Create README.md

README.md

@@ -0,0 +1,33 @@
+---
+language: es
+license: cc-by-sa-4.0
+datasets:
+- wikipedia
+- cc100
+widget:
+- text: "Yo soy"
+---
+
+## GPT2 Spanish base model (Uncased)
+
+### Model architecture
+
+This model uses GPT2 base setttings except vocabulary size.
+
+### Tokenizer
+
+Using BPE tokenizer with vocabulary size 50,000.
+
+### Training Data 
+
+* [wiki40b/es](https://www.tensorflow.org/datasets/catalog/wiki40b#wiki40bes) (Spanish Wikipedia)
+* Subset of [CC-100/es](https://data.statmt.org/cc-100/) : Monolingual Datasets from Web Crawl Data
+
+### Usage
+
+```python
+from transformers import pipeline
+
+generator = pipeline('text-generation', model='ClassCat/gpt2-base-spanish')
+generator("Yo soy ", max_length=50, num_return_sequences=5)
+```