Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
language:
|
| 5 |
+
- it
|
| 6 |
+
tags:
|
| 7 |
+
- fill-mask
|
| 8 |
+
- masked-lm
|
| 9 |
+
- long-context
|
| 10 |
+
- modernbert
|
| 11 |
+
- italian
|
| 12 |
+
pipeline_tag: fill-mask
|
| 13 |
+
---
|
| 14 |
+
|
| 15 |
+
# Italian ModernBERT
|
| 16 |
+
|
| 17 |
+
## Model Description
|
| 18 |
+
|
| 19 |
+
Italian ModernBERT (DeepMount00/Italian-ModernBERT-base) is a specialized Italian language version of ModernBERT, specifically pre-trained on Italian text corpora. This model is designed exclusively for Italian language tasks.
|
| 20 |
+
|
| 21 |
+
## Key Features
|
| 22 |
+
|
| 23 |
+
- **Architecture**: Based on ModernBERT-base (22 layers, 149M parameters)
|
| 24 |
+
- **Context Length**: 8,192 tokens
|
| 25 |
+
- **Language**: Italian-only
|
| 26 |
+
- **Tokenizer**: Custom tokenizer optimized for Italian language
|
| 27 |
+
- **Training**: Pre-trained on Italian text corpus
|
| 28 |
+
|
| 29 |
+
## Technical Details
|
| 30 |
+
|
| 31 |
+
- Uses Rotary Positional Embeddings (RoPE)
|
| 32 |
+
- Implements Local-Global Alternating Attention
|
| 33 |
+
- Supports Flash Attention 2 for optimal performance
|
| 34 |
+
- No token type IDs required
|
| 35 |
+
|
| 36 |
+
|
| 37 |
+
## Limitations
|
| 38 |
+
|
| 39 |
+
- Optimized only for Italian language processing
|
| 40 |
+
- Not suitable for other languages
|
| 41 |
+
- May reflect biases present in training data
|