Skip to content

Commit c41e5ae

Browse files
prekshivyasgithub-actions[bot]senthilr-nv
authored
fix(inference): update Muse Glimmer vLLM image for revision handling (#9675)
## Summary Replace the DGX Spark Muse Glimmer vLLM ARM64 digest with an immutable official image that preserves the requested Hugging Face model revision across vLLM engine spawning. The previous image fell back to `main` and surfaced a misleading sentencepiece/tiktoken tokenizer error even though both packages were installed. ## Related Issue Fixes #9601. ## Changes - Pin `vllm/vllm-openai@sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965`, built from vLLM commit `5a4c8d99242e9e069b604d0e9b969e77f7dd501d`. - Record the exact image manifest, configuration, source ancestry, runtime dependency versions, and revision-serialization evidence. - Protect the image, `huggingface_hub 1.28.0`, and post-pickle revision with catalog and provenance tests. - Update the vLLM setup documentation with the qualified digest and dependency fix. ## Type of Change - [ ] Code change (feature, bug fix, or refactor) - [x] Code change with doc updates - [ ] Doc only (prose changes, no code sample modifications) - [ ] Doc only (includes code sample changes) ## Quality Gates - [x] Tests added or updated for changed behavior - [ ] Existing tests cover changed behavior — justification: - [ ] Tests not applicable — justification: - [x] Sensitive paths changed (security, policy, credentials, preflight, onboarding, inference, runner, sandbox, or messaging) - [ ] Sensitive-path review completed or maintainer-approved waiver recorded — reviewer/approval link/justification: Pending maintainer review; the exact image provenance and DGX Spark qualification are recorded in this change. - [ ] Non-success, skipped, or missing CI check accepted by maintainer — check name, approval link, and follow-up issue: ## DGX Station Hardware Evidence - [ ] Tested on DGX Station - Tested commit: Not applicable - Station profile/scenario: Not applicable; this changes the DGX Spark Muse Glimmer recipe. - Result: Not applicable - Supporting evidence: Not applicable ## DGX Spark Hardware Evidence Validated on a physical ARM64 DGX Spark with an NVIDIA GB10 using model `Inferact/Muse-Glimmer-30B-NVFP4-W4A4` at revision `d35cb79050f419c457611b1cee5c5d15b176f285`. - reproduced the old image failure with `vllm/vllm-openai@sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204` - imported sentencepiece `0.2.2`, tiktoken `0.14.0`, and `huggingface_hub 1.28.0` from the replacement image - verified that the resolved revision survives a pickle round trip - started the full recipe-equivalent vLLM server from cold cache without a usable `refs/main` - verified authenticated `/v1/models`, `/tokenize`, chat reasoning, and structured Muse Glimmer tool calling - observed zero vLLM container restarts ## Verification - [x] PR description includes a `Signed-off-by:` line and every commit appears as `Verified` in GitHub - [x] Normal `pre-commit`, `commit-msg`, and `pre-push` hooks passed, or `npm run validate:pr` passed after refreshing `origin/main` when hooks were skipped or unavailable - [x] Targeted behavior tests pass for the current change set, or tests are marked not applicable above — 115 focused CLI tests, 18 provenance/compiler integration tests, and 32 growth guardrails passed. - [x] Applicable broad gate passed — Not applicable to this image-pin/provenance-only change. `npm run test:changed` selected six unrelated suites with 18 failures that reproduce identically at base commit `dbf48bae9d35beda8d781205a46e881d6f8f900a`. - [ ] Quality Gates section completed with required justifications or waivers — sensitive-path maintainer review remains pending. - [x] No secrets, API keys, or credentials committed - [ ] `npm run docs` builds without warnings (doc changes only) — `npm run docs` passed with the two existing Fern warnings. - [x] Doc pages follow the [style guide](https://github.com/NVIDIA/NemoClaw/blob/main/docs/CONTRIBUTING.md) (doc changes only) - [ ] New doc pages include SPDX header and frontmatter (new pages only) --- Signed-off-by: Prekshi Vyas <prekshiv@nvidia.com> <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Documentation** * Updated DGX Spark Muse Glimmer managed-vLLM documentation with the latest image, size, build, and dependency details. * **Bug Fixes** * Updated vLLM runtime image references and download metadata. * Improved runtime resolution, architecture handling, shared-memory configuration, local-image validation, and Lightning recipe support. * Preserved interactive runtime options. * **Tests** * Refreshed expected image metadata. * Added security coverage validating image provenance, runtime consistency, dependencies, and revision changes. <!-- end of auto-generated comment: release notes by coderabbit.ai --> --------- Signed-off-by: Senthil Ravichandran <senthilr@nvidia.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Senthil Ravichandran <senthilr@nvidia.com>
1 parent 55e772d commit c41e5ae

