Update app.py

app.py

@@ -4,41 +4,192 @@ import torch.nn as nn
 import sentencepiece as spm
 import math
 
+# Define Transformer components (unchanged)
+class MultiHeadAttention(nn.Module):
+    def __init__(self, d_model, num_heads):
+        super(MultiHeadAttention, self).__init__()
+        assert d_model % num_heads == 0
+        self.d_model = d_model
+        self.num_heads = num_heads
+        self.d_k = d_model // num_heads
+        self.W_q = nn.Linear(d_model, d_model)
+        self.W_k = nn.Linear(d_model, d_model)
+        self.W_v = nn.Linear(d_model, d_model)
+        self.W_o = nn.Linear(d_model, d_model)
+        
+    def scaled_dot_product_attention(self, Q, K, V, mask=None):
+        attn_scores = torch.matmul(Q, K.transpose(-2, -1)) / math.sqrt(self.d_k)
+        if mask is not None:
+            attn_scores = attn_scores.masked_fill(mask == 0, -1e9)
+        attn_probs = torch.softmax(attn_scores, dim=-1)
+        output = torch.matmul(attn_probs, V)
+        return output
+        
+    def split_heads(self, x):
+        batch_size, seq_length, d_model = x.size()
+        return x.view(batch_size, seq_length, self.num_heads, self.d_k).transpose(1, 2)
+        
+    def combine_heads(self, x):
+        batch_size, _, seq_length, d_k = x.size()
+        return x.transpose(1, 2).contiguous().view(batch_size, seq_length, self.d_model)
+        
+    def forward(self, Q, K, V, mask=None):
+        Q = self.split_heads(self.W_q(Q))
+        K = self.split_heads(self.W_k(K))
+        V = self.split_heads(self.W_v(V))
+        attn_output = self.scaled_dot_product_attention(Q, K, V, mask)
+        output = self.W_o(self.combine_heads(attn_output))
+        return output
+
+class PositionWiseFeedForward(nn.Module):
+    def __init__(self, d_model, d_ff):
+        super(PositionWiseFeedForward, self).__init__()
+        self.fc1 = nn.Linear(d_model, d_ff)
+        self.fc2 = nn.Linear(d_ff, d_model)
+        self.relu = nn.ReLU()
+
+    def forward(self, x):
+        return self.fc2(self.relu(self.fc1(x)))
+
+class PositionalEncoding(nn.Module):
+    def __init__(self, d_model, max_seq_length):
+        super(PositionalEncoding, self).__init__()
+        pe = torch.zeros(max_seq_length, d_model)
+        position = torch.arange(0, max_seq_length, dtype=torch.float).unsqueeze(1)
+        div_term = torch.exp(torch.arange(0, d_model, 2).float() * -(math.log(10000.0) / d_model))
+        pe[:, 0::2] = torch.sin(position * div_term)
+        pe[:, 1::2] = torch.cos(position * div_term)
+        self.register_buffer('pe', pe.unsqueeze(0))
+        
+    def forward(self, x):
+        return x + self.pe[:, :x.size(1)]
+
+class EncoderLayer(nn.Module):
+    def __init__(self, d_model, num_heads, d_ff, dropout):
+        super(EncoderLayer, self).__init__()
+        self.self_attn = MultiHeadAttention(d_model, num_heads)
+        self.feed_forward = PositionWiseFeedForward(d_model, d_ff)
+        self.norm1 = nn.LayerNorm(d_model)
+        self.norm2 = nn.LayerNorm(d_model)
+        self.dropout = nn.Dropout(dropout)
+        
+    def forward(self, x, mask):
+        attn_output = self.self_attn(x, x, x, mask)
+        x = self.norm1(x + self.dropout(attn_output))
+        ff_output = self.feed_forward(x)
+        x = self.norm2(x + self.dropout(ff_output))
+        return x
+
+class DecoderLayer(nn.Module):
+    def __init__(self, d_model, num_heads, d_ff, dropout):
+        super(DecoderLayer, self).__init__()
+        self.self_attn = MultiHeadAttention(d_model, num_heads)
+        self.cross_attn = MultiHeadAttention(d_model, num_heads)
+        self.feed_forward = PositionWiseFeedForward(d_model, d_ff)
+        self.norm1 = nn.LayerNorm(d_model)
+        self.norm2 = nn.LayerNorm(d_model)
+        self.norm3 = nn.LayerNorm(d_model)
+        self.dropout = nn.Dropout(dropout)
+        
+    def forward(self, x, enc_output, src_mask, tgt_mask):
+        attn_output = self.self_attn(x, x, x, tgt_mask)
+        x = self.norm1(x + self.dropout(attn_output))
+        attn_output = self.cross_attn(x, enc_output, enc_output, src_mask)
+        x = self.norm2(x + self.dropout(attn_output))
+        ff_output = self.feed_forward(x)
+        x = self.norm3(x + self.dropout(ff_output))
+        return x
+
+class Transformer(nn.Module):
+    def __init__(self, src_vocab_size, tgt_vocab_size, d_model, num_heads, num_layers, d_ff, max_seq_length, dropout):
+        super(Transformer, self).__init__()
+        self.encoder_embedding = nn.Embedding(src_vocab_size, d_model)
+        self.decoder_embedding = nn.Embedding(tgt_vocab_size, d_model)
+        self.positional_encoding = PositionalEncoding(d_model, max_seq_length)
+        self.encoder_layers = nn.ModuleList([EncoderLayer(d_model, num_heads, d_ff, dropout) for _ in range(num_layers)])
+        self.decoder_layers = nn.ModuleList([DecoderLayer(d_model, num_heads, d_ff, dropout) for _ in range(num_layers)])
+        self.fc = nn.Linear(d_model, tgt_vocab_size)
+        self.dropout = nn.Dropout(dropout)
+
+    def generate_mask(self, src, tgt):
+        src_mask = (src != 0).unsqueeze(1).unsqueeze(2)
+        tgt_mask = (tgt != 0).unsqueeze(1).unsqueeze(3)
+        seq_length = tgt.size(1)
+        nopeak_mask = (1 - torch.triu(torch.ones(1, seq_length, seq_length), diagonal=1)).bool()
+        tgt_mask = tgt_mask & nopeak_mask
+        return src_mask, tgt_mask
+
+    def forward(self, src, tgt):
+        src_mask, tgt_mask = self.generate_mask(src, tgt)
+        src_embedded = self.dropout(self.positional_encoding(self.encoder_embedding(src)))
+        tgt_embedded = self.dropout(self.positional_encoding(self.decoder_embedding(tgt)))
+        enc_output = src_embedded
+        for enc_layer in self.encoder_layers:
+            enc_output = enc_layer(enc_output, src_mask)
+        dec_output = tgt_embedded
+        for dec_layer in self.decoder_layers:
+            dec_output = dec_layer(dec_output, enc_output, src_mask, tgt_mask)
+        output = self.fc(dec_output)
+        return output
+
+# Set device
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
 # Load tokenizers
-sp_pseudo = spm.SentencePieceProcessor(model_file="pseudocode_tokenizer.model")  # For decoding pseudocode (target)
-sp_code = spm.SentencePieceProcessor(model_file="code_tokenizer.model")      # For encoding C++ (source)
+sp_pseudo = spm.SentencePieceProcessor(model_file="pseudo.model")  # For decoding pseudocode (target)
+sp_code = spm.SentencePieceProcessor(model_file="code.model")      # For encoding C++ (source)
 
