{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/code/extract-content","entry":"extract_content","source":"Syntology graph, per-sample; not an archive number","read_at":"2026-09-24T18:15:14+00:00","claim":"Names are grouped by exact entry-name string. Same-named routines are NOT asserted to be equivalent; 'ran' means executed on a synthesized fixture, not correctness. n_samples_ran = sum of by_status over every status except 'unverified' (ran_draft_wrong and ran_fixture are failures of Syntology's instrument, not of the code); n_papers_ran = papers with at least one such sample.","status_vocabulary":{"ran_honours":"ran, honoured the contract we drafted","ran_violates":"ran, violated the contract we drafted","ran_draft_wrong":"ran; our contract draft was wrong, not the code","ran_fixture":"ran; our fixture could not drive it","ran":"ran on a synthesized input","unverified":"unverified (harvested, no recorded run)"},"n_papers":17,"n_papers_ran":12,"units":"n_samples, n_samples_ran, n_samples_fingerprinted and by_status count distinct code bodies (code_sha256); n_places and n_places_pointer_only count places, one per (paper, code body) pair, which is also the unit of the samples list","n_samples":16,"n_samples_ran":10,"n_samples_fingerprinted":8,"n_places":18,"n_places_pointer_only":8,"by_status":{"ran_honours":1,"ran_violates":0,"ran_draft_wrong":2,"ran_fixture":0,"ran":7,"unverified":6},"syntology":{"atlas_url":null,"mcp":null,"mcp_per_sample":{"tool":"get_code","arguments_in":"samples[].mcp_get_code"},"developers":"https://syntology.ai/developers"},"samples":[{"arxiv_id":"2606.05183","paper":"/paper/arxiv-2606-05183","title":"The Granularity Gap: A Multi-Dimensional Cross-Generational Audit of Sycophancy in Gemini Models","date":null,"month_inferred_from_arxiv_id":"2026-06","title_source":"syntology","repo":"pskeough/The-Granularity-Gap","path":"pipeline/01_run_panel.py","file_url":"https://github.com/pskeough/The-Granularity-Gap/blob/HEAD/pipeline/01_run_panel.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"code_sha256_prefix":"8d568be75e039c26","mcp_get_code":{"code_sha256":"8d568be75e039c26"}},{"arxiv_id":"2510.05691","paper":"/paper/arxiv-2510-05691","title":"DecEx-RAG: Boosting Agentic Retrieval-Augmented Generation with Decision and Execution Optimization via Process Supervision","date":null,"month_inferred_from_arxiv_id":"2025-10","title_source":"syntology","repo":"sdsxdxl/DecEx-RAG","path":"rollout.py","file_url":"https://github.com/sdsxdxl/DecEx-RAG/blob/HEAD/rollout.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"507009efa4744397","mcp_get_code":{"code_sha256":"507009efa4744397"}},{"arxiv_id":"2506.04180","paper":"/paper/superwriter-reflection-driven-long-form","title":"SuperWriter: Reflection-Driven Long-Form Generation with Large Language Models","date":"2025-06-04","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"mozhu621/superwriter","path":"Agent/Super_write_agent.py","file_url":"https://github.com/mozhu621/superwriter/blob/HEAD/Agent/Super_write_agent.py","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"af5438bff5c31194","mcp_get_code":{"code_sha256":"af5438bff5c31194"}},{"arxiv_id":"2503.10582","paper":"/paper/visualwebinstruct-scaling-up-multimodal","title":"VisualWebInstruct: Scaling up Multimodal Instruction Data through Web Search","date":"2025-03-13","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"tiger-ai-lab/visualwebinstruct","path":"VisualWebInstruct/acc_tree.py","file_url":"https://github.com/tiger-ai-lab/visualwebinstruct/blob/HEAD/VisualWebInstruct/acc_tree.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"2d25b921e96b5b59","mcp_get_code":{"code_sha256":"2d25b921e96b5b59"}},{"arxiv_id":"2502.09723","paper":"/paper/making-them-a-malicious-database-exploiting","title":"QueryAttack: Jailbreaking Aligned Large Language Models Using Structured Non-natural Query Language","date":"2025-02-13","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"horizonsinzqs/queryattack","path":"evaluate_responses.py","file_url":"https://github.com/horizonsinzqs/queryattack/blob/HEAD/evaluate_responses.py","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"36d4f4fdeeda8117","mcp_get_code":{"code_sha256":"36d4f4fdeeda8117"}},{"arxiv_id":"2501.04694","paper":"/paper/epicoder-encompassing-diversity-and","title":"EpiCoder: Encompassing Diversity and Complexity in Code Generation","date":"2025-01-08","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"DeepLearnXMU/EpiCoder","path":"debug/run_test_iter0.py","file_url":"https://github.com/DeepLearnXMU/EpiCoder/blob/HEAD/debug/run_test_iter0.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"code_sha256_prefix":"e3c9ed0f508ba2ea","mcp_get_code":{"code_sha256":"e3c9ed0f508ba2ea"}},{"arxiv_id":"2501.04694","paper":"/paper/epicoder-encompassing-diversity-and","title":"EpiCoder: Encompassing Diversity and Complexity in Code Generation","date":"2025-01-08","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"microsoft/EpiCoder","path":"gen/gen_code.py","file_url":"https://github.com/microsoft/EpiCoder/blob/HEAD/gen/gen_code.