feat: checkpoint distributed gguf runtime stories

This commit is contained in:
Dobromir Popov
2026-07-15 23:42:58 +03:00
parent eaf00f6add
commit 1fe31ef38d
60 changed files with 8478 additions and 105 deletions

View File

@@ -5,6 +5,7 @@ Model ids here are arbitrary and made up on purpose: nothing in the admission or
routing path may branch on a vendor, model or kernel name.
"""
import hashlib
import json
import time
import urllib.error
@@ -18,6 +19,7 @@ from meshnet_tracker.capability import (
POLICY_ENFORCE,
STATE_ABSENT,
STATE_ADMITTED,
STATE_COMPATIBILITY_MISMATCH,
STATE_CATALOGUE_INCOMPATIBLE,
STATE_FAILED,
STATE_INVALID,
@@ -41,6 +43,14 @@ SHORT = "oracle-9b"
LAYERS = 32
def _stable_json(data: dict) -> str:
return json.dumps(data, sort_keys=True, separators=(",", ":"), ensure_ascii=False)
def _sha256_text(data: dict) -> str:
return "sha256:" + hashlib.sha256(_stable_json(data).encode("utf-8")).hexdigest()
def _post_json(url: str, payload: dict) -> dict:
data = json.dumps(payload).encode()
req = urllib.request.Request(
@@ -60,6 +70,8 @@ def _report(
model_id: str = MODEL,
start: int = 0,
end: int = 15,
owns_embedding: bool | None = None,
owns_final_head: bool | None = None,
status: str = "passed",
validated_at: float | None = None,
recipe_id: str = "baseline",
@@ -70,10 +82,48 @@ def _report(
diagnostics: list | None = None,
) -> dict:
"""A capability report shaped exactly as `meshnet_node.capability` emits it."""
return {
if owns_embedding is None:
owns_embedding = start == 0
if owns_final_head is None:
owns_final_head = end >= LAYERS - 1
artifact = {
"model_id": model_id,
"revision": None,
"artifact_hash": _sha256_text(
{
"model_id": model_id,
"shard_start": start,
"shard_end": end,
"recipe_id": recipe_id,
"recipe_version": recipe_version,
}
),
"shard_start": start,
"shard_end": end,
}
runtime_recipe = {
"weight_quantization": "bfloat16",
"activation_dtype": "bfloat16",
"compute_dtype": "bfloat16",
"kv_dtype": "bfloat16",
"kv_layout": "session-cache",
"tokenizer_revision": model_id,
"architecture_adapter": "unknown",
"backend_id": "torch-transformers",
"runtime_version": "0.1.0",
"boundary_schema_version": 1,
"cache_layout": "local-hot-kv",
}
runtime_recipe["fingerprint"] = _sha256_text(runtime_recipe)
payload = {
"schema_version": schema_version,
"model": {"model_id": model_id, "revision": None, "config_fingerprint": None},
"shard": {"start": start, "end": end},
"shard": {
"start": start,
"end": end,
"owns_embedding": owns_embedding,
"owns_final_head": owns_final_head,
},
"recipe": {
"recipe_id": recipe_id,
"recipe_version": recipe_version,
@@ -86,11 +136,24 @@ def _report(
"quantization": "bfloat16",
"runtime": {},
},
"artifact": artifact,
"runtime_recipe": runtime_recipe,
"status": status,
"validated_at": time.time() if validated_at is None else validated_at,
"duration_ms": 42,
"diagnostics": list(diagnostics or []),
}
payload["compatibility_fingerprint"] = _sha256_text(
{
"model": payload["model"],
"shard": payload["shard"],
"recipe": payload["recipe"],
"backend": payload["backend"],
"artifact": payload["artifact"],
"runtime_recipe": payload["runtime_recipe"],
}
)
return payload
def _registration(
@@ -119,6 +182,7 @@ def _registration(
report = _report(start=start, end=end)
if report is not None:
payload["capability_report"] = report
payload["compatibility_fingerprint"] = report["compatibility_fingerprint"]
if recipe_id is not None:
payload["recipe_id"] = recipe_id
if recipe_version is not None:
@@ -196,6 +260,15 @@ def test_a_report_for_a_different_recipe_than_the_node_declares_is_a_recipe_mism
assert versioned.state == STATE_RECIPE_MISMATCH
def test_a_report_for_a_different_compatibility_fingerprint_is_a_compatibility_mismatch():
"The exact artifact/runtime recipe fingerprint gates admission.\n\nTags: routing, tracker"
state = _evaluate(
_report(),
declared_compatibility_fingerprint="sha256:deadbeef",
)
assert state.state == STATE_COMPATIBILITY_MISMATCH
def test_an_older_recipe_catalogue_is_incompatible():
"Recipe ids from a catalogue older than the tracker's minimum cannot be matched.\n\nTags: routing, tracker"
state = _evaluate(_report(catalogue_version="2025.01.1"))