turboquant-skill/codemeta.json at main · Ryuketsukami/turboquant-skill · GitHub

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
{
  "@context": "https://w3id.org/codemeta/3.0",
  "@type": "SoftwareSourceCode",
  "name": "TurboQuant Skill",
  "description": "AI agent skill implementing Google's TurboQuant algorithm (ICLR 2026) for LLM KV cache compression — 6x memory reduction, 8x speedup, zero accuracy loss. Compatible with Claude Code, Codex CLI, and all Agent Skills-compatible tools.",
  "codeRepository": "https://github.com/Ryuketsukami/turboquant-skill",
  "issueTracker": "https://github.com/Ryuketsukami/turboquant-skill/issues",
  "license": "https://spdx.org/licenses/MIT",
  "version": "0.1.0",
  "dateCreated": "2026-03-28",
  "dateModified": "2026-03-28",
  "programmingLanguage": ["Python"],
  "runtimePlatform": "Python 3.10+",
  "author": [
    {
      "@type": "Person",
      "name": "Ryuketsukami",
      "url": "https://github.com/Ryuketsukami"
    }
  ],
  "keywords": [
    "turboquant",
    "agent-skill",
    "kv-cache",
    "quantization",
    "llm-compression",
    "claude-code",
    "codex",
    "ai-agent",
    "vector-quantization",
    "inference-optimization"
  ],
  "softwareRequirements": [
    "numpy>=1.24",
    "scipy>=1.10"
  ],
  "referencePublication": {
    "@type": "ScholarlyArticle",
    "name": "TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate",
    "author": [
      {"@type": "Person", "name": "Amir Zandieh"},
      {"@type": "Person", "name": "Majid Daliri"},
      {"@type": "Person", "name": "Majid Hadian"},
      {"@type": "Person", "name": "Vahab Mirrokni"}
    ],
    "url": "https://arxiv.org/abs/2504.19874",
    "datePublished": "2026"
  }
}