Instructions to use SlayerLab/NERGAL with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use SlayerLab/NERGAL with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("token-classification", model="SlayerLab/NERGAL")# Load model directly from transformers import AutoTokenizer, AutoModelForTokenClassification tokenizer = AutoTokenizer.from_pretrained("SlayerLab/NERGAL") model = AutoModelForTokenClassification.from_pretrained("SlayerLab/NERGAL", device_map="auto") - Notebooks
- Google Colab
- Kaggle
1.0.3: registry and label-note rules, placeholder and card fixes
#1
by ppuzio - opened
Draft for 1.0.3. Weights, threshold and API unchanged.
- Rules: registry and label-note fixes (rules sha256 f32d5c54...).
- Wrapper: an existing [PII]/[Telefon] placeholder no longer switches the rules off for the whole text.
- Card: weights_sha256 is split into source_checkpoint_sha256 (the source checkpoint) and model_safetensors_sha256 (the published file). Card readers that use weights_sha256 must switch keys.
- Known limitation: an identifier with a placeholder inside or right before it is still not masked.
See CHANGELOG.md. Not for merge until release sign-off.
Generated with Claude Code
ppuzio changed pull request status to open
ppuzio changed pull request status to merged
ppuzio deleted the
refs/pr/1 ref