{
  "schema_version": 1,
  "title": "Reproduction: Distributed Direct Preference Optimization",
  "emoji": "🎯",
  "space_id": "SabaPivot/repro-distributed-direct-preference-optimization",
  "paper": {
    "arxiv_id": "2605.20696"
  },
  "tags": [
    "icml2026-repro",
    "paper-ljNZyrAlaa"
  ],
  "updated_at": "2026-08-03T02:07:27.692915+00:00",
  "root": {
    "slug": "index",
    "title": "Reproduction: Distributed Direct Preference Optimization",
    "file": "pages/index.md",
    "children": [
      {
        "slug": "executive-summary",
        "title": "Executive summary",
        "file": "pages/executive-summary/page.md",
        "children": []
      },
      {
        "slug": "claim-1",
        "title": "Claim 1: Theorem 5.1 gives the first convergence bound for Federated DPO (FedDPO) under partial client participation, showing gradient-norm error scaling with local steps E, rounds R, sampled clients S, and gradient variance ζ²_g (Theorem 5.1).",
        "file": "pages/claim-1/page.md",
        "children": []
      },
      {
        "slug": "claim-2",
        "title": "Claim 2: Corollary 5.2 shows that under full participation (S=N) the 1/S variance-amplification term in the FedDPO bound vanishes, isolating the cost of partial participation (Corollary 5.2).",
        "file": "pages/claim-2/page.md",
        "children": []
      },
      {
        "slug": "claim-3",
        "title": "Claim 3: Theorem 5.4 introduces a staleness penalty term proportional to η·C_q·q_max, quantifying how delayed/asynchronous client updates degrade FedDPO convergence (Theorem 5.4).",
        "file": "pages/claim-3/page.md",
        "children": []
      },
      {
        "slug": "claim-4",
        "title": "Claim 4: Theorem 5.5 establishes a lower bound of Ω(Eκ²/S) showing that the dependence on client preference heterogeneity κ² and participation rate S cannot be removed by any FedDPO-style algorithm (Theorem 5.5).",
        "file": "pages/claim-4/page.md",
        "children": []
      },
      {
        "slug": "claim-5",
        "title": "Claim 5: Theorem 6.1 proves DecDPO (decentralized DPO) converges at rate O(1/√R + 1/(R(1−ρ²))) where ρ is the spectral gap of the communication graph, with variance and heterogeneity terms scaled by 1/(1−ρ²) (Theorem 6.1).",
        "file": "pages/claim-5/page.md",
        "children": []
      },
      {
        "slug": "claim-6",
        "title": "Claim 6: Numerical experiments on the Stanford Human Preferences dataset with N=5 agents empirically confirm the predicted effects of local step count, participation rate, staleness, and network topology on convergence (Section 7, Numerical Results).",
        "file": "pages/claim-6/page.md",
        "children": []
      },
      {
        "slug": "conclusion",
        "title": "Conclusion",
        "file": "pages/conclusion/page.md",
        "children": []
      }
    ]
  },
  "agent_view_tokens": 3400,
  "revision": "1784656265000000000",
  "evidence_provenance": "PROVENANCE.md"
}