 # Load the full saved model (architecture + weights)
-model_path = "code2pseudo.pth"
-model = torch.load(model_path, map_location=torch.device("cuda" if torch.cuda.is_available() else "cpu"), weights_only=False)
+model_path = "transformer_cpp_to_pseudo.pth"
+model = torch.load(model_path, map_location=device, weights_only=False)
 model.eval()
+model = model.to(device)
 
-# Function to generate pseudocode
 def generate_pseudocode(cpp_code, max_len):
+    """Generate pseudocode from C++ code with streaming output."""
     model.eval()
-    src = torch.tensor([sp_code.encode_as_ids(cpp_code)], dtype=torch.long)  # Tokenize C++ code
-    tgt = torch.tensor([[2]], dtype=torch.long)  # <bos_id>=2
+    src = torch.tensor([sp_code.encode_as_ids(cpp_code)], dtype=torch.long, device=device)  # Tokenize C++ code
+    tgt = torch.tensor([[2]], dtype=torch.long, device=device)  # <bos_id>=2
+    
     generated_tokens = [2]  # Start with <START>
-    for _ in range(max_len):
-        output = model(src, tgt)
-        next_token = output[:, -1, :].argmax(-1).item()
-        generated_tokens.append(next_token)
-        tgt = torch.cat([tgt, torch.tensor([[next_token]])], dim=1)
-        if next_token == 3:  # <END>=3
-            break
-    return sp_pseudo.decode_ids(generated_tokens)  # Final decoded output
-
-# Gradio interface
-demo = gr.Interface(
-    fn=generate_pseudocode,
-    inputs=[
-        gr.Textbox(placeholder="Enter C++ code here", label="C++ Code"),
-        gr.Slider(minimum=10, maximum=1000, value=50, step=1, label="Max Tokens")
-    ],
-    outputs=gr.Textbox(label="Generated Pseudocode"),
-    title="C++ to Pseudocode Converter",
-    description="Enter C++ code and get its pseudocode equivalent using a transformer model."
-)
+    response = ""
+    with torch.no_grad():
+        for _ in range(max_len):
+            output = model(src, tgt)
+            next_token = output[:, -1, :].argmax(-1).item()
+            generated_tokens.append(next_token)
+            tgt = torch.cat([tgt, torch.tensor([[next_token]], device=device)], dim=1)
+            response = sp_pseudo.decode_ids(generated_tokens)  # Decode to pseudocode
+            yield response  # Yield partial output
+            if next_token == 3:  # <END>=3 (adjust if your EOS ID differs)
+                break
+    yield response  # Final output
+
+def respond(message, history, max_tokens):
+    """Wrapper for Gradio interface."""
+    for response in generate_pseudocode(message, max_tokens):
+        yield response
+
+# Gradio UI setup with Blocks
+with gr.Blocks(title="C++ to Pseudocode Transformer") as demo:
+    gr.Markdown("## C++ to Pseudocode Converter")
+    gr.Markdown("Enter C++ code below and press Submit to generate pseudocode.")
+    cpp_input = gr.Textbox(
+        label="C++ Code",
+        placeholder="e.g., 'int x = 5; for(int i=0; i<x; i++) cout << i;'",
+        lines=5
+    )
+    submit_btn = gr.Button("Submit", variant="primary")
+    pseudocode_output = gr.Textbox(
+        label="Generated Pseudocode",
+        lines=5
+    )
+    
+    submit_btn.click(
+        fn=respond,
+        inputs=[cpp_input, gr.State(value=[]), gr.Slider(minimum=10, maximum=1000, value=50, step=1, visible=False)],
+        outputs=pseudocode_output
+    )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()