logbook.json

{
  "schema_version": 1,
  "title": "Reproduction: Learning When to Attend: Conditional Memory Access for Long-Context LLMs",
  "emoji": "🎯",
  "space_id": "keepsloading/repro-learning-when-to-attend-conditional-memory-access-for-long-context-llms",
  "paper": {
    "arxiv_id": "2603.17484"
  },
  "tags": [
    "icml2026-repro",
    "paper-UcGwFXWnbn",
    "toy-scale-verified"
  ],
  "updated_at": "2026-07-19T12:04:31+00:00",
  "root": {
    "slug": "index",
    "title": "Reproduction: Learning When to Attend: Conditional Memory Access for Long-Context LLMs",
    "file": "pages/index.md",
    "children": [
      {
        "slug": "executive-summary",
        "title": "executive-summary",
        "file": "pages/executive-summary/page.md",
        "children": []
      },
      {
        "slug": "claim-1-l2a-preserves-quality-at-toy-scale-1-7b-model-8k-context",
        "title": "Claim 1: L2A preserves quality at toy scale (1.7B model, 8K context)",
        "file": "pages/claim-1-l2a-preserves-quality-at-toy-scale-1-7b-model-8k-context/page.md",
        "children": []
      },
      {
        "slug": "claim-2-algorithmic-verification-of-l2a-gating-logic",
        "title": "Claim 2: Algorithmic verification of L2A gating logic",
        "file": "pages/claim-2-algorithmic-verification-of-l2a-gating-logic/page.md",
        "children": []
      },
      {
        "slug": "claim-3-l2a-skips-global-attention-for-80-of-tokens",
        "title": "Claim 3: L2A skips global attention for ~80% of tokens",
        "file": "pages/claim-3-l2a-skips-global-attention-for-80-of-tokens/page.md",
        "children": []
      },
      {
        "slug": "claim-4-sparse-attention-pruning-reduces-kv-cache-by-50",
        "title": "Claim 4: Sparse attention pruning reduces KV-cache by 50%",
        "file": "pages/claim-4-sparse-attention-pruning-reduces-kv-cache-by-50/page.md",
        "children": []
      },
      {
        "slug": "conclusion",
        "title": "Conclusion",
        "file": "pages/conclusion/page.md",
        "children": []
      }
    ]
  },
  "agent_view_tokens": 3338,
  "revision": "1784462671297273900"
}