Update app.py

app.py

@@ -1,39 +1,20 @@
 import gradio as gr
-import torch
-from transformers import AutoTokenizer, AutoModel
 import plotly.graph_objects as go
+import hashlib
 
-model_name = "mistralai/Mistral-7B-v0.1"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = None
-
-# Set pad token to eos token if not defined
-if tokenizer.pad_token is None:
-    tokenizer.pad_token = tokenizer.eos_token
-
-def get_embedding(text):
-    global model
-    if model is None:
-        model = AutoModel.from_pretrained(model_name)
-        model.resize_token_embeddings(len(tokenizer))
-    
-    inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True, max_length=512)
-    with torch.no_grad():
-        outputs = model(**inputs)
-    return outputs.last_hidden_state.mean(dim=1).squeeze().detach().numpy()
-
-def reduce_to_3d(embedding):
-    return embedding[:3]
+def simple_embedding(text, dim=3):
+    """A simple hash-based embedding function for demonstration purposes."""
+    hash_value = hashlib.md5(text.encode()).hexdigest()
+    return [int(hash_value[i:i+2], 16) / 255.0 for i in range(0, dim*2, 2)]
 
 def compare_embeddings(*texts):
-    embeddings = [get_embedding(text) for text in texts if text.strip()]  # Only process non-empty texts
-    embeddings_3d = [reduce_to_3d(emb) for emb in embeddings]
+    embeddings = [simple_embedding(text) for text in texts if text.strip()]  # Only process non-empty texts
     
     fig = go.Figure()
     
     colors = ['red', 'blue', 'green', 'purple', 'orange', 'cyan', 'magenta', 'yellow']
     
-    for i, emb in enumerate(embeddings_3d):
+    for i, emb in enumerate(embeddings):
         color = colors[i % len(colors)]
         fig.add_trace(go.Scatter3d(
             x=[0, emb[0]], y=[0, emb[1]], z=[0, emb[2]],
@@ -48,8 +29,9 @@ def compare_embeddings(*texts):
     return fig
 
 with gr.Blocks() as iface:
-    gr.Markdown("# 3D Embedding Comparison")
-    gr.Markdown("Compare the embeddings of multiple strings visualized in 3D space using Mistral 7B.")
+    gr.Markdown("# 3D Embedding Comparison (Simplified)")
+    gr.Markdown("Compare simplified embeddings of multiple strings visualized in 3D space.")
+    gr.Markdown("Note: This is a demonstration using a basic hash-based embedding, not a real NLP model.")
     
     with gr.Row():
         num_inputs = gr.Slider(minimum=2, maximum=10, step=1, value=2, label="Number of texts to compare")