update

app.py

@@ -1,28 +1,19 @@
 import gradio as gr
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
 import torch
-import numpy as np
-from transformers import AutoModelForSequenceClassification
 
-# Load ONLY the model, NOT the tokenizer
+# Load model and tokenizer
 model = AutoModelForSequenceClassification.from_pretrained(
     "Kevintu/Engessay_grading_ML")
+tokenizer = AutoTokenizer.from_pretrained("KevSun/Engessay_grading_ML")
 
 
-def process_embeddings(embeddings_array):
-    # Convert the received embeddings to the format expected by the model
-    embeddings_tensor = torch.tensor(embeddings_array)
-
-    # Process embeddings with the model
+def grade_essay(text):
+    encoded_input = tokenizer(
+        text, return_tensors='pt', padding=True, truncation=True, max_length=64)
     model.eval()
     with torch.no_grad():
-        # Create a dict with the expected input format
-        model_inputs = {
-            'input_ids': None,  # Not needed since we're using embeddings directly
-            'attention_mask': None,  # Not needed for this use case
-            'inputs_embeds': embeddings_tensor  # Pass embeddings directly
-        }
-        outputs = model(**model_inputs)
-
+        outputs = model(**encoded_input)
     predictions = outputs.logits.squeeze()
 
     item_names = ["cohesion", "syntax", "vocabulary",
@@ -35,14 +26,18 @@ def process_embeddings(embeddings_array):
     return results
 
 
-# Create Gradio interface for embeddings input
+# Create Gradio interface
 demo = gr.Interface(
-    fn=process_embeddings,
-    inputs=gr.JSON(label="Embeddings"),
-    outputs=gr.JSON(label="Scores"),
+    fn=grade_essay,
+    inputs=gr.Textbox(lines=10, placeholder="Enter essay text here..."),
+    outputs=gr.JSON(),
     title="Essay Grading API",
-    description="Grade essays based on precomputed embeddings"
+    description="Grade essays on six dimensions of writing quality",
+    examples=[
+        ["The English Language Learner Insight, Proficiency and Skills Evaluation (ELLIPSE) Corpus is a freely available corpus of ~6,500 ELL writing samples that have been scored for overall holistic language proficiency as well as analytic proficiency scores."]
+    ],
 )
 
+# For API access
 demo.queue()
 demo.launch()