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

# lfm2.5-Encoder-350M-datause

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` (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 | 4577 | 1091 | 2769 | 0.8075 | 0.6231 | 0.7624 | 0.7034 |
| 0.20 | 4577 | 1091 | 2769 | 0.8075 | 0.6231 | 0.7624 | 0.7034 |
| 0.30 | 4576 | 1091 | 2770 | 0.8075 | 0.6229 | 0.7623 | 0.7033 |
| 0.40 | 4565 | 1086 | 2781 | 0.8078 | 0.6214 | 0.7621 | 0.7025 |
| 0.50 | 4501 | 1033 | 2845 | 0.8133 | 0.6127 | 0.7633 | 0.6989 |
| 0.60 | 4200 | 840 | 3146 | 0.8333 | 0.5717 | 0.7635 | 0.6782 |
| 0.70 | 3877 | 687 | 3469 | 0.8495 | 0.5278 | 0.7572 | 0.6510 |

**Best F0.5**: 0.7635 (thr=0.6)
**Best F1**: 0.7034 (thr=0.1)