File size: 1,681 Bytes
795b0d8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
{
  "runtime": "Kapitsa 0.2.31 (84249c0)",
  "host": "Apple M4 Max, 40-core GPU, 64 GiB",
  "context_limit": 131072,
  "sampler": "greedy; thinking disabled",
  "implementation": "request-level strict JSON Schema via response_format; no weight or tokenizer changes; no response postprocessing",
  "schema_inputs": "Field types and structural constraints only; expected answers are excluded from requests.",
  "test_source_sha256": "b0e5013aecfe29e2b966afed82120926b700ffd8860a8129ee08e5cdecb62088",
  "schema_file": "structured_output_profiles.json",
  "fixture_seeds": {
    "original": [
      7319,
      173
    ],
    "held_out": [
      27183,
      65537,
      42017
    ]
  },
  "vision": [
    {
      "mode": "dflash2",
      "lane": "baseline",
      "passed": 4,
      "total": 8
    },
    {
      "mode": "dflash2",
      "lane": "schema-r1",
      "passed": 8,
      "total": 8
    },
    {
      "mode": "dflash2",
      "lane": "schema-r2",
      "passed": 8,
      "total": 8
    },
    {
      "mode": "dflash2",
      "lane": "holdout",
      "passed": 12,
      "total": 12
    },
    {
      "mode": "mtp3",
      "lane": "schema-mtp",
      "passed": 8,
      "total": 8
    }
  ],
  "coding_passed": 6,
  "coding_total": 8,
  "coding_grader_note": "The same standard-library typing allowance is applied to both models.",
  "speed": {
    "text_median_decode_tps": 90.2,
    "structured_json_median_decode_tps": 128.1,
    "text_probes": 2,
    "repeats": 3,
    "measurement": "Median server-reported decode; code/prose capped at 256 tokens; JSON sequence independently checked for exact correctness. Six text and three JSON measurements per model."
  }
}