# Causal Faithfulness Evaluation on FineWeb Temporal Source: [Causal Faithfulness Evaluation on FineWeb Temporal]() Collection index: [llms.txt]() Data updated: 2026-05-22T07:41:28.496Z Evaluation of model causal faithfulness on the FineWeb Temporal dataset, measured using nAOPC, nComp, and \|∆CE\| scores. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Causal Faithfulness Evaluation | | Dataset (including any stated split) | FineWeb Temporal | | Metric | nAOPC | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 28 of 28 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [nAOPC](): Higher is better; unit: SCALAR. - [nComp](): Higher is better; unit: SCALAR. - [Delta CE](): Lower is better; unit: SCALAR. ## Recorded results | Method | nAOPC | Result date | Report paper | | --- | --- | --- | --- | | GAE | 0.985 | Not recorded | 6a0fe320f984739ed6739e05 | | Retrain | 0.983 | Not recorded | 6a0fe320f984739ed6739e05 | | FaithfulSAE | 0.982 | Not recorded | 6a0fe320f984739ed6739e05 | | SAEBoost | 0.971 | Not recorded | 6a0fe320f984739ed6739e05 | | TERM | 0.965 | Not recorded | 6a0fe320f984739ed6739e05 | | Finetune | 0.963 | Not recorded | 6a0fe320f984739ed6739e05 | | Fixed | 0.962 | Not recorded | 6a0fe320f984739ed6739e05 | | GAE | 0.96 | Not recorded | 6a0fe320f984739ed6739e05 | | SAEBoost | 0.958 | Not recorded | 6a0fe320f984739ed6739e05 | | FaithfulSAE | 0.936 | Not recorded | 6a0fe320f984739ed6739e05 | | GAE | 0.915 | Not recorded | 6a0fe320f984739ed6739e05 | | SAEBoost | 0.908 | Not recorded | 6a0fe320f984739ed6739e05 | | Retrain | 0.895 | Not recorded | 6a0fe320f984739ed6739e05 | | Retrain | 0.894 | Not recorded | 6a0fe320f984739ed6739e05 | | Finetune | 0.859 | Not recorded | 6a0fe320f984739ed6739e05 | | FaithfulSAE | 0.858 | Not recorded | 6a0fe320f984739ed6739e05 | | Fixed | 0.857 | Not recorded | 6a0fe320f984739ed6739e05 | | Finetune | 0.856 | Not recorded | 6a0fe320f984739ed6739e05 | | TERM | 0.853 | Not recorded | 6a0fe320f984739ed6739e05 | | TERM | 0.845 | Not recorded | 6a0fe320f984739ed6739e05 | | Fixed | 0.839 | Not recorded | 6a0fe320f984739ed6739e05 | | GAE | 0.768 | Not recorded | 6a0fe320f984739ed6739e05 | | Retrain | 0.766 | Not recorded | 6a0fe320f984739ed6739e05 | | TERM | 0.741 | Not recorded | 6a0fe320f984739ed6739e05 | | Fixed | 0.735 | Not recorded | 6a0fe320f984739ed6739e05 | | Finetune | 0.725 | Not recorded | 6a0fe320f984739ed6739e05 | | FaithfulSAE | 0.725 | Not recorded | 6a0fe320f984739ed6739e05 | | SAEBoost | 0.704 | Not recorded | 6a0fe320f984739ed6739e05 | ### GAE nAOPC: 0.985. Result ID: 6a1008a4fbc373592c08afce. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### Retrain nAOPC: 0.983. Result ID: 6a1008a4fbc373592c08afcb. [Reporting paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### FaithfulSAE nAOPC: 0.982. Result ID: 6a1008a4fbc373592c08afcd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### SAEBoost nAOPC: 0.971. Result ID: 6a1008a4fbc373592c08afcc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### TERM nAOPC: 0.965. Result ID: 6a1008a4fbc373592c08afc9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### Finetune nAOPC: 0.963. Result ID: 6a1008a4fbc373592c08afca. [Reporting paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### Fixed nAOPC: 0.962. Result ID: 6a1008a4fbc373592c08afc8. [Reporting paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### GAE nAOPC: 0.96. Result ID: 6a1008a4fbc373592c08af9d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### SAEBoost nAOPC: 0.958. Result ID: 6a1008a4fbc373592c08af9b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### FaithfulSAE nAOPC: 0.936. Result ID: 6a1008a4fbc373592c08af9c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### GAE nAOPC: 0.915. Result ID: 6a1008a4fbc373592c08afa4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### SAEBoost nAOPC: 0.908. Result ID: 6a1008a4fbc373592c08afa2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### Retrain nAOPC: 0.895. Result ID: 6a1008a4fbc373592c08af9a. [Reporting paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### Retrain nAOPC: 0.894. Result ID: 6a1008a4fbc373592c08afa1. [Reporting paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### Finetune nAOPC: 0.859. Result ID: 6a1008a4fbc373592c08afa0. [Reporting paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### FaithfulSAE nAOPC: 0.858. Result ID: 6a1008a4fbc373592c08afa3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### Fixed nAOPC: 0.857. Result ID: 6a1008a4fbc373592c08af97. [Reporting paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### Finetune nAOPC: 0.856. Result ID: 6a1008a4fbc373592c08af99. [Reporting paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### TERM nAOPC: 0.853. Result ID: 6a1008a4fbc373592c08af98. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### TERM nAOPC: 0.845. Result ID: 6a1008a4fbc373592c08af9f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### Fixed nAOPC: 0.839. Result ID: 6a1008a4fbc373592c08af9e. [Reporting paper]() Recorded evaluation configuration: {"Model":"Pythia-1.4B"}. This record does not assert independent reproduction. ### GAE nAOPC: 0.768. Result ID: 6a1008a4fbc373592c08afc7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### Retrain nAOPC: 0.766. Result ID: 6a1008a4fbc373592c08afc4. [Reporting paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### TERM nAOPC: 0.741. Result ID: 6a1008a4fbc373592c08afc2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### Fixed nAOPC: 0.735. Result ID: 6a1008a4fbc373592c08afc1. [Reporting paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### Finetune nAOPC: 0.725. Result ID: 6a1008a4fbc373592c08afc3. [Reporting paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### FaithfulSAE nAOPC: 0.725. Result ID: 6a1008a4fbc373592c08afc6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. ### SAEBoost nAOPC: 0.704. Result ID: 6a1008a4fbc373592c08afc5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"GPT-2 Small"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.