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | a28a594e3a0d5fb09da196e928e3984387d946ac9f460d3e3f0e129c54ced762 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_fa6354a4633d090b |
false | true | new_image_or_mixed | fam2_train_p022 | false | fam2_train_p022_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p022",
"generation_request_sha256": "4fc236f5b8dab819eedeb1c74ee36069f5ae21ea5250d9c3aac8f37d1466fe50",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 520ee57824c9288a9a35c0c7b826a27e9131df64e876a6a749bab3596816a81d | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_fa6354a4633d090b |
false | true | ordinary_image | fam2_train_p023 | false | fam2_train_p023_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you look at this wedding picture again?",
"role": "user"
},
{
"content": "I read the t... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p023",
"generation_request_sha256": "83d22a6da1aa308560293024c27dba50fe1c0f32f6a8048d437f66317853b207",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | cb8fd28273ae07d922fa22577df58b01f0e64d42fe7d050c1e2bc3fc5e7f995a | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_378710acea056b3d |
false | true | explicit_reanalysis | fam2_train_p023 | false | fam2_train_p023_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Run the deepfake check on the same image again, please.",
"role": "user"
},
{
"content": "... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p023",
"generation_request_sha256": "83d22a6da1aa308560293024c27dba50fe1c0f32f6a8048d437f66317853b207",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 2e36ea476d0b06753da69abbf8e6f1dafc67796889cfd950761c391e47c2d3ef | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_378710acea056b3d |
false | true | explain_existing | fam2_train_p024 | false | fam2_train_p024_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Why did it come back fake earlier?",
"role": "user"
},
{
"content": "Would you like me to ... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p024",
"generation_request_sha256": "83d22a6da1aa308560293024c27dba50fe1c0f32f6a8048d437f66317853b207",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 3f1c5ad2cc0833af6f941fd0a019798c80a62efbc5d16ef91ecde51955e68018 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_feb3dcbca106f648 |
false | true | explicit_reanalysis | fam2_train_p024 | false | fam2_train_p024_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Let's try a new authenticity scan this time.",
"role": "user"
},
{
"content": "Would you l... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p024",
"generation_request_sha256": "83d22a6da1aa308560293024c27dba50fe1c0f32f6a8048d437f66317853b207",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | c047a22b5a97ce547a2192a1f1416154ed1c290e6c315c899bccfc980649e1b3 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_feb3dcbca106f648 |
false | true | ordinary_image | fam2_train_p025 | false | fam2_train_p025_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p025",
"generation_request_sha256": "83d22a6da1aa308560293024c27dba50fe1c0f32f6a8048d437f66317853b207",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 770ed09d2c0c889cc6b092e18dd4eee9e1ca5f38e44c10a03b1d3fdc6d795628 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_2d38335f012ea5fb |
false | true | new_image_or_mixed | fam2_train_p025 | false | fam2_train_p025_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p025",
"generation_request_sha256": "83d22a6da1aa308560293024c27dba50fe1c0f32f6a8048d437f66317853b207",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 30405621fcc2cc5322c69b25466c1afc8ebd1ce9abf949a94093d22d4d6bf19d | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_2d38335f012ea5fb |
false | true | ordinary_image | fam2_train_p026 | false | fam2_train_p026_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Look at the text in this train station image again.",
"role": "user"
},
{
"content": "I re... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p026",
"generation_request_sha256": "83d22a6da1aa308560293024c27dba50fe1c0f32f6a8048d437f66317853b207",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | a09c163739bcba717a213704df5ca326bccb8f73c25a74fc3cc54448dee2bf28 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_eba9a21dc39c6f91 |
false | true | explicit_reanalysis | fam2_train_p026 | false | fam2_train_p026_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Redo the deepfake check for this platform photo.",
"role": "user"
},
{
"content": "I ran t... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p026",
"generation_request_sha256": "83d22a6da1aa308560293024c27dba50fe1c0f32f6a8048d437f66317853b207",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 10c7a9def4b3097452807dd8f97292f601358a7255193d90b7b1f18516c8790a | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_eba9a21dc39c6f91 |