8 files changed

Lines changed: 113 additions & 50 deletions

‎ci/source-shape-test-budget.json‎

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -81,6 +81,11 @@
8181
"test": "restores exact locked posture after root-separated repair and later failure (#7033)",
8282
"category": "security"
8383
},
84+
{
85+
"file": "test/muse-glimmer-vllm-image-provenance.test.ts",
86+
"test": "binds the checked-in provenance to the selected runtime",
87+
"category": "security"
88+
},
8489
{
8590
"file": "test/muse-glimmer-vllm-image-provenance.test.ts",
8691
"test": "rejects %s",

‎docs/inference/set-up-vllm.mdx‎

Lines changed: 3 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -158,8 +158,9 @@ Managed profiles and model-specific recipes use immutable image digests:
158158

159159
- DGX Spark and DGX Station models without a model-specific runtime use the `linux/arm64` digest `sha256:9204569b17ee4c0eff75194b8e6e458479c8aee18953b5ab9cf359fcdac659e2` with a compressed layer size of `9.60 GB` under `nvcr.io/nvidia/vllm:26.05.post1-py3`.
160160
N1x uses the same image.
161-
- The DGX Spark Muse Glimmer recipe uses the `linux/arm64` post-merge vLLM nightly digest `sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204` with a compressed layer size of `9.70 GB` under `vllm/vllm-openai`.
162-
The image was built from vLLM commit `ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9`, which contains merged Muse Glimmer support commit `6adad08767583f52eb4d2122111af0bf638ed5e6`.
161+
- The DGX Spark Muse Glimmer recipe uses the `linux/arm64` post-fix vLLM nightly digest `sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965` with a compressed layer size of `9.71 GB` under `vllm/vllm-openai`.
162+
The image was built from vLLM commit `5a4c8d99242e9e069b604d0e9b969e77f7dd501d`, which contains merged Muse Glimmer support commit `6adad08767583f52eb4d2122111af0bf638ed5e6` and the vLLM dependency fix from commit `90984ddbed27a09409506d6d6c0eea87f54b04b5`.
163+
The image includes `huggingface_hub` `1.28.0`, which preserves the pinned model revision when vLLM transfers configuration to the engine process.
163164
It pins Hugging Face revision `d35cb79050f419c457611b1cee5c5d15b176f285` for the approximately `25.45 GB` model download.
164165
- The DGX Spark Nemotron 3.5 Lightning recipe uses the `linux/arm64` digest `sha256:3af90144a0926e5c5fe46ee16e5201e763dd854538b9d7ce433755f11dadaf78` with a compressed layer size of approximately `12.69 GB` under `vllm/vllm-openai`.
165166
It pins Hugging Face revision `0dcd680e5585c791728c83342b311d0a0026dbeb` for the approximately `21.56 GB` model download.

