train.py

import os
import math
import time
import inspect
from dataclasses import dataclass
import torch
import torch.nn as nn
from torch.nn import functional as F
import tiktoken
import numpy as np
from huggingface_hub import HfApi, Repository
import gradio as gr
from tqdm import tqdm


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 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 # max sequence length
    vocab_size: int = 50257 # number of tokens: 50,000 BPE merges + 256 bytes tokens + 1 <|endoftext|> token
    n_layer: int = 12 # number of layers
    n_head: int = 12 # number of heads
    n_embd: int = 768 # embedding dimension


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)

        # weight sharing
        self.transformer.wte.weight = self.lm_head.weight

        # weight initialization
        self.apply(self._init_weights)

    def generate(self, idx, max_new_tokens):
        # idx is (B, T) array of indices in the current context
        for _ in range(max_new_tokens):
            # crop idx to the last block_size tokens
            idx_cond = idx[:, -self.config.block_size:]
            # get the predictions
            logits, loss = self(idx_cond)
            # focus only on the last time step
            logits = logits[:, -1, :] # becomes (B, C)
            # apply softmax to get probabilities
            probs = F.softmax(logits, dim=-1) # (B, C)
            # sample from the distribution
            idx_next = torch.multinomial(probs, num_samples=1) # (B, 1)
            # append sampled index to the running sequence
            idx = torch.cat((idx, idx_next), dim=1) # (B, T+1)
        return idx

    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):
        # idx is of shape (B, T)
        B, T = idx.size()
        assert T <= self.config.block_size, f"Cannot forward sequence of length {T}, block size is only {self.config.block_size}"
        # forward the token and posisition embeddings
        pos = torch.arange(0, T, dtype=torch.long, device=idx.device) # shape (T)
        pos_emb = self.transformer.wpe(pos) # position embeddings of shape (T, n_embd)
        tok_emb = self.transformer.wte(idx) # token embeddings of shape (B, T, n_embd)
        x = tok_emb + pos_emb
        # forward the blocks of the transformer
        for block in self.transformer.h:
            x = block(x)
        # forward the final layernorm and the classifier
        x = self.transformer.ln_f(x)
        logits = self.lm_head(x) # (B, T, vocab_size)
        loss = None
        if targets is not None:
            loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets.view(-1))
        return logits, loss

    @classmethod
    def from_pretrained(cls, model_type):
        """Loads pretrained GPT-2 model weights from huggingface"""
        assert model_type in {'gpt2', 'gpt2-medium', 'gpt2-large', 'gpt2-xl'}
        from transformers import GPT2LMHeadModel
        print("loading weights from pretrained gpt: %s" % model_type)

        # n_layer, n_head and n_embd are determined from model_type
        config_args = {
            'gpt2':         dict(n_layer=12, n_head=12, n_embd=768),  # 124M params
            'gpt2-medium':  dict(n_layer=24, n_head=16, n_embd=1024), # 350M params
            'gpt2-large':   dict(n_layer=36, n_head=20, n_embd=1280), # 774M params
            'gpt2-xl':      dict(n_layer=48, n_head=25, n_embd=1600), # 1558M params
        }[model_type]
        config_args['vocab_size'] = 50257 # always 50257 for GPT model checkpoints
        config_args['block_size'] = 1024 # always 1024 for GPT model checkpoints
        # create a from-scratch initialized minGPT model
        config = GPTConfig(**config_args)
        model = GPT(config)
        sd = model.state_dict()
        sd_keys = sd.keys()
        sd_keys = [k for k in sd_keys if not k.endswith('.attn.bias')] # discard this mask / buffer, not a param

        # init a huggingface/transformers model
        model_hf = GPT2LMHeadModel.from_pretrained(model_type)
        sd_hf = model_hf.state_dict()

        # copy while ensuring all of the parameters are aligned and match in names and shapes
        sd_keys_hf = sd_hf.keys()
        sd_keys_hf = [k for k in sd_keys_hf if not k.endswith('.attn.masked_bias')] # ignore these, just a buffer
        sd_keys_hf = [k for k in sd_keys_hf if not k.endswith('.attn.bias')] # same, just the mask (buffer)
        transposed = ['attn.c_attn.weight', 'attn.c_proj.weight', 'mlp.c_fc.weight', 'mlp.c_proj.weight']
        # basically the openai checkpoints use a "Conv1D" module, but we only want to use a vanilla Linear
        # this means that we have to transpose these weights when we import them
        assert len(sd_keys_hf) == len(sd_keys), f"mismatched keys: {len(sd_keys_hf)} != {len(sd_keys)}"
        for k in sd_keys_hf:
            if any(k.endswith(w) for w in transposed):
                # special treatment for the Conv1D weights we need to transpose
                assert sd_hf[k].shape[::-1] == sd[k].shape
                with torch.no_grad():
                    sd[k].copy_(sd_hf[k].t())
            else:
                # vanilla copy over the other parameters
                assert sd_hf[k].shape == sd[k].shape
                with torch.no_grad():
                    sd[k].copy_(sd_hf[k])

        return model

# model = GPT.from_pretrained('gpt2')

device = 'cpu'
if torch.cuda.is_available():
    device = 'cuda'
elif hasattr(torch.backends, "mps") and torch.backends.mps.is_available():
    device = "mps"
print(f"using device: {device}")