false | true | explain_existing | fam2_train_p027 | false | fam2_train_p027_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Could you tell me why it was marked real?",
"role": "user"
},
{
"content": "I can explain ... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p027",
"generation_request_sha256": "e24ffc46b9ff304db6ec3959035392c549af3132a42115a2e0cf9089ac52d131",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 1c4da88fa428f9fa334af8423f391e148e86eff687ce2b89248a8d14ccd5ea3a | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_4f6ea9cf6f9963f8 |
false | true | explicit_reanalysis | fam2_train_p027 | false | fam2_train_p027_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can we do another authenticity check on this one?",
"role": "user"
},
{
"content": "I can ... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p027",
"generation_request_sha256": "e24ffc46b9ff304db6ec3959035392c549af3132a42115a2e0cf9089ac52d131",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 58a0ada22c06d23524bc0f7b4d916f12977ededdcfc3b28fbc09a954642230e8 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_4f6ea9cf6f9963f8 |
false | true | ordinary_image | fam2_train_p028 | false | fam2_train_p028_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p028",
"generation_request_sha256": "e24ffc46b9ff304db6ec3959035392c549af3132a42115a2e0cf9089ac52d131",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | da2667dc93f404e7a7b10134eac4c69aff46d3a27b59bcfcb4d1d0081e8b5cca | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_5c3fa1c4c3ba6e37 |
false | true | new_image_or_mixed | fam2_train_p028 | false | fam2_train_p028_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p028",
"generation_request_sha256": "e24ffc46b9ff304db6ec3959035392c549af3132a42115a2e0cf9089ac52d131",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 5a99ea927c0d23c59a84cbaa47a9943bb7fab34e6121d781086ca0f94d6f72a8 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_5c3fa1c4c3ba6e37 |
false | true | ordinary_image | fam2_train_p029 | false | fam2_train_p029_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you read what's written in the image again?",
"role": "user"
},
{
"content": "I read t... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p029",
"generation_request_sha256": "e24ffc46b9ff304db6ec3959035392c549af3132a42115a2e0cf9089ac52d131",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | dc23e44d829bac9fca1b5a7f8fc97a8d20dc68ff92ce5ffdcef08b491313b022 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_06c8833f13e2f812 |
false | true | explicit_reanalysis | fam2_train_p029 | false | fam2_train_p029_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Could you run the authenticity test once more?",
"role": "user"
},
{
"content": "I ran the... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p029",
"generation_request_sha256": "e24ffc46b9ff304db6ec3959035392c549af3132a42115a2e0cf9089ac52d131",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 88d5dd23730344b96b19076452f4e4bda59c33e364ac764e5636830af50d112b | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_06c8833f13e2f812 |
false | true | explain_existing | fam2_train_p030 | false | fam2_train_p030_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I'd like to know why it was marked fake.",
"role": "user"
},
{
"content": "Would you like ... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p030",
"generation_request_sha256": "e24ffc46b9ff304db6ec3959035392c549af3132a42115a2e0cf9089ac52d131",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 799bb405a85d9e983292d9d76bb12b56fd2bb80da9a09682559e5e7b117bb53d | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_e2cb809b8971ee0f |
false | true | explicit_reanalysis | fam2_train_p030 | false | fam2_train_p030_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Let's check its authenticity again please.",
"role": "user"
},
{
"content": "Would you lik... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p030",
"generation_request_sha256": "e24ffc46b9ff304db6ec3959035392c549af3132a42115a2e0cf9089ac52d131",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | f7ddfbec0392a085e25cc321b43e1d8445916e348389c825d30895c08ca4b94d | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_e2cb809b8971ee0f |
false | true | new_image_or_mixed | fam2_train_p031 | false | fam2_train_p031_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p031",
"generation_request_sha256": "4b886e1175b6abc39453ca00baaabdfba1004de1a03c2c71e81c1012b77751bf",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | bc45db5ac72e147b412934279790a5f167323141d04e81d2de6f153d0f3061ed | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_2714c0443bf273ca |
