---
license: apache-2.0
pipeline_tag: token-classification
tags:
  - ner
  - bio
  - lfm2
  - data-use
---

# lfm2.5-Encoder-350M-datause-extended

Fine-tune of `LiquidAI/LFM2.5-Encoder-350M` for BIO data-use mention tagging
(dataset / survey / census / registry mentions in economics research papers).

## Labels
- `NAMED_DATA` — a proper name, title, or acronym of a specific data source
- `DESCRIPTIVE_DATA` — a source described in words but not named
- `VAGUE_DATA` — generic data wording with no identifiable source

## Training
- base model: `LiquidAI/LFM2.5-Encoder-350M`
- dataset: `rafmacalaba/data-use-mentions-extended` (bio config)
- epochs: 5
- learning rate: 2e-05
- batch size: 16
- precision: bf16

## Evaluation (holdout, label-agnostic)

| thr | tp | fp | fn | precision | recall | f0.5 | f1 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 0.10 | 7803 | 1715 | 4761 | 0.8198 | 0.6211 | 0.7705 | 0.7067 |
| 0.20 | 7803 | 1715 | 4761 | 0.8198 | 0.6211 | 0.7705 | 0.7067 |
| 0.30 | 7803 | 1715 | 4761 | 0.8198 | 0.6211 | 0.7705 | 0.7067 |
| 0.40 | 7787 | 1702 | 4777 | 0.8206 | 0.6198 | 0.7707 | 0.7062 |
| 0.50 | 7658 | 1625 | 4906 | 0.8249 | 0.6095 | 0.7705 | 0.7011 |
| 0.60 | 7083 | 1304 | 5481 | 0.8445 | 0.5638 | 0.7680 | 0.6761 |
| 0.70 | 6560 | 1057 | 6004 | 0.8612 | 0.5221 | 0.7622 | 0.6501 |

**Best F0.5**: 0.7707 (thr=0.4)
**Best F1**: 0.7067 (thr=0.1)