‎internal/security-reviews/muse-glimmer-vllm-image-provenance-v1.json‎

Lines changed: 38 additions & 25 deletions
Original file line numberDiff line numberDiff line change
@@ -11,51 +11,64 @@
1111
"repository": "vllm/vllm-openai"
1212
},
1313
"image": {
14-
"reference": "vllm/vllm-openai@sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204",
15-
"manifestDigest": "sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204",
14+
"reference": "vllm/vllm-openai@sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965",
15+
"manifestDigest": "sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965",
1616
"manifestMediaType": "application/vnd.docker.distribution.manifest.v2+json",
17-
"manifestUrl": "https://registry-1.docker.io/v2/vllm/vllm-openai/manifests/sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204",
18-
"configDigest": "sha256:c3f199e54a26d2d7a9a41115cd07ce9d90a6488c5a4e75b17129e1006ce533fd",
17+
"manifestUrl": "https://registry-1.docker.io/v2/vllm/vllm-openai/manifests/sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965",
18+
"configDigest": "sha256:49d2eb65dc2a8dea24e43c27b226f650481ac97d4ba9c567b6e1ca08bc472303",
1919
"configMediaType": "application/vnd.docker.container.image.v1+json",
20-
"configSizeBytes": 34762,
21-
"configUrl": "https://registry-1.docker.io/v2/vllm/vllm-openai/blobs/sha256:c3f199e54a26d2d7a9a41115cd07ce9d90a6488c5a4e75b17129e1006ce533fd",
20+
"configSizeBytes": 34738,
21+
"configUrl": "https://registry-1.docker.io/v2/vllm/vllm-openai/blobs/sha256:49d2eb65dc2a8dea24e43c27b226f650481ac97d4ba9c567b6e1ca08bc472303",
2222
"layerCount": 32,
23-
"compressedLayerSizeBytes": 9699710136,
24-
"createdAt": "2026-08-14T05:33:50.528328374Z",
23+
"compressedLayerSizeBytes": 9706339423,
24+
"createdAt": "2026-08-19T05:34:00.562836767Z",
2525
"platform": {
2626
"os": "linux",
2727
"architecture": "arm64"
2828
}
2929
},
3030
"build": {
31-
"sourceRepository": "https://github.com/vllm-project/vllm",
32-
"sourceRevision": "ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
33-
"sourceRevisionUrl": "https://github.com/vllm-project/vllm/commit/ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
34-
"imageTag": "vllm/vllm-openai:nightly-ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
35-
"pipelineId": "019d130e-464e-4ff7-b84b-492992c0c06b",
36-
"pipelineUrl": "https://buildkite.com/vllm/release-v2/builds/5174"
31+
"sourceRevision": "5a4c8d99242e9e069b604d0e9b969e77f7dd501d",
32+
"imageTag": "vllm/vllm-openai:nightly-5a4c8d99242e9e069b604d0e9b969e77f7dd501d",
33+
"pipelineId": "019d130e-464e-4ff7-b84b-492992c0c06b"
3734
},
3835
"upstreamSupport": {
3936
"museMergeCommit": "6adad08767583f52eb4d2122111af0bf638ed5e6",
40-
"museMergeCommitUrl": "https://github.com/vllm-project/vllm/commit/6adad08767583f52eb4d2122111af0bf638ed5e6",
41-
"comparisonUrl": "https://github.com/vllm-project/vllm/compare/6adad08767583f52eb4d2122111af0bf638ed5e6...ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
42-
"relationship": "direct-descendant",
43-
"aheadBy": 1
37+
"relationship": "descendant",
38+
"aheadBy": 173,
39+
"revisionFixCommit": "90984ddbed27a09409506d6d6c0eea87f54b04b5",
40+
"revisionFixRelationship": "descendant",
41+
"revisionFixAheadBy": 33
4442
},
4543
"reportedLabels": {
46-
"ai.vllm.build.commit": "ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
44+
"ai.vllm.build.commit": "5a4c8d99242e9e069b604d0e9b969e77f7dd501d",
4745
"ai.vllm.build.pipeline": "019d130e-464e-4ff7-b84b-492992c0c06b",
48-
"ai.vllm.build.url": "https://buildkite.com/vllm/release-v2/builds/5174",
49-
"ai.vllm.image.tag": "vllm/vllm-openai:nightly-ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
50-
"org.opencontainers.image.revision": "ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
51-
"org.opencontainers.image.source": "https://github.com/vllm-project/vllm"
46+
"ai.vllm.image.tag": "vllm/vllm-openai:nightly-5a4c8d99242e9e069b604d0e9b969e77f7dd501d",
47+
"org.opencontainers.image.revision": "5a4c8d99242e9e069b604d0e9b969e77f7dd501d"
48+
},
49+
"runtimeDependencies": {
50+
"vllmVersion": "0.26.1rc1.dev942+g5a4c8d992",
51+
"huggingfaceHubVersion": "1.28.0",
52+
"sentencepieceVersion": "0.2.2",
53+
"tiktokenVersion": "0.14.0"
54+
},
55+
"revisionSerialization": {
56+
"model": "Inferact/Muse-Glimmer-30B-NVFP4-W4A4",
57+
"requestedRevision": "d35cb79050f419c457611b1cee5c5d15b176f285",
58+
"resolvedRevisionBeforePickle": "d35cb79050f419c457611b1cee5c5d15b176f285",
59+
"resolvedRevisionAfterPickle": "d35cb79050f419c457611b1cee5c5d15b176f285",
60+
"preserved": true
5261
},
5362
"verification": {
54-
"observedAt": "2026-08-14T16:33:50Z",
63+
"observedAt": "2026-08-19T22:53:46Z",
5564
"methods": [
5665
"docker-buildx-imagetools-inspect-raw",
5766
"docker-image-inspect",
58-
"github.qkg1.toppare-api"
67+
"github.qkg1.toppare-api",
68+
"container-package-imports",
69+
"resolved-revision-pickle-round-trip",
70+
"dgx-spark-cold-cache-startup",
71+
"vllm-openai-api-validation"
5972
],
6073
"signedProvenanceAttestation": "not-available"
6174
}