false | true | ordinary_image | fam2_train_p032 | false | fam2_train_p032_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "What does the text in the photo say?",
"role": "user"
},
{
"content": "I read the text in ... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p032",
"generation_request_sha256": "4b886e1175b6abc39453ca00baaabdfba1004de1a03c2c71e81c1012b77751bf",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | a5c88e30925bcb4b425fd880d8e876b7616d9a7792467e8db500038e59149d12 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_d904f31ffd75e757 |
false | true | explicit_reanalysis | fam2_train_p032 | false | fam2_train_p032_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you verify this image again?",
"role": "user"
},
{
"content": "I ran the authenticity ... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p032",
"generation_request_sha256": "4b886e1175b6abc39453ca00baaabdfba1004de1a03c2c71e81c1012b77751bf",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | f6564d76e4da9a94ece40935e11f1e3df196bcc4849ed1a34671eb2c2d1ec0b6 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_d904f31ffd75e757 |
false | true | explain_existing | fam2_train_p033 | false | fam2_train_p033_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Could you clarify how it was verified?",
"role": "user"
},
{
"content": "I can explain the... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p033",
"generation_request_sha256": "4b886e1175b6abc39453ca00baaabdfba1004de1a03c2c71e81c1012b77751bf",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | f3315b45e3fd77a1efd010ac315a2f1a63c89209b0b17f5ee3a72f689c42a7a3 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_a37b9fae15710c7a |
false | true | explicit_reanalysis | fam2_train_p033 | false | fam2_train_p033_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Would you run a new check on this one?",
"role": "user"
},
{
"content": "I can run a fresh... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p033",
"generation_request_sha256": "4b886e1175b6abc39453ca00baaabdfba1004de1a03c2c71e81c1012b77751bf",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 7fbf5ed0a32782427bae3898efcb6a9f17cfc876d70ec6cd910e1ba9c0e65f07 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_a37b9fae15710c7a |
false | true | ordinary_image | fam2_train_p034 | false | fam2_train_p034_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p034",
"generation_request_sha256": "4b886e1175b6abc39453ca00baaabdfba1004de1a03c2c71e81c1012b77751bf",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 06ff3057337deb1b09457536a4c4ba1d01c8ed74a354c1a01878bcfb686b6d94 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_acd163d9321cfa2a |
false | true | new_image_or_mixed | fam2_train_p034 | false | fam2_train_p034_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p034",
"generation_request_sha256": "4b886e1175b6abc39453ca00baaabdfba1004de1a03c2c71e81c1012b77751bf",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | b24e3fbe604007c5ee0fd5e1d3331a9b1d181ec1b907b309c5360f5208c37b35 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_acd163d9321cfa2a |
false | true | ordinary_image | fam2_train_p035 | false | fam2_train_p035_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you read the text in this image again?",
"role": "user"
},
{
"content": "I read the te... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p035",
"generation_request_sha256": "fb8076b10d43da016b6fd706a60c76c667d9588895c8a59f91252e7f84b5e1e1",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 36a38ea4895f67530e97b32e885687378285dfbea3b30a3a8710e5c0effad6cd | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_8f91f702935a602e |
false | true | explicit_reanalysis | fam2_train_p035 | false | fam2_train_p035_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Could you check if this has been altered?",
"role": "user"
},
{
"content": "I ran the auth... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p035",
"generation_request_sha256": "fb8076b10d43da016b6fd706a60c76c667d9588895c8a59f91252e7f84b5e1e1",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | e840afd453a58a18e72248fd03c10df6279907fd9a2888819f102fc748acf5d0 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_8f91f702935a602e |
false | true | explain_existing | fam2_train_p036 | false | fam2_train_p036_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "What was the result for that image earlier?",
"role": "user"
},
{
"content": "Would you li... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p036",
"generation_request_sha256": "fb8076b10d43da016b6fd706a60c76c667d9588895c8a59f91252e7f84b5e1e1",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 7874136103b98974817e82fe81fb6791bbec199397ff0b0eefdf4fd8f4c5d632 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_c72260e8b4f68f79 |
