File size: 3,891 Bytes
377660b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
bc9139e
 
 
 
 
 
 
 
 
 
 
 
 
 
377660b
 
 
 
bc9139e
377660b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
bc9139e
 
 
 
 
 
377660b
 
 
254d03a
bc9139e
377660b
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
96
97
98
99
100
101
102
103
104
105
106
107
{
  "table": "Table 2",
  "metric": "Avg@8 (%)",
  "average": "Equal weight over ten benchmarks",
  "checkpoint_selection": {
    "F0": {
      "checkpoint_step": 175,
      "weights": "Actor"
    },
    "M1": {
      "checkpoint_step": 100,
      "weights": "Actor",
      "reason": "Single checkpoint with all fourteen benchmarks; also the highest recorded CLEAN2K score."
    }
  },
  "rows": {
    "RAVEC-3B": {
      "source_scores_percent": {
        "Geometry3K": "33.28",
        "MathVista": "59.78",
        "We-Math": "61.68",
        "MathVerse": "57.43",
        "LogicVista": "41.67",
        "MMMU-Pro": "28.38",
        "MathVerse-V": "54.44",
        "C-VQA-Real": "66.5414281170483500",
        "MARS-Bench": "53.0454990215264200",
        "POPE": "87.9833333333333400"
      },
      "benchmark_count": 10,
      "missing_benchmarks": [],
      "mean_percent": "54.4230260471908110",
      "display_mean_percent": "54.42"
    },
    "RAVEC-7B": {
      "source_scores_percent": {
        "Geometry3K": "43.20",
        "MathVista": "66.44",
        "We-Math": "70.73",
        "MathVerse": "67.53",
        "LogicVista": "46.28",
        "MMMU-Pro": "36.09",
        "MathVerse-V": "65.03",
        "C-VQA-Real": "70.69",
        "MARS-Bench": "55.99",
        "POPE": "87.99"
      },
      "benchmark_count": 10,
      "missing_benchmarks": [],
      "mean_percent": "60.997",
      "display_mean_percent": "61.00",
      "raw_scores_percent": {
        "geo3k": 43.19883527454243,
        "mathvista": 66.4375,
        "wemath": 70.72557471264368,
        "mathverse": 67.53440366972477,
        "logicvista": 46.28076062639821,
        "mmmu-pro": 36.09104046242775,
        "mathverse-v": 65.03249235474006,
        "c-vqa-real": 70.68821564885496,
        "mars-bench": 55.990704500978474,
        "pope": 87.99166666666667
      },
      "raw_mean_percent": 60.997119391697694
    }
  },
  "availability": {
    "RAVEC-3B": "included",
    "RAVEC-7B": "included; M1 step100 Actor, verified against evaluation-time SHA-256 hashes"
  },
  "provenance": [
    {
      "source_in_project": "paper/tables/main_results.tex",
      "sha256": "6f7c33a0076b6e649c3bc44a4ee600731d3a8f722be5aacf23f567ded0745361"
    },
    {
      "source_in_project": "paper/scripts/build_experiment_tables.py",
      "sha256": "fd2a5c3e3d16d9a2501351729b5d5f6099174bc921f82b15094de83a3627fc95"
    },
    {
      "source_in_project": "paper/data/table_statistics_20261107.json",
      "sha256": "5e42aba7c5ebf532153f9a14e45665c589cc611e6a1f9151c64a3ef4a6d2f80d"
    },
    {
      "source_in_project": "paper/data/ravec3b_extra_scores_20261108.json",
      "sha256": "0fece581e263c3bbc7fd0054cb7938af45b4088ac7234abb332e05b72bede3a8"
    },
    {
      "source_in_project": "checkpoints/ravec_main/step175/manifest.json",
      "sha256": "e95a6a5de5794b5f10f1cbad73f4dd6ed22308dd4a397abf0deda4cce64374e2"
    },
    {
      "source_in_project": "RAVEC_GRPO_7B_RESULTS_20261106.md",
      "sha256": "4ecf81e9dfe50c71c16fc51a0c92134807bc41234ad43cec8f471908404b6a35"
    },
    {
      "source_in_project": "runtime/assets/runs/formal_7b_4gpu_20261029_000339/M1/eval_step100_suite14/238fbe483d8179bb/summary.json",
      "sha256": "87f95c152d4ebc6d193018d7e1648677d2f375012fd35e00c7e6144fcf797329",
      "model_fingerprint": "238fbe483d8179bbbab0ce9f4ec044c09bb43661f4f319698f063beb2f6199d7",
      "release_manifest": "RAVEC-7B/checkpoint_manifest.json"
    }
  ],
  "precision_note": "3B original seven displayed tasks and 7B tasks use reported precision; 3B CVQA/MARS/POPE use raw scoring precision.",
  "weight_label_note": "Historical 3B records use EMA_G3R / EMA labels; current paper and table-generation metadata identify this exact step175 export as Actor/student. File hashes identify the released artifact unambiguously.",
  "release_scope": "3B and 7B main-result checkpoints"
}