Create app.py

app.py

@@ -0,0 +1,69 @@
+import torch
+import tiktoken
+from model import *
+import gradio as gr
+
+enc = tiktoken.get_encoding('gpt2') 
+model = torch.load('model.pt',map_location='cpu') 
+
+
+def response(message = "Hello, I'm a language model", num_return_sequences = 5,max_length = 30,top_k = 50):
+  tokens = enc.encode(message)
+  tokens = torch.tensor(tokens, dtype= torch.long) # (8,) #check tiktoken app
+  tokens = tokens.unsqueeze(0).repeat(num_return_sequences, 1) # (5, 8)
+  x = tokens.to('cpu')
+  torch.manual_seed(42)
+  torch.cuda.manual_seed(42)
+  while x.size(1) < max_length:
+      # forward the model to get the logits
+      with torch.no_grad():
+          logits = model(x)[0] # (B, T, vocab_size)
+          # take the logits at the last position
+          logits = logits[:, -1, :] # (B, vocab_size)
+          # get the probabilities
+          probs = F.softmax(logits, dim=-1)
+          # do top-k sampling of 50 (huggingface pipeline default)
+          # topk_probs here becomes (5, 50), topk_indices is (5, 50)
+          topk_probs, topk_indices = torch.topk(probs, top_k, dim=-1)
+          # select a token from the top-k probabilities
+          # note: multinomial does not demand the input to sum to 1
+          ix = torch.multinomial(topk_probs, 1) # (B, 1)
+          # gather the corresponding indices
+          xcol = torch.gather(topk_indices, -1, ix) # (B, 1)
+          # append to the sequence
+          x = torch.cat((x, xcol), dim=1)
+
+  # print the generated text
+  return_text = ""
+  for i in range(num_return_sequences):
+      tokens = x[i, :max_length].tolist()
+      decoded = enc.decode(tokens)
+      return_text = return_text + ">"+ decoded +"\n"
+  return return_text
+
+
+
+# Function to generate text
+def generate_text(prompt, top_k, max_return_sequences, max_tokens):
+    ans = response(message = prompt, num_return_sequences = max_return_sequences,max_length = max_tokens,top_k = top_k)
+    
+
+    return ans
+
+# Create Gradio interface
+iface = gr.Interface(
+    fn=generate_text,
+    inputs=[
+        gr.Textbox(lines=5, label="Input Text"),
+        gr.Slider(minimum=1, maximum=100, value=50, step=1, label="Top-k"),
+        gr.Slider(minimum=1, maximum=5, value=1, step=1, label="Return Sequences"),
+        gr.Slider(minimum=10, maximum=150, value=20, step=5, label="Max Tokens")
+    ],
+    outputs=gr.Textbox(label="Generated Text"),
+    title="GPT-2 Text Generator",
+    description="Generate text using GPT-2 model with adjustable parameters.",
+    examples=[["Hello, I'm a language model"]],
+)
+
+# Launch the interface
+iface.launch()