false | true | explicit_reanalysis | fam2_train_p036 | false | fam2_train_p036_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Should I do another check on that picture?",
"role": "user"
},
{
"content": "Would you lik... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p036",
"generation_request_sha256": "fb8076b10d43da016b6fd706a60c76c667d9588895c8a59f91252e7f84b5e1e1",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | cb0e0a3701272f6568f18d5a70dc0e701d0f85deae6083d48ff690ca539c396d | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_c72260e8b4f68f79 |
false | true | ordinary_image | fam2_train_p037 | false | fam2_train_p037_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p037",
"generation_request_sha256": "fb8076b10d43da016b6fd706a60c76c667d9588895c8a59f91252e7f84b5e1e1",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | c9a6e2fae38e58332bfbc6ec9dc4cfabc6c6d4ca8959f95a7ffb9883b7915e7b | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_0875693d4de8a9a8 |
false | true | new_image_or_mixed | fam2_train_p037 | false | fam2_train_p037_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p037",
"generation_request_sha256": "fb8076b10d43da016b6fd706a60c76c667d9588895c8a59f91252e7f84b5e1e1",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 3a25fc1ba973b9ff486058d2fcf8ed4fdf179813a9f8287eb2de5808425dec53 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_0875693d4de8a9a8 |
false | true | ordinary_image | fam2_train_p038 | false | fam2_train_p038_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I’d like to see the text in this image again.",
"role": "user"
},
{
"content": "I read the... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p038",
"generation_request_sha256": "fb8076b10d43da016b6fd706a60c76c667d9588895c8a59f91252e7f84b5e1e1",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 0ebaeac4df99421e96b9f11d561ba0ab18de7261653a1233491fdf5d6bb6b96d | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_1448c77e1c0c1959 |
false | true | explicit_reanalysis | fam2_train_p038 | false | fam2_train_p038_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you verify the authenticity of this picture?",
"role": "user"
},
{
"content": "I ran t... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p038",
"generation_request_sha256": "fb8076b10d43da016b6fd706a60c76c667d9588895c8a59f91252e7f84b5e1e1",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 3ec652589a0818eef1836cc00a48e615d923cc4ae7955b9cfe27b6b2a42058c1 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_1448c77e1c0c1959 |
false | true | explain_existing | fam2_train_p039 | false | fam2_train_p039_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you tell me what you saw in that picture?",
"role": "user"
},
{
"content": "I can expl... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p039",
"generation_request_sha256": "d81b8ec673b4fb88d371b5349393a7d3ca0239ac6e61b3584c4033c0c0a9e3f7",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | b982a537ee43b0ea645c2f0d79ca7e64c428056f77aeba9ff8174eda29cd5088 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_c3d3302afe069f6b |
false | true | explicit_reanalysis | fam2_train_p039 | false | fam2_train_p039_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I'd like to run the check again, please.",
"role": "user"
},
{
"content": "I can run a fre... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p039",
"generation_request_sha256": "d81b8ec673b4fb88d371b5349393a7d3ca0239ac6e61b3584c4033c0c0a9e3f7",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 187ca4095ce43e107f4a31c4f2e0e2a0a177905d7f0cec092a024125c2cad159 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_c3d3302afe069f6b |
false | true | ordinary_image | fam2_train_p040 | false | fam2_train_p040_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p040",
"generation_request_sha256": "d81b8ec673b4fb88d371b5349393a7d3ca0239ac6e61b3584c4033c0c0a9e3f7",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | d0c6176ee2192ec2bb417a5f6f869cf66908a1b24d03fa9f66e90afa4037712f | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_a018e218c7ea46e8 |
false | true | new_image_or_mixed | fam2_train_p040 | false | fam2_train_p040_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p040",
"generation_request_sha256": "d81b8ec673b4fb88d371b5349393a7d3ca0239ac6e61b3584c4033c0c0a9e3f7",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 5c61c86bb015e36f8cf1c80438965d8636d2e9ff295143ee0f2859a5895610f2 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_a018e218c7ea46e8 |
false | true | ordinary_image | fam2_train_p041 | false | fam2_train_p041_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you read the text in the picture again?",
"role": "user"
},
