Update app.py

app.py

@@ -1,9 +1,83 @@
 import streamlit as st
 import torch
-import tiktoken
-from dataclasses import dataclass
 import torch.nn as nn
 from torch.nn import functional as F
+import tiktoken
+import sys
+import os
+import logging
+import warnings
+from dataclasses import dataclass
+import math
+
+class MLP(nn.Module):
+
+    def __init__(self, config):
+        super().__init__()
+        self.c_fc    = nn.Linear(config.n_embd, 4 * config.n_embd)
+        self.gelu    = nn.GELU(approximate='tanh')
+        self.c_proj  = nn.Linear(4 * config.n_embd, config.n_embd)
+        self.c_proj.NANOGPT_SCALE_INIT = 1
+
+    def forward(self, x):
+        x = self.c_fc(x)
+        x = self.gelu(x)
+        x = self.c_proj(x)
+        return x
+
+
+class CausalSelfAttention(nn.Module):
+
+    def __init__(self, config):
+        super().__init__()
+        assert config.n_embd % config.n_head == 0
+        # key, query, value projections for all heads, but in a batch
+        self.c_attn = nn.Linear(config.n_embd, 3 * config.n_embd)
+        # output projection
+        self.c_proj = nn.Linear(config.n_embd, config.n_embd)
+        self.c_proj.NANGPT_SCALE_INIT = 1
+        # regularization
+        self.n_head = config.n_head
+        self.n_embd = config.n_embd
+        self.register_buffer("bias", torch.tril(torch.ones(config.block_size, config.block_size)).view(1, 1, config.block_size, config.block_size))
+
+    def forward(self, x):
+        B, T, C = x.size() # batch size, sequence length, embedding dimensionality (n_embd)
+        # calculate query, key, values for all heads in batch and move head forward to be the batch dim
+        # nh is "number of heads", hs is "head size", and C (number of channels) = nh * hs
+        # e.g. in GPT-2 (124M), n_head=12, hs=64, so nh*hs=C=768 channels in the Transformer
+        qkv = self.c_attn(x)
+        q, k, v = qkv.split(self.n_embd, dim=2)
+        k = k.view(B, T, self.n_head, C // self.n_head).transpose(1, 2) # (B, nh, T, hs)
+        q = q.view(B, T, self.n_head, C // self.n_head).transpose(1, 2) # (B, nh, T, hs)
+        v = v.view(B, T, self.n_head, C // self.n_head).transpose(1, 2) # (B, nh, T, hs)
+
+        att = (q @ k.transpose(-2, -1)) * (1.0 / math.sqrt(k.size(-1)))
+        att = att.masked_fill(self.bias[:, :, :T, :T] == 0, float('-inf'))
+        att = F.softmax(att, dim=-1)
+        y = att @ v # (B, nh, T, T) x (B, nh, T, hs) -> (B, nh, T, hs)
+
+        y = y.transpose(1, 2).contiguous().view(B, T, C) # re-assemble all head outputs side by side
+        # output projection
+        y = self.c_proj(y)
+        return y
+
+
+
+class Block(nn.Module):
+
+    def __init__(self, config):
+        super().__init__()
+        self.ln_1 = nn.LayerNorm(config.n_embd)
+        self.attn = CausalSelfAttention(config)
+        self.ln_2 = nn.LayerNorm(config.n_embd)
+        self.mlp = MLP(config)
+
+    def forward(self, x):
+        x = x + self.attn(self.ln_1(x))
+        x = x + self.mlp(self.ln_2(x))
+        return x
+
 
 @dataclass
 class GPTConfig:
@@ -119,65 +193,108 @@ class GPT(nn.Module):
         return model
 
