Step 2: tool results

get_claim

{
  "include": [
    "arguments",
    "subclaims",
    "related",
    "dependents"
  ],
  "claim_id": "a5665880-9436-4582-a906-e2161fe8e71a"
}

{"claim":{"id":"a5665880-9436-4582-a906-e2161fe8e71a","kind":"claim","canonical_form":"With features selected and labelled by a supervised pipeline, SAE steering approaches LoRA performance on AxBench.","claim_type":"empirical_verifiable","state":"active","decomposition_status":"complete","importance":0.5,"contestation":0.55,"domains":[],"tags":[{"slug":"large-language-models","name":"Large language models"},{"slug":"mechanistic-interpretability","name":"Mechanistic Interpretability"},{"slug":"ai-capability-evaluation","name":"AI Capability Evaluation"}],"steward_state":"running","stewarded_at":"2026-11-15T07:15:48.797Z","canonical_direction_note":"The canonical form states the positive performance result the source reports, rather than its negation.","created_at":"2026-11-13T01:05:17.078Z","updated_at":"2026-11-15T07:15:45.589Z","page_url":"https://minerval.ai/claims/a5665880-9436-4582-a906-e2161fe8e71a"},"assessment":null,"coverage":{"examined":false,"on":[],"note":"No judgment of this node is in force. That is not a finding either way."},"quantity":null,"threshold_of":null,"question":null,"formalization":null,"formalization_pending":null,"formalization_history":[],"verification":null,"lean_checks":[],"lean_checks_total":0,"waiting_messages":[{"trigger":"structure_and_assess","from":"the system","sender_node_id":null,"sent_at":"2026-11-13T01:05:17.168Z","paid":false,"summary":"New claim onboarded. Give it the structure and assessment its importance warrants; a minor or settled claim may need no decomposition."},{"trigger":"mandate_plan","from":"a mandate's plan","sender_node_id":null,"sent_at":"2026-11-15T07:15:45.596Z","paid":true,"summary":"First assessment. Read the source reporting this result and compare its setup with the original steering benchmark (arXiv 2501.17148; not held as a source, open its abstract or HTML page directly): mo…"}],"requests":[],"requests_total":0,"subclaim_count":0,"instance_count":1,"source_map":null,"arguments":[],"dependents":[{"id":"cb2d56c9-fcce-47d3-b2b3-5127f10f3293","text":"Under what conditions, if any, do sparse-autoencoder and other dictionary-learning features outperform linear probes and difference-of-means steering on labeled language-model tasks?","claim_type":"empirical_derived","node_kind":"question","relation_type":"answers","reasoning":"A second conditional answer for steering, from the 2026 rebuttal titled as a claim that SAEs can outperform simple baselines: when features are matched to concepts with labeled data, SAE steering approaches the fine-tuning reference on the same benchmark.","importance":0.4,"assessment_status":null,"assessment_confidence":null,"assessment_credence":null,"bounty_micro_usd":null,"checked":null,"formal":false}],"dependents_total":1,"related":[{"id":"7c1050dc-7ca9-492e-8d3f-ade8d793de4c","text":"On AxBench, SAE steering underperforms both prompting and difference-in-means representations.","link_kind":"related","reasoning":"A 2026 partial rebuttal of the AxBench SAE steering result that changes how features are selected and labelled; it bears on how the original result generalizes, not on whether it was measured correctly.","assessment_status":"verified","assessment_confidence":0.85,"assessment_credence":0.92}],"subclaims":[],"subclaims_assessed":0,"unplaced_argument_passages":[{"instance_id":"d95ab2cf-ba02-4520-aebc-a00bc1ef6f3f","source":{"id":"ff9222f2-759d-4782-a7ff-8c811f759780","url":"https://arxiv.org/pdf/2605.31183","title":"Steering LLMs? Actually, Sparse Autoencoders can outperform simple baselines"},"verbatim_text":"Figure 2 depicts the overall steering performance and shows a general improvement over those reported in Wu\net al. (2025). We observe that multiple fora provide labels that result in feature-steering performance on par\nwith the LoRA reference.","context":"Results on AxBench aggregated rating","stance":"advances","direction":"for"}],"unplaced_argument_passages_total":1}