{
"content": "I read the t... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p041",
"generation_request_sha256": "d81b8ec673b4fb88d371b5349393a7d3ca0239ac6e61b3584c4033c0c0a9e3f7",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 66c57767b0a024e14e51185a64ce0912f52f73cce2efa2db9c7d49a640939050 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_709bee6809fa98fe |
false | true | explicit_reanalysis | fam2_train_p041 | false | fam2_train_p041_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I'd like to check it again.",
"role": "user"
},
{
"content": "I ran the authenticity check... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p041",
"generation_request_sha256": "d81b8ec673b4fb88d371b5349393a7d3ca0239ac6e61b3584c4033c0c0a9e3f7",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | fe608ff1f9248346fa15b1c80327838679b99260fa06bbd0ac5c30a12a0a93be | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_709bee6809fa98fe |
false | true | explain_existing | fam2_train_p042 | false | fam2_train_p042_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I want to know why it was marked as fake.",
"role": "user"
},
{
"content": "Would you like... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p042",
"generation_request_sha256": "d81b8ec673b4fb88d371b5349393a7d3ca0239ac6e61b3584c4033c0c0a9e3f7",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | fd95e66b09107e04e550828c5d3c3402d1ff9939808f7b717a24e2dceadf1d5d | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_209c014ee0dc27e2 |
false | true | explicit_reanalysis | fam2_train_p042 | false | fam2_train_p042_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you test this photo again?",
"role": "user"
},
{
"content": "Would you like me to run ... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p042",
"generation_request_sha256": "d81b8ec673b4fb88d371b5349393a7d3ca0239ac6e61b3584c4033c0c0a9e3f7",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | b1b1f4873f8c306f37e5cef60e45e5f411518c861f1bbc1990ed56a79d2d595c | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_209c014ee0dc27e2 |
false | true | ordinary_image | fam2_train_p043 | false | fam2_train_p043_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p043",
"generation_request_sha256": "234cf5ebac779cfef9cd2340dfb54bf9c4526ef074f40e0dad0034f97ea7c1f5",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 0addc78b78610151ed08a57a7f5d637d6e33ee93bd3eb89415f0108e675da9e2 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_71ad469b34ed191d |
false | true | new_image_or_mixed | fam2_train_p043 | false | fam2_train_p043_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p043",
"generation_request_sha256": "234cf5ebac779cfef9cd2340dfb54bf9c4526ef074f40e0dad0034f97ea7c1f5",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 35c83b72d8d6f425feaf823f1c8d8e8ad2baedfa9f826c0004967b8d5c98164e | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_71ad469b34ed191d |
false | true | ordinary_image | fam2_train_p044 | false | fam2_train_p044_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I want to read the text in this picture again.",
"role": "user"
},
{
"content": "I read th... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p044",
"generation_request_sha256": "234cf5ebac779cfef9cd2340dfb54bf9c4526ef074f40e0dad0034f97ea7c1f5",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 5b76b78b6bacf570f7d8444257c64feea6a6917edede1c07a8c8f6c491938395 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_d61ecfab1c57a8eb |
false | true | explicit_reanalysis | fam2_train_p044 | false | fam2_train_p044_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Run another authenticity check on it.",
"role": "user"
},
{
"content": "I ran the authenti... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p044",
"generation_request_sha256": "234cf5ebac779cfef9cd2340dfb54bf9c4526ef074f40e0dad0034f97ea7c1f5",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 83bd5f3754cc97885210b89038b953a6177bb6f7d58583a623b2f508908ac3fe | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_d61ecfab1c57a8eb |
false | true | explain_existing | fam2_train_p045 | false | fam2_train_p045_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Could you explain the result?",
"role": "user"
},
{
"content": "I can explain the stored r... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p045",
"generation_request_sha256": "234cf5ebac779cfef9cd2340dfb54bf9c4526ef074f40e0dad0034f97ea7c1f5",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | a0d5a85f56699533af6ced152f5e25ea86cb99f483874c2958df07779bc3f454 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_9031fd7941fc7743 |
false | true | explicit_reanalysis | fam2_train_p045 | false | fam2_train_p045_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can I have a fresh authenticity check?",
