Instructions to use arthrod/gliner-mmbert-small-ptbr-pii-full-3x-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- GLiNER
How to use arthrod/gliner-mmbert-small-ptbr-pii-full-3x-v1 with GLiNER:
from gliner import GLiNER model = GLiNER.from_pretrained("arthrod/gliner-mmbert-small-ptbr-pii-full-3x-v1") - Notebooks
- Google Colab
- Kaggle
card: add tiktoken count (~113M tokens)
Browse files
README.md
CHANGED
|
@@ -95,7 +95,7 @@ Additional label `berco-de-tiradentes` was integrated **from step 1** — not a
|
|
| 95 |
- **Focal loss**: α=0.75, γ=2.0, reduction=mean
|
| 96 |
- **LR**: 1.5e-5 (encoder) / 5e-5 (others), weight decay 0.01
|
| 97 |
- **Precision**: bf16 with `HIPBLASLT_ALLOW_TF32=0` (MI300X single-GPU partition)
|
| 98 |
-
- **Data**: `data/splits/train_with_ritapolis.jsonl` — ~
|
| 99 |
|
| 100 |
## Usage
|
| 101 |
|
|
|
|
| 95 |
- **Focal loss**: α=0.75, γ=2.0, reduction=mean
|
| 96 |
- **LR**: 1.5e-5 (encoder) / 5e-5 (others), weight decay 0.01
|
| 97 |
- **Precision**: bf16 with `HIPBLASLT_ALLOW_TF32=0` (MI300X single-GPU partition)
|
| 98 |
+
- **Data**: `data/splits/train_with_ritapolis.jsonl` — 986 491 rows, **~113.3M tokens** (cl100k_base; mean 114.8 tokens/row, max 9 908) — PT-BR PII + 2 000 Ritápolis (`berco-de-tiradentes`)
|
| 99 |
|
| 100 |
## Usage
|
| 101 |
|