Hugging Face's logo

Spaces:
vimarsh
/
repro-float8-2bits-entropy-coding-enables-data-free-model-compression
Running

App Files Community
repro-float8-2bits-entropy-coding-enables-data-free-model-compression
File size: 3,122 Bytes
43eff34
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
{
  "schema_version": 1,
  "title": "Repro - Float8@2bits: Entropy Coding Enables Data-Free Model Compression",
  "emoji": "🎯",
  "space_id": "vimarsh/repro-float8-2bits-entropy-coding-enables-data-free-model-compression",
  "paper": {
    "openreview_id": "15SdFmonYz"
  },
  "tags": [
    "icml2026-repro",
    "paper-15SdFmonYz"
  ],
  "updated_at": "2026-07-17T06:08:09+00:00",
  "root": {
    "slug": "index",
    "title": "Repro - Float8@2bits: Entropy Coding Enables Data-Free Model Compression",
    "file": "pages/index.md",
    "children": [
      {
        "slug": "claim-1-compression-time-and-gpu-ans",
        "title": "Claim 1: compression time and GPU ANS",
        "file": "pages/claim-1-compression-time-and-gpu-ans/page.md",
        "children": []
      },
      {
        "slug": "claim-2-llama-2-7b-c4-perplexity",
        "title": "Claim 2: LLaMA-2 7B C4 perplexity",
        "file": "pages/claim-2-llama-2-7b-c4-perplexity/page.md",
        "children": []
      },
      {
        "slug": "claim-3-calibrated-method-comparison",
        "title": "Claim 3: calibrated-method comparison",
        "file": "pages/claim-3-calibrated-method-comparison/page.md",
        "children": []
      },
      {
        "slug": "claim-4-representational-expressivity",
        "title": "Claim 4: representational expressivity",
        "file": "pages/claim-4-representational-expressivity/page.md",
        "children": []
      },
      {
        "slug": "claim-5-inference-throughput",
        "title": "Claim 5: inference throughput",
        "file": "pages/claim-5-inference-throughput/page.md",
        "children": []
      },
      {
        "slug": "claim-6-instruction-tuned-benchmarks",
        "title": "Claim 6: instruction-tuned benchmarks",
        "file": "pages/claim-6-instruction-tuned-benchmarks/page.md",
        "children": []
      },
      {
        "slug": "conclusion",
        "title": "Conclusion",
        "file": "pages/conclusion/page.md",
        "children": []
      },
      {
        "slug": "float8-2bits-repro",
        "title": "float8-2bits-repro",
        "file": "pages/float8-2bits-repro/page.md",
        "children": []
      },
      {
        "slug": "claim-1-llpf-m2m-low-loss-connectivity",
        "title": "Claim 1: LLPF_M2M low-loss connectivity",
        "file": "pages/claim-1-llpf-m2m-low-loss-connectivity/page.md",
        "children": []
      },
      {
        "slug": "claim-2-layer-wise-distance-consistency",
        "title": "Claim 2: Layer-wise distance consistency",
        "file": "pages/claim-2-layer-wise-distance-consistency/page.md",
        "children": []
      },
      {
        "slug": "claim-3-llpf-versus-autoneb",
        "title": "Claim 3: LLPF versus AutoNEB",
        "file": "pages/claim-3-llpf-versus-autoneb/page.md",
        "children": []
      },
      {
        "slug": "claim-4-llpf-m2o-toward-origin",
        "title": "Claim 4: LLPF_M2O toward origin",
        "file": "pages/claim-4-llpf-m2o-toward-origin/page.md",
        "children": []
      }
    ]
  },
  "agent_view_tokens": 3286,
  "revision": "1784268489639454773"
}