"role": "user"
},
{
"content": "I can run a fresh... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p045",
"generation_request_sha256": "234cf5ebac779cfef9cd2340dfb54bf9c4526ef074f40e0dad0034f97ea7c1f5",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 6b898a54254ab15c94195fbf1cba3e1f39df4b4cd9f62539c704422d69529baa | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_9031fd7941fc7743 |
false | true | ordinary_image | fam2_train_p046 | false | fam2_train_p046_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p046",
"generation_request_sha256": "234cf5ebac779cfef9cd2340dfb54bf9c4526ef074f40e0dad0034f97ea7c1f5",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 3c2108e6016a575234e85dabc56937e47b7d4e6eed470539564a523725e42bac | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_c114072588a2feae |
false | true | new_image_or_mixed | fam2_train_p046 | false | fam2_train_p046_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p046",
"generation_request_sha256": "234cf5ebac779cfef9cd2340dfb54bf9c4526ef074f40e0dad0034f97ea7c1f5",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | fa38014e949f422eafd709aa7d20ca8e07eadaeb80fd068b6d9bd34835f3a3dd | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_c114072588a2feae |
false | true | ordinary_image | fam2_train_p047 | false | fam2_train_p047_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Could you look at this again?",
"role": "user"
},
{
"content": "I read the text in the ima... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p047",
"generation_request_sha256": "18808e4e0b4c02c73f277656e10cba4229b8d15e3e88de0c20d8c525cd5de6d3",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 30955e0dd70e53bc0839f718f59ae865ae6f04a7b0f7294745973561b7d47d73 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_63f05f5cba05997d |
false | true | explicit_reanalysis | fam2_train_p047 | false | fam2_train_p047_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you check it once more?",
"role": "user"
},
{
"content": "I ran the authenticity check... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p047",
"generation_request_sha256": "18808e4e0b4c02c73f277656e10cba4229b8d15e3e88de0c20d8c525cd5de6d3",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | fbbb635c93c5d34f9b1f9dcf63bf02cf0fa75d7e49d48f738460cc36768cb086 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_63f05f5cba05997d |
false | true | explain_existing | fam2_train_p048 | false | fam2_train_p048_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I need an explanation of the result.",
"role": "user"
},
{
"content": "Would you like me t... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p048",
"generation_request_sha256": "18808e4e0b4c02c73f277656e10cba4229b8d15e3e88de0c20d8c525cd5de6d3",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | 6dded2586916c1d089ba0732c2a63b389cfb3c7e91f5d75dffaa0266b7b244eb | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_84fde2a30a66f146 |
false | true | explicit_reanalysis | fam2_train_p048 | false | fam2_train_p048_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Would you like a second authenticity check?",
"role": "user"
},
{
"content": "Would you li... | ACCEPT_CLEAR_OFFER / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p048",
"generation_request_sha256": "18808e4e0b4c02c73f277656e10cba4229b8d15e3e88de0c20d8c525cd5de6d3",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | ba86125d8f7372efdb6a185a96d8ab90c852689a4a2c8fb42e5e60482a543efd | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_84fde2a30a66f146 |
false | true | ordinary_image | fam2_train_p049 | false | fam2_train_p049_b | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p049",
"generation_request_sha256": "18808e4e0b4c02c73f277656e10cba4229b8d15e3e88de0c20d8c525cd5de6d3",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 7a57518a448850273566e46448cd12295a5747dcd3681fb5431e68ca1f0da907 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_c6822e173c62c4b6 |
false | true | new_image_or_mixed | fam2_train_p049 | false | fam2_train_p049_d | {
"active_image_id": "img_2",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
},
{
"availability": "available",
"id": "img_2",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"... | NEW_TARGET_CONTINUATION / new_image_or_mixed generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p049",
"generation_request_sha256": "18808e4e0b4c02c73f277656e10cba4229b8d15e3e88de0c20d8c525cd5de6d3",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | NEW_TARGET_CONTINUATION | retain | null | 6ca196b81fd1a8c7f551382ac761b3e4ef3def565298cb4fc7bb6fed06abcf39 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_c6822e173c62c4b6 |
