Hugging Face's logo

Spaces:
bertfil
/
DfhMMHXDuu
Running

App Files Community
DfhMMHXDuu
File size: 3,253 Bytes
5504be5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
73
74
75
76
77
78
79
80
{
  "schema_version": 1,
  "title": "Repro - Mitigating Reward Hacking in RLHF via Bayesian Non-negative Reward Modeling",
  "emoji": "\ud83d\udd2c",
  "space_id": "bertfil/DfhMMHXDuu",
  "paper": {
    "arxiv_id": "2602.10623",
    "openreview_id": "DfhMMHXDuu",
    "title": "Mitigating Reward Hacking in RLHF via Bayesian Non-negative Reward Modeling",
    "url": "https://arxiv.org/abs/2602.10623"
  },
  "tags": [
    "icml2026-repro",
    "paper-DfhMMHXDuu"
  ],
  "updated_at": "2026-07-20T03:52:36+00:00",
  "root": {
    "slug": "index",
    "title": "Repro - Mitigating Reward Hacking in RLHF via Bayesian Non-negative Reward Modeling",
    "file": "pages/index.md",
    "children": [
      {
        "slug": "00-scored-evidence-summary",
        "title": "00 - Scored evidence summary",
        "file": "pages/00-scored-evidence-summary/page.md",
        "children": []
      },
      {
        "slug": "claim-1-bnrm-replaces-deterministic-scalar-reward-outputs",
        "title": "Claim 1 - BNRM replaces deterministic scalar reward outputs with spars\u2026",
        "file": "pages/claim-1-bnrm-replaces-deterministic-scalar-reward-outputs/page.md",
        "children": []
      },
      {
        "slug": "claim-2-bnrm-extends-the-bradley-terry-preference-model-wi",
        "title": "Claim 2 - BNRM extends the Bradley-Terry preference model with Gamma p\u2026",
        "file": "pages/claim-2-bnrm-extends-the-bradley-terry-preference-model-wi/page.md",
        "children": []
      },
      {
        "slug": "claim-3-with-40k-training-examples-bt-bnrm-improves-over-t",
        "title": "Claim 3 - With 40K training examples, BT-BNRM improves over the baseli\u2026",
        "file": "pages/claim-3-with-40k-training-examples-bt-bnrm-improves-over-t/page.md",
        "children": []
      },
      {
        "slug": "claim-4-bnrm-matches-the-performance-of-a-baseline-trained",
        "title": "Claim 4 - BNRM matches the performance of a baseline trained on 20K sa\u2026",
        "file": "pages/claim-4-bnrm-matches-the-performance-of-a-baseline-trained/page.md",
        "children": []
      },
      {
        "slug": "claim-5-bnrm-reduces-the-pearson-correlation-between-rewar",
        "title": "Claim 5 - BNRM reduces the Pearson correlation between reward and resp\u2026",
        "file": "pages/claim-5-bnrm-reduces-the-pearson-correlation-between-rewar/page.md",
        "children": []
      },
      {
        "slug": "claim-6-in-rlhf-policy-evaluation-policies-trained-with-bn",
        "title": "Claim 6 - In RLHF policy evaluation, policies trained with BNRM-derive\u2026",
        "file": "pages/claim-6-in-rlhf-policy-evaluation-policies-trained-with-bn/page.md",
        "children": []
      },
      {
        "slug": "reproduction-protocol-and-provenance",
        "title": "Reproduction protocol and provenance",
        "file": "pages/reproduction-protocol-and-provenance/page.md",
        "children": []
      },
      {
        "slug": "conclusion",
        "title": "Conclusion",
        "file": "pages/conclusion/page.md",
        "children": []
      }
    ]
  },
  "agent_view_tokens": 5888,
  "revision": "1784519556257555900"
}