‎managed-inference/recipes/vllm.muse-glimmer-30b-nvfp4-w4a4.spark-single.v1.yaml‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -13,8 +13,8 @@ spec:
1313

1414
modelRef: vllm.muse-glimmer-30b-nvfp4-w4a4.v1
1515
runtime:
16-
image: vllm/vllm-openai@sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204
17-
imageDownloadSizeBytes: 9699710136
16+
image: vllm/vllm-openai@sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965
17+
imageDownloadSizeBytes: 9706339423
1818
minimumComputeCapability: 121
1919
minimumGpuMemoryBytes: 25447097878
2020
pullTimeoutSeconds: 3600

‎src/lib/inference/vllm.test.ts‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -300,9 +300,9 @@ describe("vLLM profile detection", () => {
300300
const resolved = resolveVllmModelRuntime(profile!, muse!);
301301
const runtime = resolved.profile;
302302
expect(runtime.image).toBe(
303-
"vllm/vllm-openai@sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204",
303+
"vllm/vllm-openai@sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965",
304304
);
305-
expect(runtime.imageDownloadSizeBytes).toBe(9_699_710_136);
305+
expect(runtime.imageDownloadSizeBytes).toBe(9_706_339_423);
306306
expect(runtime.modelDownloadSizeBytes).toBe(25_447_097_878);
307307

308308
const apiKey = "a".repeat(64);

‎test/managed-inference-catalog-compiler.test.ts‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -295,7 +295,7 @@ describe("managed inference YAML profile contract", () => {
295295
runtime: {
296296
architecture: "arm64",
297297
image:
298-
"vllm/vllm-openai@sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204",
298+
"vllm/vllm-openai@sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965",
299299
},
300300
execution: {
301301
materializerRef: "vllm.host-local/v1",

‎test/muse-glimmer-vllm-image-provenance.test.ts‎

Lines changed: 26 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -35,6 +35,26 @@ const recipe = YAML.parse(readFileSync(RECIPE_PATH, "utf8")) as {
3535
};
3636