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"d0f9fa36c5d6a449","mcp_get_code":{"code_sha256":"d0f9fa36c5d6a449"}},{"arxiv_id":"2410.10343","paper":"/paper/locking-down-the-finetuned-llms-safety","title":"Locking Down the Finetuned LLMs Safety","date":"2024-10-14","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"zhu-minjun/safetylock","path":"eval/eval_utils/openai_gpt4_judge.py","file_url":"https://github.com/zhu-minjun/safetylock/blob/HEAD/eval/eval_utils/openai_gpt4_judge.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"e3b723ac5b19f39c","mcp_get_code":{"code_sha256":"e3b723ac5b19f39c"}},{"arxiv_id":"2406.16306","paper":"/paper/cascade-reward-sampling-for-efficient","title":"Cascade Reward Sampling for Efficient Decoding-Time Alignment","date":"2024-06-24","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"lblaoke/CARDS","path":"evaluation/eval_ai.py","file_url":"https://github.com/lblaoke/CARDS/blob/HEAD/evaluation/eval_ai.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"cf6e7912b92ce301","mcp_get_code":{"code_sha256":"cf6e7912b92ce301"}},{"arxiv_id":"2406.12091","paper":"/paper/is-poisoning-a-real-threat-to-llm-alignment","title":"Is poisoning a real threat to LLM alignment? Maybe more so than you think","date":"2024-06-17","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"pankayaraj/RLHFPoisoning","path":"evaluation/gpt_4_evaluation_script.py","file_url":"https://github.com/pankayaraj/RLHFPoisoning/blob/HEAD/evaluation/gpt_4_evaluation_script.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"code_sha256_prefix":"cf6e7912b92ce301","mcp_get_code":{"code_sha256":"cf6e7912b92ce301"}},{"arxiv_id":"2404.01099","paper":"/paper/what-s-in-your-safe-data-identifying-benign","title":"What is in Your Safe Data? Identifying Benign Data that Breaks Safety","date":"2024-04-01","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"princeton-nlp/benign-data-breaks-safety","path":"safety_evaluation/eval_utils/openai_gpt4_judge.py","file_url":"https://github.com/princeton-nlp/benign-data-breaks-safety/blob/HEAD/safety_evaluation/eval_utils/openai_gpt4_judge.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"e28da1bf4c00b938","mcp_get_code":{"code_sha256":"e28da1bf4c00b938"}},{"arxiv_id":"2403.04783","paper":"/paper/autodefense-multi-agent-llm-defense-against","title":"AutoDefense: Multi-Agent LLM Defense against Jailbreak Attacks","date":"2024-03-02","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"xhmy/autodefense","path":"evaluator/gpt4_evaluator.py","file_url":"https://github.com/xhmy/autodefense/blob/HEAD/evaluator/gpt4_evaluator.py","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"b5d083a56fe636a1","mcp_get_code":{"code_sha256":"b5d083a56fe636a1"}},{"arxiv_id":"2402.11753","paper":"/paper/artprompt-ascii-art-based-jailbreak-attacks","title":"ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMs","date":"2024-02-19","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"uw-nsl/ArtPrompt","path":"jailbreak/safe_eval.py","file_url":"https://github.com/uw-nsl/ArtPrompt/blob/HEAD/jailbreak/safe_eval.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"48fa5a3bd20a9af2","mcp_get_code":{"code_sha256":"48fa5a3bd20a9af2"}},{"arxiv_id":"2402.02172","paper":"/paper/collaborative-agents-for-software-engineering","title":"CodeAgent: Autonomous Communicative Agents for Code Review","date":null,"month_inferred_from_arxiv_id":"2024-02","title_source":"archive","repo":"code4agent/codeagent","path":"Baselines/codeBERT.py","file_url":"https://github.com/code4agent/codeagent/blob/HEAD/Baselines/codeBERT.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"7744c07018b338f3","mcp_get_code":{"code_sha256":"7744c07018b338f3"}},{"arxiv_id":"2401.06373","paper":"/paper/how-johnny-can-persuade-llms-to-jailbreak","title":"How Johnny Can Persuade LLMs to Jailbreak Them: Rethinking Persuasion to Challenge AI Safety by Humanizing LLMs","date":"2024-01-12","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"chats-lab/persuasive_jailbreaker","path":"PAP_Better_Incontext_Sample/mutation_utils.py","file_url":"https://github.com/chats-lab/persuasive_jailbreaker/blob/HEAD/PAP_Better_Incontext_Sample/mutation_utils.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"code_sha256_prefix":"04e9ab3c39976275","mcp_get_code":{"code_sha256":"04e9ab3c39976275"}},{"arxiv_id":"2025.findings-emnlp.479","paper":null,"title":"arXiv:2025.findings-emnlp.479","date":null,"month_inferred_from_arxiv_id":null,"title_source":null,"repo":"FoundationAgents/SPO","path":"utils/llm_client.py","file_url":"https://github.com/FoundationAgents/SPO/blob/HEAD/utils/llm_client.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"d78a68f48b9d2a41","mcp_get_code":{"code_sha256":"d78a68f48b9d2a41"}},{"arxiv_id":"2025.findings-acl.298","paper":null,"title":"arXiv:2025.findings-acl.298","date":null,"month_inferred_from_arxiv_id":null,"title_source":null,"repo":"horizonsinzqs/QueryAttack","path":"evaluate_responses.py","file_url":"https://github.com/horizonsinzqs/QueryAttack/blob/HEAD/evaluate_responses.py","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"36d4f4fdeeda8117","mcp_get_code":{"code_sha256":"36d4f4fdeeda8117"}},{"arxiv_id":"2025.acl-long.851","paper":null,"title":"arXiv:2025.acl-long.851","date":null,"month_inferred_from_arxiv_id":null,"title_source":null,"repo":"uw-nsl/SafeDecoding","path":"exp/safe_eval.py","file_url":"https://github.com/uw-nsl/SafeDecoding/blob/HEAD/exp/safe_eval.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"00a0bbf79e4f82c8","mcp_get_code":{"code_sha256":"00a0bbf79e4f82c8"}}]}