Step 2: tool results

get_claim

{
  "include": [
    "arguments",
    "subclaims",
    "dependents"
  ],
  "claim_id": "738d9786-0b87-4b75-9c33-cabe43c903d5"
}

{"claim":{"id":"738d9786-0b87-4b75-9c33-cabe43c903d5","kind":"claim","canonical_form":"Simple non-SAE baselines match SAE results on dataset-quality, spurious-correlation, and multi-token probing tasks.","claim_type":"empirical_derived","state":"active","decomposition_status":"complete","importance":0.4,"contestation":0.45,"domains":[],"tags":[{"slug":"mechanistic-interpretability","name":"Mechanistic Interpretability"},{"slug":"shortcut-learning","name":"Shortcut Learning"}],"steward_state":"running","stewarded_at":"2026-11-15T07:16:54.311Z","canonical_direction_note":"The affirmative form states the source's specific finding that simple non-SAE baselines can achieve results similar to SAEs across these probing applications.","created_at":"2026-11-13T00:52:22.170Z","updated_at":"2026-11-15T07:15:45.626Z","page_url":"https://minerval.ai/claims/738d9786-0b87-4b75-9c33-cabe43c903d5"},"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":"2026-11-13T00:52:22.325Z","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":"2026-11-15T07:15:45.628Z","paid":true,"summary":"First assessment. The claim covers three task families. For each, find the study that reported an SAE result and the study that ran a non-SAE baseline on the same task, and compare: data, model, how t…"}],"requests":[],"requests_total":0,"subclaim_count":0,"instance_count":1,"source_map":null,"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":"The deflationary answer to the candidate conditions: in the settings where SAE probes looked advantageous (spotting dataset defects, spurious correlations, pooling across tokens), a simple non-SAE baseline matched them once it was tried.","importance":0.4,"assessment_status":null,"assessment_confidence":null,"assessment_credence":null,"bounty_micro_usd":null,"checked":null,"formal":false}],"dependents_total":1,"arguments":[],"subclaims":[],"subclaims_assessed":0,"unplaced_argument_passages":[{"instance_id":"1175c448-86b3-406f-8812-847c680986d6","source":{"id":"b45014e2-63ec-41af-8a6f-59db01b1e0fc","url":"https://arxiv.org/pdf/2502.16681","title":"Are Sparse Autoencoders Useful? A Case Study in Sparse Probing"},"verbatim_text":"Although we first found this result\nwith SAE probes, we later found that logistic regression was\nalso capable of the same identification.","context":"CoLA mislabeled examples.","stance":"advances","direction":"for"},{"instance_id":"79baa35d-2745-4fc6-a02c-379ff0242b70","source":{"id":"b45014e2-63ec-41af-8a6f-59db01b1e0fc","url":"https://arxiv.org/pdf/2502.16681","title":"Are Sparse Autoencoders Useful? A Case Study in Sparse Probing"},"verbatim_text":"Although we discovered this by inves-\ntigating SAE latents, we could reach the same conclusion\nwith baselines","context":"AI vs human text dataset ending-punctuation spurious feature.","stance":"advances","direction":"for"}],"unplaced_argument_passages_total":2}