3737
describe("Muse Glimmer vLLM image provenance", () => {
38+
// source-shape-contract: security -- The checked-in provenance and recipe must resolve to the same reviewed external runtime image and size.
39+
it("binds the checked-in provenance to the selected runtime", () => {
40+
verifyMuseGlimmerVllmImageProvenance(provenance);
41+
42+
expect(recipe.spec.runtime).toMatchObject({
43+
architecture: "arm64",
44+
image: MUSE_GLIMMER_VLLM_IMAGE_REFERENCE,
45+
imageDownloadSizeBytes: 9_706_339_423,
46+
});
47+
48+
const profile = detectVllmProfile({ platform: "spark", type: "nvidia" });
49+
const model = VLLM_MODELS.find(({ envValue }) => envValue === "muse-glimmer-30b");
50+
expect(profile).not.toBeNull();
51+
expect(model).toBeDefined();
52+
expect(resolveVllmRuntimeProfile(profile!, model!)).toMatchObject({
53+
image: MUSE_GLIMMER_VLLM_IMAGE_REFERENCE,
54+
imageDownloadSizeBytes: 9_706_339_423,
55+
});
56+
});
57+
3858
// source-shape-contract: security -- Mutating each trust field proves the reviewed provenance record fails closed before a substituted external runtime can be published.
3959
it.each([
4060
["publisher drift", ["publisher", "namespace"], "attacker"],
@@ -43,6 +63,12 @@ describe("Muse Glimmer vLLM image provenance", () => {
4363
["source drift", ["build", "sourceRevision"], "0".repeat(40)],
4464
["label drift", ["reportedLabels", "org.opencontainers.image.revision"], "0".repeat(40)],
4565
["support ancestry drift", ["upstreamSupport", "relationship"], "unverified"],
66+
["runtime dependency drift", ["runtimeDependencies", "huggingfaceHubVersion"], "1.27.0"],
67+
[
68+
"revision serialization drift",
69+
["revisionSerialization", "resolvedRevisionAfterPickle"],
70+
"main",
71+
],
4672
] as const)("rejects %s", (_name, pathParts, replacement) => {
4773
const changed = structuredClone(provenance) as JsonRecord;
4874
let target = changed;

‎test/support/muse-glimmer-vllm-image-provenance-test-support.ts‎

Lines changed: 36 additions & 18 deletions
Original file line numberDiff line numberDiff line change
@@ -3,13 +3,14 @@
33

44
import assert from "node:assert/strict";
55

6-
const MANIFEST_DIGEST = "sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204";
6+
const MANIFEST_DIGEST = "sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965";
77
export const MUSE_GLIMMER_VLLM_IMAGE_REFERENCE = `vllm/vllm-openai@${MANIFEST_DIGEST}`;
8-
const CONFIG_DIGEST = "sha256:c3f199e54a26d2d7a9a41115cd07ce9d90a6488c5a4e75b17129e1006ce533fd";
9-
const SOURCE_REVISION = "ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9";
8+
const CONFIG_DIGEST = "sha256:49d2eb65dc2a8dea24e43c27b226f650481ac97d4ba9c567b6e1ca08bc472303";
9+
const SOURCE_REVISION = "5a4c8d99242e9e069b604d0e9b969e77f7dd501d";
1010
const MUSE_MERGE_COMMIT = "6adad08767583f52eb4d2122111af0bf638ed5e6";
11+
const REVISION_FIX_COMMIT = "90984ddbed27a09409506d6d6c0eea87f54b04b5";
12+
const MODEL_REVISION = "d35cb79050f419c457611b1cee5c5d15b176f285";
1113
const PIPELINE_ID = "019d130e-464e-4ff7-b84b-492992c0c06b";
12-
const PIPELINE_URL = "https://buildkite.com/vllm/release-v2/builds/5174";
1314

