Fill-Mask
Transformers
Safetensors
Urdu
modernbert
urdu
encoder
masked-language-modeling
long-context
8k-context
urblimp
zero-shot
benchmark
Instructions to use ProximaAI/urnova-95m with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use ProximaAI/urnova-95m with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("fill-mask", model="ProximaAI/urnova-95m")# Load model directly from transformers import AutoTokenizer, AutoModelForMaskedLM tokenizer = AutoTokenizer.from_pretrained("ProximaAI/urnova-95m") model = AutoModelForMaskedLM.from_pretrained("ProximaAI/urnova-95m", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| urnova-95m | |
| Copyright 2026 The urnova authors | |
| The original urnova model weights, tokenizer artifacts, code, documentation, | |
| benchmark reports, and visualizations included in this repository are made | |
| available under the Apache License, Version 2.0, to the extent of the rights | |
| held by their respective copyright holders. | |
| The model was trained from scratch using an Urdu corpus derived from HPLT 3.0. | |
| HPLT licenses the packaging of its dataset under CC0 but states that it does | |
| not own the underlying extracted web text. | |
| The Apache License does not grant rights in the underlying training text, | |
| third-party content, trademarks, privacy rights, publicity rights, database | |
| rights, or text that a model may reproduce. | |
| No raw HPLT training documents are distributed in this model repository. | |
| See DATA_AND_THIRD_PARTY_NOTICE.md for provenance, limitations, and the | |
| takedown procedure. | |