File size: 3,028 Bytes
5baae82 ea3a71e 5baae82 dd90a4c 5baae82 dd90a4c 5baae82 dd90a4c 5baae82 ea3a71e 5baae82 ea3a71e 5baae82 ea3a71e 5baae82 ea3a71e dd90a4c ea3a71e 5baae82 ea3a71e 5baae82 ea3a71e dd90a4c | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 | {
"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"
} |