Update README.md

README.md

@@ -11,6 +11,53 @@ tags: []
 
 ## Model Details
 
+### Code to create models
+
+```python
+import torch
+
+from transformers import (
+    AutoTokenizer,
+    EuroBertConfig,
+    EuroBertForMaskedLM,
+    EuroBertForSequenceClassification,
+    EuroBertForTokenClassification,
+    EuroBertModel,
+)
+
+
+classes = [
+    EuroBertForTokenClassification,
+    EuroBertModel,
+    EuroBertForMaskedLM,
+    EuroBertForSequenceClassification,
+]
+
+for klass in classes:
+    model_id = "EuroBERT/EuroBERT-210m"
+    config = EuroBertConfig.from_pretrained(
+        model_id,
+        hidden_size=32,
+        intermediate_size=64,
+        num_hidden_layers=4,
+        num_attention_heads=2,
+        num_key_value_heads=2,
+    )
+    del config.auto_map  # Avoid custom code loading
+
+    # Create model and randomize all weights
+    model = klass(config)
+
+    torch.manual_seed(0)  # Set for reproducibility
+    for name, param in model.named_parameters():
+        param.data = torch.randn_like(param)
+
+    tokenizer = AutoTokenizer.from_pretrained(model_id, config=config)
+
+    model.push_to_hub(f"tiny-random-{klass.__name__}")
+    tokenizer.push_to_hub(f"tiny-random-{klass.__name__}")
+```
+
 ### Model Description
 
 <!-- Provide a longer summary of what this model is. -->