app.py

"""
Shakespeare Text Generator - Hugging Face Gradio App
=====================================================
A GPT model trained on Shakespeare's works to generate text in Shakespearean style.
"""

import gradio as gr
import torch
import torch.nn as nn
from torch.nn import functional as F
import math
from dataclasses import dataclass
import tiktoken

# ============================================================================
# MODEL ARCHITECTURE (Same as training)
# ============================================================================

class CausalSelfAttention(nn.Module):
    def __init__(self, config):
        super().__init__()
        assert config.n_embd % config.n_head == 0
        self.c_attn = nn.Linear(config.n_embd, 3 * config.n_embd)
        self.c_proj = nn.Linear(config.n_embd, config.n_embd)
        self.c_proj.NANGPT_SCALE_INIT = 1
        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()
        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)
        q = q.view(B, T, self.n_head, C // self.n_head).transpose(1, 2)
        v = v.view(B, T, self.n_head, C // self.n_head).transpose(1, 2)
        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
        y = y.transpose(1, 2).contiguous().view(B, T, C)
        y = self.c_proj(y)
        return y

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 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:
    block_size: int = 1024
    vocab_size: int = 50257
    n_layer: int = 12
    n_head: int = 12
    n_embd: int = 768

class GPT(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.config = config
        self.transformer = nn.ModuleDict(dict(
            wte = nn.Embedding(config.vocab_size, config.n_embd),
            wpe = nn.Embedding(config.block_size, config.n_embd),
            h = nn.ModuleList([Block(config) for _ in range(config.n_layer)]),
            ln_f = nn.LayerNorm(config.n_embd),
        ))
        self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
        self.transformer.wte.weight = self.lm_head.weight
        self.apply(self._init_weights)

    def _init_weights(self, module):
        if isinstance(module, nn.Linear):
            std = 0.02
            if hasattr(module, 'NANGPT_SCALE_INIT'):
                std *= (2 * self.config.n_layer) ** -0.5
            torch.nn.init.normal_(module.weight, mean=0.0, std=std)
            if module.bias is not None:
                torch.nn.init.zeros_(module.bias)
        elif isinstance(module, nn.Embedding):
            torch.nn.init.normal_(module.weight, mean=0.0, std=0.02)

    def forward(self, idx, targets=None):
        B, T = idx.size()
        assert T <= self.config.block_size
        pos = torch.arange(0, T, dtype=torch.long, device=idx.device)
        pos_emb = self.transformer.wpe(pos)
        tok_emb = self.transformer.wte(idx)
        x = tok_emb + pos_emb
        for block in self.transformer.h:
            x = block(x)
        x = self.transformer.ln_f(x)
        logits = self.lm_head(x)
        loss = None
        if targets is not None:
            loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets.view(-1))
        return logits, loss

# ============================================================================
# LOAD MODEL
# ============================================================================

print("Loading model...")
device = 'cuda' if torch.cuda.is_available() else 'cpu'
print(f"Using device: {device}")

model = GPT(GPTConfig())
# Load your trained checkpoint here
checkpoint = torch.load('shakespeare_gpt_fp16.pt', map_location=device)
model.load_state_dict({k: v.float() for k, v in checkpoint.items()})
model.to(device)
model.eval()

enc = tiktoken.get_encoding('gpt2')

# ============================================================================
# GENERATION FUNCTION
# ============================================================================

def generate_shakespeare(prompt, max_length=100, temperature=0.8, top_k=50, num_samples=1):
    """
    Generate Shakespeare-style text from a prompt.
    
    Args:
        prompt: Starting text
        max_length: Maximum number of tokens to generate
        temperature: Sampling temperature (higher = more random)
        top_k: Number of top tokens to sample from
        num_samples: Number of different samples to generate
    """
    if not prompt.strip():
        return "Please enter a prompt to generate text."
    
    try:
        # Encode the prompt
        tokens = enc.encode(prompt)
        if len(tokens) == 0:
            return "Invalid prompt. Please try again."
        
        outputs = []
        
        for _ in range(num_samples):
            x = torch.tensor(tokens, dtype=torch.long, device=device).unsqueeze(0)
            
            with torch.no_grad():
                for _ in range(max_length):
                    # Forward pass
                    logits = model(x)[0]
                    logits = logits[:, -1, :] / temperature
                    