 
-# Load the trained model
+# Configure logging and warnings
+logging.getLogger('streamlit').setLevel(logging.ERROR)
+warnings.filterwarnings('ignore', message='.*torch.classes.*')
+warnings.filterwarnings('ignore', category=FutureWarning)
+
+# Add the project root to Python path
+sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+
+
 @st.cache_resource
 def load_model():
+    device = "cpu"
     config = GPTConfig()
     model = GPT(config)
-    try:
-        # Load the model with map_location to handle CPU-only environments
-        model.load_state_dict(torch.load('trained_model_quantized.pt', map_location=torch.device('cpu')), strict=False)
-        model.eval()  # Set the model to evaluation mode
-        st.success("Model loaded successfully!")
-    except Exception as e:
-        st.error(f"Error loading model: {e}")
-    return model
-
-# Load the tokenizer
-def load_tokenizer():
-    return tiktoken.get_encoding('gpt2')
-
-# Generate text function
-def generate_text(model, tokenizer, input_text, length, num_sequences):
-    # Encode the input text
-    input_ids = tokenizer.encode(input_text)
-    input_tensor = torch.tensor(input_ids).unsqueeze(0)  # Add batch dimension (shape: [1, T])
-
-    generated_sequences = []
-    for _ in range(num_sequences):
-        # Generate additional tokens
-        with torch.no_grad():
-            for _ in range(length):
-                logits = model(input_tensor)[0]  # Get logits
-                next_token_logits = logits[:, -1, :]  # Get the last token's logits
-                next_token_probs = torch.softmax(next_token_logits, dim=-1)
-                next_token = torch.multinomial(next_token_probs, num_samples=1)  # Sample from the distribution
-                
-                # Ensure the next_token has the correct shape for concatenation
-                next_token = next_token.view(1, -1)  # Reshape to [1, 1] if necessary
-                input_tensor = torch.cat((input_tensor, next_token), dim=1)  # Append the new token
-
-        # Decode the generated tokens
-        generated_sequences.append(tokenizer.decode(input_tensor[0].tolist()))
-
-    return generated_sequences
-
-# Streamlit app layout
+    
+    # Load the trained weights from root directory
+    checkpoint = torch.load('trained_model_quantized.pt', map_location=device, weights_only=True)
+    
+    # Handle pruned weights
+    state_dict = checkpoint['model_state_dict']
+    new_state_dict = {}
+    
+    for key in model.state_dict().keys():
+        if key.endswith('.weight'):
+            # Check if this is a pruned weight
+            orig_key = key[:-7] + '.weight_orig' if key.endswith('.weight') else key
+            mask_key = key[:-7] + '.weight_mask' if key.endswith('.weight') else key
+            
+            if orig_key in state_dict and mask_key in state_dict:
+                # Reconstruct the pruned weight
+                new_state_dict[key] = state_dict[orig_key] * state_dict[mask_key]
+            else:
+                # Use the weight as is
+                new_state_dict[key] = state_dict[key] if key in state_dict else model.state_dict()[key]
+        else:
+            # Copy non-weight parameters as is
+            new_state_dict[key] = state_dict[key] if key in state_dict else model.state_dict()[key]
+    
+    # Load the processed state dict
+    model.load_state_dict(new_state_dict)
+    
+    # Convert back to float32 for inference
+    model = model.float()
+    model.to(device)
+    model.eval()
+    
+    return model, device
+
+def generate_text(model, prompt, max_length=100, num_return_sequences=1, device='cpu'):
+    tokenizer = tiktoken.get_encoding('gpt2')
+    input_tokens = tokenizer.encode(prompt)
+    x = torch.tensor(input_tokens).unsqueeze(0).repeat(num_return_sequences, 1)
+    x = x.to(device)
+    
+    # Calculate final length (input length + requested additional tokens)
+    input_length = x.size(1)
+    target_length = input_length + max_length
+    
+    # Generate text
+    with torch.no_grad():
+        while x.size(1) < target_length:
+            logits = model(x)[0]
+            next_token_logits = logits[:, -1, :]
+            probs = torch.softmax(next_token_logits, dim=-1)
+            next_token = torch.multinomial(probs, num_samples=1)
+            x = torch.cat((x, next_token), dim=1)
+    
+    # Print token information once before generating sequences
+    st.text(f"Size of Input tokens: {input_length}, Additional tokens to be predicted: {max_length}, Total tokens to be generated: {x.size(1)}")
+    
+    # Decode generated sequences
+    generated_texts = []
+    for i in range(num_return_sequences):
+        tokens = x[i].tolist()
+        text = tokenizer.decode(tokens)
+        generated_texts.append(text)
+    
+    return generated_texts
+
+# Streamlit UI
 st.title("GPT Text Generator")
-st.write("Enter your text and specify the length of additional text to generate.")
 
-input_text = st.text_area("Input Text", "Once upon a time", max_chars=512)  # Limit to 512 characters
-length = st.slider("Predict Additional Text of Length", 1, 50, 10)
-num_sequences = st.slider("Number of Sequences to Generate", 1, 5, 1)
+# Load model
+model, device = load_model()
+
+# Input form
+prompt = st.text_area("Enter your prompt:", "Once upon a time")
+max_length = st.slider("Predict additional text of length:", min_value=1, max_value=50, value=5)
+num_sequences = st.slider("Number of sequences to generate:", 1, 5, 1)
 
 if st.button("Generate"):
-    model = load_model()  # Load the model for inference
-    tokenizer = load_tokenizer()  # Load the tokenizer
-    st.write("Generating text...")
-    generated_texts = generate_text(model, tokenizer, input_text, length, num_sequences)
-    st.write("Text generation complete.")
-
-    st.write("Generated Texts:")
-    for i, text in enumerate(generated_texts):
-        st.subheader(f"Sequence {i + 1}")
-        st.write(text)
+    with st.spinner("Generating text..."):
+        generated_texts = generate_text(
+            model=model,
+            prompt=prompt,
+            max_length=max_length,
+            num_return_sequences=num_sequences,
+            device=device
+        )
+        
+        # Display results
+        for i, text in enumerate(generated_texts, 1):
+            st.write(f"\nSequence {i}:")
+            st.write(text)