# SEED
torch.manual_seed(1337)
if torch.cuda.is_available():
    torch.cuda.manual_seed(1337)

# STOP
num_return_sequences = 5
max_length = 30


class DataLoaderLite:
    def __init__(self, B, T):
        self.B = B
        self.T = T

        # at init load tokens from disk and store them in memory
        with open('/content/drive/My Drive/ERAV3/Assign12/input.txt', 'r') as f:
            text = f.read()
        enc = tiktoken.get_encoding('gpt2')
        tokens = enc.encode(text)
        self.tokens = torch.tensor(tokens)
        print(f'loaded {len(self.tokens)} tokens')
        print(f'1 epoch = {len(self.tokens) // (B * T)} batches')

        # state
        self.current_position = 0

    def next_batch(self):
        B, T = self.B, self.T
        buf = self.tokens[self.current_position: self.current_position + B * T + 1]
        x = (buf[:-1]).view(B, T) # inputs
        y = (buf[1:]).view(B, T) # targets
        # advance the position in the tensor
        self.current_position += B*T
        # if loading the next batch would be out of bounds, reset
        if self.current_position + (B * T + 1) > len(self.tokens):
            self.current_position = 0
        return x, y


model = GPT(GPTConfig())
model.to(device)

train_loader = DataLoaderLite(B = 4, T = 32)

# Calculate number of epochs
total_tokens = len(train_loader.tokens)
batches_per_epoch = total_tokens // (4 * 32)
total_epochs = 5000 / batches_per_epoch
print(f'\nTraining for approximately {total_epochs:.2f} epochs')
print(f'Total tokens: {total_tokens:,}')
print(f'Batches per epoch: {batches_per_epoch}')
print(f'Total steps: 5,000\n')

# Continue with training loop
optimizer = torch.optim.AdamW(model.parameters(), lr = 3e-4)

# Calculate total epochs and steps per epoch
total_steps = 5000
steps_per_epoch = batches_per_epoch
num_epochs = total_steps // steps_per_epoch
remaining_steps = total_steps % steps_per_epoch

print(f"Training for {num_epochs} full epochs plus {remaining_steps} steps")
print(f"Steps per epoch: {steps_per_epoch}\n")

step = 0
for epoch in range(num_epochs + 1):
    # Determine steps for this epoch
    if epoch == num_epochs:
        if remaining_steps == 0:
            break
        current_steps = remaining_steps
    else:
        current_steps = steps_per_epoch
    
    print(f"\nEpoch {epoch+1}/{num_epochs + (1 if remaining_steps > 0 else 0)}")
    epoch_loss = 0
    
    # Use tqdm for progress bar
    pbar = tqdm(range(current_steps), desc=f'Training', 
                leave=True, ncols=100)
    
    for i in pbar:
        x, y = train_loader.next_batch()
        x, y = x.to(device), y.to(device)
        
        optimizer.zero_grad()
        logits, loss = model(x, y)
        loss.backward()
        optimizer.step()
        
        epoch_loss += loss.item()
        step += 1
        
        # Update progress bar description with current loss
        pbar.set_description(f'Loss: {loss.item():.4f}')
    
    # Print epoch summary
    avg_epoch_loss = epoch_loss / current_steps
    print(f'\nEpoch {epoch+1} completed. Average Loss: {avg_epoch_loss:.4f}')
    print(f'Total steps completed: {step}/{total_steps}')

# For even smaller file size, quantize the model to 8-bit
model_save_path = '/content/drive/My Drive/ERAV3/Assign12/gpt_model_quantized.pt'
try:
    # Quantize weights to 8-bit
    state_dict = model.state_dict()
    quantized_dict = {}
    
    for key, param in state_dict.items():
        if param.dtype == torch.float32 or param.dtype == torch.float16:
            # Quantize to 8-bit
            param_np = param.cpu().numpy()
            scale = np.max(np.abs(param_np)) / 127
            quantized = np.round(param_np / scale).astype(np.int8)
            quantized_dict[key] = {
                'data': quantized,
                'scale': scale
            }
        else:
            quantized_dict[key] = param

    # Save quantized weights
    torch.save(quantized_dict, model_save_path)
    print(f'\nQuantized model saved successfully to {model_save_path}')
except Exception as e:
    print(f'\nError saving model: {e}')

# To load the quantized model:
# def dequantize_model(model, quantized_dict):
#     state_dict = {}
#     for key, value in quantized_dict.items():
#         if isinstance(value, dict):
#             # Dequantize
#             state_dict[key] = torch.tensor(
#                 value['data'].astype(np.float32) * value['scale']
#             )
#         else:
#             state_dict[key] = value
#     model.load_state_dict(state_dict)
#     return model

context = torch.zeros((1, 1), dtype=torch.long, device=device)
enc = tiktoken.get_encoding('gpt2')
print(enc.decode(model.generate(context, max_new_tokens=500)[0].tolist()))