Model cardKnowledgator
GLiClass large v3.0 model card
Measured on: 14-dataset zero-shot classification average (F1)
- Zero-shot F1 average
- 0.490
Author-reported; standard text-classification sets, not decisions.
GLiClass edge v3.0
KnowledgatorOpen weights
The 33M latency floor of the family: about 40 ms per three-question case on four CPU cores.
- typed-decisions
- —
- Metatext suite
- 0.418
- ECE untuned
- 0.287
- p50 per case
- 41 ms
not run
2/7 over baseline
four CPU cores
Specifications
- Checkpoint
- knowledgator/gliclass-edge-v3.0
- Publisher
- Knowledgator
- Mechanism
- Label-relevance classifier
- Backbone
- small encoder
- Parameters
- 33M
- License
- Apache 2.0
- Runs as
- open weights, self-hosted
- Library
- gliclass
- Primitives
- Choice, Noul
- Pricing
- none — your compute
Metatext decisionsrank 20 of 21
Full leaderboard →| Configuration | Mode | Decisions | Beats baseline | ECE |
|---|---|---|---|---|
| criteria + single-hypothesis Noul (correct payload)headline | serial | 0.418 | 2/7 | 0.287 |
| default payload | serial | 0.386 | 2/7 | 0.484 |
| criteria folded into labels | serial | 0.383 | 2/7 | 0.387 |
Per-question accuracy — criteria + single-hypothesis Noul (correct payload), serial mode, 120 judgments
| Task | Question | Type | Accuracy | Baseline | Beats? |
|---|---|---|---|---|---|
| agent_traces | outcome | choice | 0.143 | 0.357 | ❌ |
| agent_traces | needs_human | noul | 0.643 | 0.714 | ❌ |
| customer_service | department | choice | 0.450 | 0.400 | ✅ |
| customer_service | refund_requested | noul | 0.800 | 0.700 | ✅ |
| customer_service | frustration | score | 0.300 | 0.500 | ❌ |
| security_incidents | severity | choice | 0.250 | 0.312 | ❌ |
| security_incidents | actively_exploited | noul | 0.438 | 0.688 | ❌ |
Published numbers
Reported by the authors or by third parties, on their own suites. Not comparable to the leaderboard above; attributed so you can check them.
Frequently asked questions
How does GLiClass edge v3.0 compare to Jev?
On the Metatext suite GLiClass edge v3.0 scores 0.418 against Jev's 0.863.
Is GLiClass edge v3.0 open source?
Yes. knowledgator/gliclass-edge-v3.0 is released under Apache 2.0 and runs on your own hardware with the gliclass library.
Can GLiClass edge v3.0 extract entities?
No. GLiClass edge v3.0 answers questions about text and never returns spans; use a GLiNER-family model for extraction.