Model page

hugging-quants/gemma-2-9b-it-AWQ-INT4 pass

No issues found by the checks that have run.

Could not load this model from the Hugging Face API (private, gated, or nonexistent). Findings below are from our archive.

Scan coverageStatic battery2026-08-21Weights battery2026-08-21Behavioral batterynot rundetails
BatteryLooks atStatus
Static batteryMetadata & packagingcomplete 2026-08-21
Weights batteryWeights forensics — no GPU, no downloadcomplete 2026-08-21256,000-token embedding scanned · 0 undertrained
Behavioral batteryLive-inference differentialsnot run

Ingot runs three batteries against a model. What each one checks →

Findings

Scanned 2026-08-21 · published from a community scan.

info Embedding-norm glitch scan clean

No undertrained tokens found: every non-special token's embedding norm is above 0.3× the vocabulary median (1.701). The glitch-token data-corruption class has no candidate surface in this model.

Remediation guidance addresses the documented findings only. It is evidence-driven repair, not a safety certification of the model.

Fingerprint

The durable profile of this model: measured weights-and-metadata facts, rebuilt on every scan and battery run. Updated 2026-08-21.

architecturegemma2 · 42 layers · 3584-dim
parameters9241.7M
vocabulary256,000 tokens
licensegemma
serializationsafetensors
chat templatepresent · sha256:ecd6ae513fe103f0
claimed lineagegoogle/gemma-2-9b-it
lineage verifiedunverified — weights battery pending
glitch-token surfaceclean no undertrained tokens
Full measured fingerprint
architecturesGemma2ForCausalLM
librarytransformers
pipelinetext-generation
repo files11
revision6e62725da8e9
HF snapshot445.7k downloads · 10 likes · updated 2024-10-17 · captured 2026-08-21
embedding tensormodel.embed_tokens.weight · BF16 · 256,000×3584
embedding normsmedian 1.701 · mean 1.7215
lineage checkmodel metadata unreadable
Battery runs (1)the run trace behind the findings — what each job measured
batterystatusqueueddurationattempts
weightscomplete2026-08-21 05:082m1
weights run 2026-08-21 measurements
probes runglitch-norm-scan, lineage-norm-correlation
embedding tensormodel.embed_tokens.weight · BF16 · 256,000×3584
glitch surface0 undertrained, 0 plain-ASCII
lineage checknot checked (model metadata unreadable)

Verdict badge

Ship the verdict in your README — it always shows the latest published analysis:

Ingot verdict: pass

[![Ingot scan](https://ingot.tools/api/v1/models/hugging-quants/gemma-2-9b-it-AWQ-INT4/badge.svg)](https://ingot.tools/models/hugging-quants/gemma-2-9b-it-AWQ-INT4)
Gate it in CI