                    # Top-k sampling
                    probs = F.softmax(logits, dim=-1)
                    topk_probs, topk_indices = torch.topk(probs, min(top_k, probs.size(-1)), dim=-1)
                    ix = torch.multinomial(topk_probs, 1)
                    xcol = torch.gather(topk_indices, -1, ix)
                    x = torch.cat((x, xcol), dim=1)
                    
                    # Stop if we exceed block size
                    if x.size(1) >= model.config.block_size:
                        break
            
            # Decode the output
            output_tokens = x[0].tolist()
            generated_text = enc.decode(output_tokens)
            outputs.append(generated_text)
        
        # Return all samples separated by dividers
        if num_samples == 1:
            return outputs[0]
        else:
            return "\n\n" + "="*60 + "\n\n".join(outputs)
    
    except Exception as e:
        return f"Error generating text: {str(e)}"

# ============================================================================
# GRADIO INTERFACE
# ============================================================================

# Create the interface
with gr.Blocks() as demo:
    gr.Markdown(
        """
        # 🎭 Shakespeare Text Generator
        
        Generate text in the style of William Shakespeare using a GPT model trained on his complete works.
        Enter a prompt and watch the Bard's AI apprentice continue the story!
        
        **Model Details**: GPT-2 124M architecture trained on Shakespeare's plays and sonnets (Loss: 0.095)
        """
    )
    
    with gr.Row():
        with gr.Column(scale=1):
            prompt_input = gr.Textbox(
                label="📝 Enter Your Prompt",
                placeholder="To be or not to be...",
                lines=4,
                value="To be or not to be"
            )
            
            with gr.Accordion("⚙️ Advanced Settings", open=False):
                max_length_slider = gr.Slider(
                    minimum=20,
                    maximum=300,
                    value=100,
                    step=10,
                    label="Max Length (tokens)",
                    info="Maximum number of tokens to generate"
                )
                
                temperature_slider = gr.Slider(
                    minimum=0.1,
                    maximum=1.5,
                    value=0.8,
                    step=0.1,
                    label="Temperature",
                    info="Higher = more creative, Lower = more focused"
                )
                
                top_k_slider = gr.Slider(
                    minimum=10,
                    maximum=100,
                    value=50,
                    step=10,
                    label="Top-K",
                    info="Number of top tokens to sample from"
                )
                
                num_samples_slider = gr.Slider(
                    minimum=1,
                    maximum=3,
                    value=1,
                    step=1,
                    label="Number of Samples",
                    info="Generate multiple variations"
                )
            
            generate_btn = gr.Button("🎨 Generate", variant="primary", size="lg")
        
        with gr.Column(scale=1):
            output_text = gr.Textbox(
                label="📜 Generated Text",
                lines=15
            )
    
    # Examples
    gr.Markdown("### 📚 Try These Examples:")
    gr.Examples(
        examples=[
            ["To be or not to be", 100, 0.8, 50, 1],
            ["What's in a name?", 120, 0.7, 40, 1],
            ["All the world's a stage", 150, 0.9, 50, 1],
            ["Romeo, Romeo, wherefore art thou", 100, 0.8, 50, 1],
            ["Friends, Romans, countrymen", 130, 0.75, 45, 1],
            ["Now is the winter of our discontent", 110, 0.85, 50, 1],
        ],
        inputs=[prompt_input, max_length_slider, temperature_slider, top_k_slider, num_samples_slider],
        outputs=output_text,
        fn=generate_shakespeare,
        cache_examples=False
    )
    
    # Connect the button
    generate_btn.click(
        fn=generate_shakespeare,
        inputs=[prompt_input, max_length_slider, temperature_slider, top_k_slider, num_samples_slider],
        outputs=output_text
    )
    
    gr.Markdown(
        """
        ---
        ### 💡 Tips for Best Results:
        - Start with famous Shakespeare quotes for coherent continuations
        - Use **lower temperature** (0.5-0.7) for more focused, coherent text
        - Use **higher temperature** (0.9-1.2) for more creative, diverse outputs
        - Adjust **Top-K** to control vocabulary diversity
        - Try generating multiple samples to see different variations
        
        ### ⚠️ Note:
        This model was trained on Shakespeare's works and will generate text in Early Modern English style.
        Results may vary based on the prompt and parameters.
        """
    )

# ============================================================================
# LAUNCH
# ============================================================================

if __name__ == "__main__":
    demo.launch(server_port=7860, share=True)