false | true | ordinary_image | fam2_train_p050 | false | fam2_train_p050_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you read through the text in this again?",
"role": "user"
},
{
"content": "I read the ... | REPEAT_PREVIOUS_OPERATION / ordinary_image generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p050",
"generation_request_sha256": "18808e4e0b4c02c73f277656e10cba4229b8d15e3e88de0c20d8c525cd5de6d3",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | c0d62c956205d0bec07c346741a5de2702fcd6e4ebe798f0ba121a00fa7bca63 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_26714639ebc14048 |
false | true | explicit_reanalysis | fam2_train_p050 | false | fam2_train_p050_d | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "I'd like another authenticity check on it.",
"role": "user"
},
{
"content": "I ran the aut... | REPEAT_PREVIOUS_OPERATION / explicit_reanalysis generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | DEEPFAKE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | DEEPFAKE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p050",
"generation_request_sha256": "18808e4e0b4c02c73f277656e10cba4229b8d15e3e88de0c20d8c525cd5de6d3",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | REPEAT_PREVIOUS_OPERATION | retain | null | 4fb030921f226970afdead3b14257191f05ecc2e066f524f3d6821d44cfc3b23 | DEEPFAKE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_26714639ebc14048 |
false | true | explain_existing | fam2_train_p051 | false | fam2_train_p051_b | {
"active_image_id": "img_1",
"assets": [
{
"availability": "available",
"id": "img_1",
"kind": "image",
"support": "supported"
}
],
"history": [
{
"content": "Can you tell me why it was marked real?",
"role": "user"
},
{
"content": "I can explain th... | ACCEPT_CLEAR_OFFER / explain_existing generation condition; code-owned offer, roles and neutral continuation | generation_condition_ai_check_pending | context_pair | BASE | Read the immediately preceding operation/offer to resolve the neutral current reply; BASE for the ordinary/explanation task, DEEPFAKE for the fresh authenticity operation. Stored verdict metadata is not sticky intent. | codex_full_input_policy_adjudication_2026_11_17 | BASE | train | {
"assistant_turn_owner": "code",
"family_id": "fam2_train_p051",
"generation_request_sha256": "6a635a7d94b8b890e3126a7b964a6c393046fc5d5914e175ade4b2190a05484d",
"generator": "qwen3:32b-q4_K_M",
"generator_digest": "030ee887880fc378860c2dd35101da424377520441ae4bfe7be6deff8ade7840",
"source_slot": "fam2_tra... | ACCEPT_CLEAR_OFFER | retain | null | d0df98a23d2e35f1ca7b899fa9652f130d5593f7c2dbb0c8bc92befc3250ba62 | BASE | TRAIN_ONLY | AI_REVIEWED_EXPLORATORY_PILOT | scenario_train_b0a7405d6c1bbb33 |
End of preview. Expand in Data Studio
Polyvalent Router Requests Dataset
The reviewed training, development, and evaluation splits used to train the Polyvalent LAYA dynamic router.
Dataset Summary
- Total Reviewed Rows: 2,500
train.jsonl: 2,000 rowsdev.jsonl: 200 rowstest.jsonl: 300 rows
- Supplementary Evidence: 84 added boundary edge cases (
added_training_cases.jsonl), review decisions (review_decisions.jsonl), exclusion records (exclusions.jsonl), and prompt-critic audit trails (dataset_evidence/).
Generation and AI-Review Statement
This dataset contains synthetic user dialogue states generated to model realistic multimodal tool-interaction contexts:
- User Wording & Dialogue Generation: Synthesized using Qwen3-32B (Apache-2.0).
- Critique & Conflict Identification: Filtered and critiqued using Gemma3-27B (Gemma Terms of Use: outputs are not Model Derivatives; subject to prohibited-use policy).
- Policy Adjudication & Edge-Case Insertion: Reviewed, audited, and supplemented with 84 hard boundary cases using OpenAI Codex under policy supervision.
Data Schema
Each record is a JSON object with:
id: Unique record identifier (e.g.fam2_train_p001_b)category: Scenario category (e.g.ordinary_image,new_image_or_mixed,forensic_inquiry)input: 5-field runtime state:message: Current user query stringhistory: List of prior user and assistant turnsassets: List of uploaded assets withid,kind,support,availabilityactive_image_id: Currently selected image asset or nulllast_analysis: Prior verdict record or null
proposed_route: Target class (BASEorDEEPFAKE)review_decision: Audit decision (retain,modify)
See DATA_DICTIONARY.md for full field specifications.
Licence and Notices
- Dataset licence: CC BY 4.0
- Third-party attributions:
- Dialogue synthesis assisted by Qwen3-32B (Apache-2.0).
- Validation critique assisted by Gemma3-27B (governed by Google Gemma Terms of Use).
- Policy review assisted by OpenAI Codex.
Citation
@article{polyvalent2026,
title={Polyvalent: Single-Base Multi-Specialist Architecture with Dynamic Adapter Residency},
author={Polyvalent Research Team},
year={2026}
}
- Downloads last month
- -