README.md

---
library_name: gliner2
license: other
license_name: unverified-review-required
base_model: jhu-clsp/mmBERT-base
language:
  - en
tags:
  - gliner2
  - information-extraction
  - named-entity-recognition
  - relation-extraction
  - event-extraction
  - text-classification
metrics:
  - f1
  - precision
  - recall
pipeline_tag: token-classification
---

# mmbert_base_rams

A [GLiNER2](https://github.com/fastino-ai/GLiNER2) multi-task information-extraction model (entities, relations, events, and classification) fine-tuned from `jhu-clsp/mmBERT-base`.

## ⚠️ License at a glance

- **Effective license:** Unverified — review required
- **Commercial use:** Unverified
- **All dataset licenses verified:** No

See [License](#license) for the full determination and per-dataset terms.

## Model details

- **Base model:** [`jhu-clsp/mmBERT-base`](https://huggingface.co/jhu-clsp/mmBERT-base)
- **Library:** `gliner2`
- **Tasks:** entity, relation, event, and classification extraction
- **Experiment:** `mmbert_base_rams`

## Training data

**1** dataset used for this run. 7,329 training records (val: 924, test: 871).

| Dataset | Task(s) | Train | Val | Test | Language | License | Source |
|---|---|--:|--:|--:|---|---|---|
| RAMS | Event extraction (trigger + args) | 7,329 | 924 | 871 | en | see source | [link](https://nlp.jhu.edu/rams/) |

**Dataset notes**

- **RAMS** — Multi-sentence event extraction with triggers and typed arguments; 139 event types, 65 argument roles.

## Training procedure

| Setting | Value |
|---|---|
| Trained on | 2026-07-31 |
| Duration | 59m 2s |
| Throughput | 31.0 samples/s |
| Epochs | 15 |
| Batch size | 8 (× 4 grad-accum) |
| Encoder LR | 2e-05 |
| Task-head LR | 0.0005 |
| Weight decay | 0.01 |
| Scheduler | cosine_restarts (warmup 0.05) |
| Precision | bf16 |
| Max grad norm | 1.0 |
| Best-checkpoint metric | eval_event_argument_strict_micro_f1 |
| Seed | 42 |
| Architecture | `max_width=20`, `max_len=8192`, `struct_loss=bce_posweight`, `struct_pos_weight=4.0` |

## Evaluation

Decision threshold: **0.5** (calibrated against the validation set).

### Blind test (held-out test splits)

Micro precision / recall / F1, strict → relaxed.

| Category | Precision | Recall | F1 | Support |
|---|--:|--:|--:|--:|
| event_type | 1.000 → 1.000 | 0.930 → 0.930 | 0.964 → 0.964 | 848 |
| event_trigger | 0.482 → 0.484 | 0.834 → 0.837 | 0.611 → 0.613 | 848 |
| event_argument | 0.031 → 0.133 | 0.125 → 0.540 | 0.050 → 0.213 | 2016 |
| event | 0.167 → 0.247 | 0.471 → 0.697 | 0.247 → 0.365 | 3712 |

### Best checkpoint (validation)

Micro precision / recall / F1, strict → relaxed.

| Category | Precision | Recall | F1 | Support |
|---|--:|--:|--:|--:|
| event_type | 1.000 → 1.000 | 0.808 → 0.808 | 0.894 → 0.894 | 896 |
| event_trigger | 0.556 → 0.562 | 0.688 → 0.695 | 0.615 → 0.622 | 896 |
| event_argument | 0.049 → 0.185 | 0.072 → 0.271 | 0.059 → 0.220 | 2182 |
| event | 0.298 → 0.385 | 0.377 → 0.488 | 0.333 → 0.431 | 3974 |

## License

**Effective license: Unverified — review required.** This model is a derivative of its base model and every training dataset, so the most restrictive term across all of them governs the whole model.

- **Commercial use:** Unverified
- **Share-alike obligation:** No
- **All licenses verified:** No
- **Base model:** mmBERT-base — see model card

**Unverified — verify the upstream terms before redistribution**
- RAMS (see source)
- mmBERT-base (see model card)

> License strings are copied verbatim from each dataset's card/source and from `tools/train/dataset_registry.yaml`. "see card"/"see source"/"other" mean the upstream declares no clear license — treat as unverified. This summary is informational, not legal advice; confirm terms before redistribution or commercial use.

## Citation

If you use this model, please cite GLiNER2 and the underlying datasets (linked in [Training data](#training-data)).

---
_Model card generated automatically at the end of training (2026-07-31)._