first update

README.md

@@ -11,4 +11,33 @@ license: apache-2.0
 short_description: ES→PT Translation (BART from scratch)
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+## Overview
+This Space provides a simple **Spanish → Portuguese (ES→PT)** translation demo using a **BART-style encoder–decoder Transformer trained from scratch**.
+
+- **Dataset:** Helsinki-NLP/Tatoeba (es-pt)
+- **Tokenizer:** Subword BPE (32k vocab)
+- **Metric reported:** chrF (beam search)
+
+## Model
+This demo loads the model from the Hugging Face Hub:
+
+**Model card:** REPLACE_WITH_YOUR_MODEL_LINK  
+(e.g., `https://huggingface.co/<username>/<model-repo>`)
+
+## How to use
+1. Enter a sentence in **Spanish (es)**.
+2. Click **Submit**.
+3. The app returns the model’s **Portuguese (pt)** translation.
+
+## Notes / Limitations
+- The model was trained on Tatoeba-style sentence pairs, so very long or highly technical sentences may perform worse.
+- Minor punctuation differences or occasional repetition can occur with beam search.
+
+## Example inputs
+- `¡Intentemos algo!`
+- `Las personas dicen que estoy loco.`
+- `¿Cuál es el significado de la vida en la tierra?`
+
+## References
+- Dataset: https://huggingface.co/datasets/Helsinki-NLP/tatoeba
+- Spaces config reference: https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,42 @@
+import gradio as gr
+import torch
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+
+MODEL_ID = "liansheng06/bart-tatoeba-es-pt"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_ID)
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+model.to(device)
+model.eval()
+
+def translate_es_to_pt(text):
+    if not text or not text.strip():
+        return ""
+    inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=256).to(device)
+    with torch.no_grad():
+        outputs = model.generate(
+            **inputs,
+            max_new_tokens=128,
+            num_beams=5,
+            length_penalty=1.0,
+            early_stopping=True,
+        )
+    return tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+demo = gr.Interface(
+    fn=translate_es_to_pt,
+    inputs=gr.Textbox(lines=4, label="Spanish (es) input"),
+    outputs=gr.Textbox(lines=4, label="Portuguese (pt) output"),
+    title="ES → PT Translator (BART from scratch)",
+    description="Model trained from scratch on Helsinki-NLP/Tatoeba (es-pt).",
+    examples=[
+        ["¡Intentemos algo!"],
+        ["Las personas dicen que estoy loco."],
+        ["¿Cuál es el significado de la vida en la tierra?"],
+    ],
+)
+
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,6 @@
+gradio
+torch
+transformers
+sentencepiece
+safetensors
+accelerate