Upload 4 files

app.py

@@ -0,0 +1,47 @@
+import gradio as gr
+from corrector import GrammarCorrector
+import difflib
+
+corrector = GrammarCorrector()
+
+def highlight_diffs(original, corrected):
+    diff = difflib.ndiff(original.split(), corrected.split())
+    result = []
+    for word in diff:
+        if word.startswith("-"):
+            result.append(f"~~{word[2:]}~~")
+        elif word.startswith("+"):
+            result.append(f"**{word[2:]}**")
+        elif word.startswith(" "):
+            result.append(word[2:])
+    return " ".join(result)
+
+def fix_sentence(text):
+    corrected = corrector.correct(text)
+    highlighted = highlight_diffs(text, corrected)
+    return corrected, highlighted
+
+gr.Interface(
+    fn=fix_sentence,
+    inputs=gr.Textbox(lines=2, label="Input Sentence"),
+    outputs=[
+        gr.Textbox(label="Corrected Sentence"),
+        gr.Markdown(label="Changes Highlighted")
+    ],
+    title="Context-Aware Grammar & Spell Checker",
+    description="Fixes grammar and spelling using a T5-based model.",
+
+    examples=[
+        ["She go to school every day."],
+        ["I can has cheeseburger?"],
+        ["The cat sleeped on the mat."],
+        ["We was going to the park yesterday."],
+        ["This is teh best day of my life!"],
+        ["He no went to office today."],
+        ["Their coming too the party."],
+        ["I hopes you gets better soon."],
+        ["Where is you going now?"],
+        ["He do not likes pizza."]
+    ]
+).launch()
+

corrector.py

@@ -0,0 +1,13 @@
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+
+class GrammarCorrector:
+    def __init__(self, model_name="vennify/t5-base-grammar-correction"):
+        self.tokenizer = AutoTokenizer.from_pretrained(model_name)
+        self.model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
+
+    def correct(self, text):
+        input_text = "gec: " + text # gec: grammar error correction task
+        input_ids = self.tokenizer.encode(input_text, return_tensors="pt")
+        outputs = self.model.generate(input_ids, max_length=128, num_beams=5, early_stopping=True)
+        corrected_text = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
+        return corrected_text

examples.txt

@@ -0,0 +1,15 @@
+She go to school every day.
+I can has cheeseburger?
+The cat sleeped on the mat.
+We was going to the park yesterday.
+This is teh best day of my life!
+He no went to office today.
+Their coming too the party.
+I hopes you gets better soon.
+Where is you going now?
+He do not likes pizza.
+They was happy with there results.
+This are the answers to your questions.
+My mom sayed I can go.
+Your the best friend I ever had.
+He eated all the cake himself.

requirements.txt

@@ -0,0 +1,4 @@
+transformers
+torch
+gradio
+sentencepiece