1415
const EXPECTED_PROVENANCE = {
1516
schemaVersion: 1,
@@ -30,42 +31,59 @@ const EXPECTED_PROVENANCE = {
3031
manifestUrl: `https://registry-1.docker.io/v2/vllm/vllm-openai/manifests/${MANIFEST_DIGEST}`,
3132
configDigest: CONFIG_DIGEST,
3233
configMediaType: "application/vnd.docker.container.image.v1+json",
33-
configSizeBytes: 34_762,
34+
configSizeBytes: 34_738,
3435
configUrl: `https://registry-1.docker.io/v2/vllm/vllm-openai/blobs/${CONFIG_DIGEST}`,
3536
layerCount: 32,
36-
compressedLayerSizeBytes: 9_699_710_136,
37-
createdAt: "2026-08-14T05:33:50.528328374Z",
37+
compressedLayerSizeBytes: 9_706_339_423,
38+
createdAt: "2026-08-19T05:34:00.562836767Z",
3839
platform: {
3940
os: "linux",
4041
architecture: "arm64",
4142
},
4243
},
4344
build: {
44-
sourceRepository: "https://github.com/vllm-project/vllm",
4545
sourceRevision: SOURCE_REVISION,
46-
sourceRevisionUrl: `https://github.com/vllm-project/vllm/commit/${SOURCE_REVISION}`,
4746
imageTag: `vllm/vllm-openai:nightly-${SOURCE_REVISION}`,
4847
pipelineId: PIPELINE_ID,
49-
pipelineUrl: PIPELINE_URL,
5048
},
5149
upstreamSupport: {
5250
museMergeCommit: MUSE_MERGE_COMMIT,
53-
museMergeCommitUrl: `https://github.com/vllm-project/vllm/commit/${MUSE_MERGE_COMMIT}`,
54-
comparisonUrl: `https://github.com/vllm-project/vllm/compare/${MUSE_MERGE_COMMIT}...${SOURCE_REVISION}`,
55-
relationship: "direct-descendant",
56-
aheadBy: 1,
51+
relationship: "descendant",
52+
aheadBy: 173,
53+
revisionFixCommit: REVISION_FIX_COMMIT,
54+
revisionFixRelationship: "descendant",
55+
revisionFixAheadBy: 33,
5756
},
5857
reportedLabels: {
5958
"ai.vllm.build.commit": SOURCE_REVISION,
6059
"ai.vllm.build.pipeline": PIPELINE_ID,
61-
"ai.vllm.build.url": PIPELINE_URL,
6260
"ai.vllm.image.tag": `vllm/vllm-openai:nightly-${SOURCE_REVISION}`,
6361
"org.opencontainers.image.revision": SOURCE_REVISION,
64-
"org.opencontainers.image.source": "https://github.com/vllm-project/vllm",
62+
},
63+
runtimeDependencies: {
64+
vllmVersion: "0.26.1rc1.dev942+g5a4c8d992",
65+
huggingfaceHubVersion: "1.28.0",
66+
sentencepieceVersion: "0.2.2",
67+
tiktokenVersion: "0.14.0",
68+
},
69+
revisionSerialization: {
70+
model: "Inferact/Muse-Glimmer-30B-NVFP4-W4A4",
71+
requestedRevision: MODEL_REVISION,
72+
resolvedRevisionBeforePickle: MODEL_REVISION,
73+
resolvedRevisionAfterPickle: MODEL_REVISION,
74+
preserved: true,
6575
},
6676
verification: {
67-
observedAt: "2026-08-14T16:33:50Z",
68-
methods: ["docker-buildx-imagetools-inspect-raw", "docker-image-inspect", "github.qkg1.toppare-api"],
77+
observedAt: "2026-08-19T22:53:46Z",
78+
methods: [
79+
"docker-buildx-imagetools-inspect-raw",
80+
"docker-image-inspect",
81+
"github.qkg1.toppare-api",
82+
"container-package-imports",
83+
"resolved-revision-pickle-round-trip",
84+
"dgx-spark-cold-cache-startup",
85+
"vllm-openai-api-validation",
86+
],
6987
signedProvenanceAttestation: "not-available",
7088
},
7189
} as const;

0 commit comments

Comments
 (0)