02 · Anatomy

One file. Seven layers.

A .kolm is a deterministic zip with a fixed schema (RS-1). Every layer is reproducible from (task, examples, base_model, recipe_registry@version). Two compiles with the same inputs produce byte-identical artifacts.

01 · What’s inside

The seven parts.

Listed in the order kolm writes them. Sizes are an example from a typical task; small artifacts run ~30 MB, large ones cap around 3 GB.

  • 01

    model.gguf

    The base model, INT4-quantized via llama.cpp. Pinned by hash to a known open-weight release. The student.

    2.1 GB
  • 02

    lora.bin

    The personal adapter. Distilled from k-sample frontier verified inference on your task. The 90% lift in 100 MB.

    82 MB
  • 03

    recipes.json

    The recipe pack — deterministic-token drafts indexed by prefix-shape and embedding-shape. Drives speculative decoding for the structured-output subset.

    7.2 MB
  • 04

    index.sqlite-vec

    Multimodal embedding index over your corpus. Queryable offline. Grounds every inference in your own data.

    11 MB
  • 05

    tests.jsonl

    The eval suite the K-score is computed against. Travels with the artifact. kolm verify re-runs them locally; no eval, no ship.

    1.4 MB
  • 06

    manifest.json

    Task description, base hash, training stats, registry version, K-score, gate, recall model hashes. The contract for what this file is.

    12 KB
  • 07

    signature.sig

    Hmac chain over every preceding layer. Anchored to the public Kolmogorov registry. Tampering breaks kolm verify.

    256 B
// manifest.json — the contract
{
  "spec": "rs-1-manifest",
  "task": "triage support email,
            output {label,severity}",
  "base_model": {
    "id": "qwen2.5-coder-7b-instruct-q4_0",
    "hash": "sha256:9f1a…b3"
  },
  "k_score": 0.94,
  "k_components": {
    "accuracy":  0.974,
    "size":      0.95,
    "latency":   0.94,
    "cost":      1.00,
    "coverage":  0.92
  },
  "gate": 0.70,
  "training": {
    "distilled_pairs": 3120,
    "verifier_accepted": true,
    "pass_rate_positive": 0.98,
    "latency_p50_us": 50
  },
  "recipe_registry": "rs-1@2026-05-06",
  "runtime": "kolm/0.4.0",
  "created_at": "2026-05-07T03:14:00Z"
}
02 · Three invariants

Three rules every .kolm respects.

Reproducible

Same (task, examples, base, registry@v)  →  byte-identical zip. No hidden randomness; sampling seeds are pinned in the manifest.

Self-evaluating

The artifact carries the tests it was scored against. kolm verify recomputes K offline. The number on the box matches the number you measure.

Tamper-evident

Hmac chain over all seven layers. One byte changes anywhere and kolm verify refuses to load the artifact. No silent corruption.

03 · Inspect

Open the box.

Every artifact is a normal zip. kolm inspect formats the contents for humans; unzip -l works fine if you don’t have kolm installed.

$ kolm inspect support-triage.kolm

  layer            size    sha256
  ───────────────  ──────  ─────────────────
  model.gguf       2.1 GB  9f1a8c…b3
  lora.bin         82  MB  4d22ef…c1
  recipes.json     7.2 MB  ab02f1…7e
  index.sqlite-vec 11  MB  8b73e9…02
  tests.jsonl      1.4 MB  ef44a3…c8
  manifest.json    12 KB   1a92c4…aa
  signature.sig    256 B   verified

  total            2.21 GB
  k_score          0.94   gate passed
  signature        ok     anchored to rs-1@2026-05-06