import gradio as gr import torch import torch.nn as nn from torch.nn import functional as F import json import os # --- Model Hyperparameters (same as before) --- batch_size = 32 block_size = 8 n_embd = 32 n_head = 4 n_layer = 4 dropout = 0.0 max_iters = 3000 eval_interval = 300 learning_rate = 1e-2 eval_iters = 200 # --- Data Preparation & Vocabulary Creation --- file_path = 'dataset.jsonl' corpus = "" try: with open(file_path, 'r') as f: for line in f: data_point = json.loads(line) corpus += data_point['header'] + '\n' + data_point['formal_statement'] + '\n' except FileNotFoundError: print(f"Error: The file '{file_path}' was not found.") exit() except (json.JSONDecodeError, KeyError): print(f"Error: There was a problem parsing a line in '{file_path}'. Check for malformed JSON or missing keys.") exit() if not corpus: print("Error: The corpus is empty.") exit() chars = sorted(list(set(corpus))) vocab_size = len(chars) stoi = {ch: i for i, ch in enumerate(chars)} itos = {i: ch for i, ch in enumerate(chars)} encode = lambda s: [stoi.get(c, 0) for c in s] decode = lambda l: ''.join([itos[i] for i in l]) device = 'cuda' if torch.cuda.is_available() else 'cpu' # Split the data for training and validation data = torch.tensor(encode(corpus), dtype=torch.long) n = int(0.9 * len(data)) train_data = data[:n] val_data = data[n:] def get_batch(split): data = train_data if split == 'train' else val_data ix = torch.randint(len(data) - block_size, (batch_size,)) x = torch.stack([data[i:i + block_size] for i in ix]) y = torch.stack([data[i + 1:i + block_size + 1] for i in ix]) x, y = x.to(device), y.to(device) return x, y # --- Model Definition (same as before) --- class Head(nn.Module): def __init__(self, head_size): super().__init__() self.key = nn.Linear(n_embd, head_size, bias=False) self.query = nn.Linear(n_embd, head_size, bias=False) self.value = nn.Linear(n_embd, head_size, bias=False) self.register_buffer('tril', torch.tril(torch.ones(block_size, block_size))) self.dropout = nn.Dropout(dropout) def forward(self, x): B, T, C = x.shape k = self.key(x) q = self.query(x) wei = q @ k.transpose(-2, -1) * C**-0.5 wei = wei.masked_fill(self.tril[:T, :T] == 0, float('-inf')) wei = F.softmax(wei, dim=-1) self.dropout(wei) v = self.value(x) out = wei @ v return out class MultiHeadAttention(nn.Module): def __init__(self, num_heads, head_size): super().__init__() self.heads = nn.ModuleList([Head(head_size) for _ in range(num_heads)]) self.proj = nn.Linear(num_heads * head_size, n_embd) self.dropout = nn.Dropout(dropout) def forward(self, x): out = torch.cat([h(x) for h in self.heads], dim=-1) out = self.dropout(self.proj(out)) return out class FeedFoward(nn.Module): def __init__(self, n_embd): super().__init__() self.net = nn.Sequential( nn.Linear(n_embd, 4 * n_embd), nn.ReLU(), nn.Linear(4 * n_embd, n_embd), nn.Dropout(dropout), ) def forward(self, x): return self.net(x) class TransformerBlock(nn.Module): def __init__(self, n_embd, n_head): super().__init__() head_size = n_embd // n_head self.sa = MultiHeadAttention(n_head, head_size) self.ffwd = FeedFoward(n_embd) self.ln1 = nn.LayerNorm(n_embd) self.ln2 = nn.LayerNorm(n_embd) def forward(self, x): x = x + self.sa(self.ln1(x)) x = x + self.ffwd(self.ln2(x)) return x class LanguageModel(nn.Module): def __init__(self, vocab_size, block_size, n_embd, n_head, n_layer, dropout): super().__init__() self.token_embedding_table = nn.Embedding(vocab_size, n_embd) self.position_embedding_table = nn.Embedding(block_size, n_embd) self.blocks = nn.Sequential(*[TransformerBlock(n_embd, n_head) for _ in range(n_layer)]) self.ln_f = nn.LayerNorm(n_embd) self.lm_head = nn.Linear(n_embd, vocab_size) self.block_size = block_size self.vocab_size = vocab_size def forward(self, idx, targets=None): B, T = idx.shape tok_emb = self.token_embedding_table(idx) pos_emb = self.position_embedding_table(torch.arange(T, device=idx.device)) x = tok_emb + pos_emb x = self.blocks(x) x = self.ln_f(x) logits = self.lm_head(x) loss = None if targets is not None: B, T, C = logits.shape logits = logits.view(B * T, C) targets = targets.view(B * T) loss = F.cross_entropy(logits, targets) return logits, loss def generate(self, idx, max_new_tokens): for _ in range(max_new_tokens): idx_cond = idx[:, -self.block_size:] logits, loss = self(idx_cond) logits = logits[:, -1, :] probs = F.softmax(logits, dim=-1) idx_next = torch.multinomial(probs, num_samples=1) idx = torch.cat((idx, idx_next), dim=1) return idx # --- Training and Generation --- model = LanguageModel(vocab_size, block_size, n_embd, n_head, n_layer, dropout) m = model.to(device) # --- Check if a trained model exists, otherwise train a new one --- model_file = 'model.pt' if os.path.exists(model_file): print(f"Loading existing model from {model_file}") try: model.load_state_dict(torch.load(model_file, map_location=device)) except RuntimeError as e: print(f"Error loading model: {e}") print("Model file might be incompatible with current vocabulary. Retraining...") # If loading fails, fall through to training logic model.train() # Set back to train mode just in case else: print("No trained model found. Starting a new training session...") # Define a helper function for loss estimation @torch.no_grad() def estimate_loss(): out = {} model.eval() for split in ['train', 'val']: losses = torch.zeros(eval_iters) for k in range(eval_iters): X, Y = get_batch(split) logits, loss = model(X, Y) losses[k] = loss.item() out[split] = losses.mean() model.train() return out # The training loop optimizer = torch.optim.AdamW(model.parameters(), lr=learning_rate) for iter in range(max_iters): if iter % eval_interval == 0: losses = estimate_loss() print(f"step {iter}: train loss {losses['train']:.4f}, val loss {losses['val']:.4f}") xb, yb = get_batch('train') logits, loss = model(xb, yb) optimizer.zero_grad(set_to_none=True) loss.backward() optimizer.step() torch.save(m.state_dict(), model_file) print(f"Training complete. Model saved to {model_file}") model.eval() model.to(device) # --- Gradio UI & Inference function --- def generate_text_chat(message, history): prompt = message max_new_tokens = 50 encoded_prompt = [stoi.get(c, 0) for c in prompt] if not encoded_prompt: return "Prompt is empty or contains unknown characters." context = torch.tensor(encoded_prompt, dtype=torch.long, device=device).unsqueeze(0) generated_text_indices = model.generate(context, max_new_tokens=max_new_tokens) generated_text = decode(generated_text_indices[0].tolist()) return generated_text[len(prompt):] demo = gr.ChatInterface( fn=generate_text_chat, title="Tiny Language Model Chat", description="A simple character-level language model trained in PyTorch, now with a chat interface.", chatbot=gr.Chatbot(height="500px"), textbox=gr.Textbox(placeholder="Ask me anything...", container=False, scale=7), theme="soft", ) demo.launch()