Skip to content

GLiClass instruct base v1.0

KnowledgatorOpen weights

The base-size instruction-tuned GLiClass.

typed-decisions
—
not run
Metatext suite
0.559
3/7 over baseline
ECE untuned
0.390
p50 per case
323 ms
four CPU cores

Specifications

Checkpoint
knowledgator/gliclass-instruct-base-v1.0
Publisher
Knowledgator
Mechanism
Label-relevance classifier
Backbone
DeBERTa-v3-base, instruction-tuned
Parameters
~190M
License
Apache 2.0
Runs as
open weights, self-hosted
Library
gliclass
Primitives
Choice, Noul
Pricing
none — your compute

Metatext decisionsrank 10 of 21

Full leaderboard →
ConfigurationModeDecisionsBeats baselineECE
criteria + single-hypothesis Noul (correct payload)headlineserial0.5593/70.390
default payloadserial0.3952/70.510
criteria folded into labelsserial0.3682/70.583
Per-question accuracy — criteria + single-hypothesis Noul (correct payload), serial mode, 120 judgments
TaskQuestionTypeAccuracyBaselineBeats?
agent_tracesoutcomechoice0.4290.357✅
agent_tracesneeds_humannoul0.7860.714✅
customer_servicedepartmentchoice0.8000.400✅
customer_servicerefund_requestednoul0.7000.700❌
customer_servicefrustrationscore0.3000.500❌
security_incidentsseveritychoice0.2500.312❌
security_incidentsactively_exploitednoul0.6880.688❌

Frequently asked questions

How does GLiClass instruct base v1.0 compare to Jev?

On the Metatext suite GLiClass instruct base v1.0 scores 0.559 against Jev's 0.863.

Is GLiClass instruct base v1.0 open source?

Yes. knowledgator/gliclass-instruct-base-v1.0 is released under Apache 2.0 and runs on your own hardware with the gliclass library.

Can GLiClass instruct base v1.0 extract entities?

No. GLiClass instruct base v1.0 answers questions about text and never returns spans; use a GLiNER-family model for extraction.