| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  | # Copyright (c) Sebastian Raschka under Apache License 2.0 (see LICENSE.txt). | 
					
						
							|  |  |  | # Source for "Build a Large Language Model From Scratch" | 
					
						
							|  |  |  | #   - https://www.manning.com/books/build-a-large-language-model-from-scratch | 
					
						
							|  |  |  | # Code: https://github.com/rasbt/LLMs-from-scratch | 
					
						
							|  |  |  | # | 
					
						
							|  |  |  | # This file collects all the relevant code that we covered thus far | 
					
						
							|  |  |  | # throughout Chapters 2-5. | 
					
						
							|  |  |  | # This file can be run as a standalone script. | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | import numpy as np | 
					
						
							|  |  |  | import tiktoken | 
					
						
							|  |  |  | import torch | 
					
						
							|  |  |  | import torch.nn as nn | 
					
						
							| 
									
										
										
										
											2024-08-06 19:54:34 -05:00
										 |  |  | from torch.utils.data import Dataset, DataLoader | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  | 
 | 
					
						
							|  |  |  | ##################################### | 
					
						
							|  |  |  | # Chapter 2 | 
					
						
							|  |  |  | ##################################### | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | class GPTDatasetV1(Dataset): | 
					
						
							|  |  |  |     def __init__(self, txt, tokenizer, max_length, stride): | 
					
						
							|  |  |  |         self.input_ids = [] | 
					
						
							|  |  |  |         self.target_ids = [] | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Tokenize the entire text | 
					
						
							| 
									
										
										
										
											2024-06-08 14:57:34 -05:00
										 |  |  |         token_ids = tokenizer.encode(txt, allowed_special={"<|endoftext|>"}) | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  | 
 | 
					
						
							|  |  |  |         # Use a sliding window to chunk the book into overlapping sequences of max_length | 
					
						
							|  |  |  |         for i in range(0, len(token_ids) - max_length, stride): | 
					
						
							|  |  |  |             input_chunk = token_ids[i:i + max_length] | 
					
						
							|  |  |  |             target_chunk = token_ids[i + 1: i + max_length + 1] | 
					
						
							|  |  |  |             self.input_ids.append(torch.tensor(input_chunk)) | 
					
						
							|  |  |  |             self.target_ids.append(torch.tensor(target_chunk)) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     def __len__(self): | 
					
						
							|  |  |  |         return len(self.input_ids) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     def __getitem__(self, idx): | 
					
						
							|  |  |  |         return self.input_ids[idx], self.target_ids[idx] | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | def create_dataloader_v1(txt, batch_size=4, max_length=256, | 
					
						
							| 
									
										
										
										
											2024-06-20 00:36:46 +02:00
										 |  |  |                          stride=128, shuffle=True, drop_last=True, num_workers=0): | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  |     # Initialize the tokenizer | 
					
						
							|  |  |  |     tokenizer = tiktoken.get_encoding("gpt2") | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     # Create dataset | 
					
						
							|  |  |  |     dataset = GPTDatasetV1(txt, tokenizer, max_length, stride) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     # Create dataloader | 
					
						
							|  |  |  |     dataloader = DataLoader( | 
					
						
							| 
									
										
										
										
											2024-06-20 00:36:46 +02:00
										 |  |  |         dataset, batch_size=batch_size, shuffle=shuffle, drop_last=drop_last, num_workers=num_workers) | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  | 
 | 
					
						
							|  |  |  |     return dataloader | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | ##################################### | 
					
						
							|  |  |  | # Chapter 3 | 
					
						
							|  |  |  | ##################################### | 
					
						
							|  |  |  | class MultiHeadAttention(nn.Module): | 
					
						
							| 
									
										
										
										
											2024-05-18 17:03:36 -05:00
										 |  |  |     def __init__(self, d_in, d_out, context_length, dropout, num_heads, qkv_bias=False, disable_causal_mask=False): | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  |         super().__init__() | 
					
						
							|  |  |  |         assert d_out % num_heads == 0, "d_out must be divisible by n_heads" | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         self.d_out = d_out | 
					
						
							|  |  |  |         self.num_heads = num_heads | 
					
						
							|  |  |  |         self.head_dim = d_out // num_heads  # Reduce the projection dim to match desired output dim | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         self.W_query = nn.Linear(d_in, d_out, bias=qkv_bias) | 
					
						
							|  |  |  |         self.W_key = nn.Linear(d_in, d_out, bias=qkv_bias) | 
					
						
							|  |  |  |         self.W_value = nn.Linear(d_in, d_out, bias=qkv_bias) | 
					
						
							|  |  |  |         self.out_proj = nn.Linear(d_out, d_out)  # Linear layer to combine head outputs | 
					
						
							|  |  |  |         self.dropout = nn.Dropout(dropout) | 
					
						
							| 
									
										
										
										
											2024-05-18 17:03:36 -05:00
										 |  |  | 
 | 
					
						
							|  |  |  |         if not disable_causal_mask: | 
					
						
							|  |  |  |             self.register_buffer('mask', torch.triu(torch.ones(context_length, context_length), diagonal=1)) | 
					
						
							|  |  |  |         self.disable_causal_mask = disable_causal_mask | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  | 
 | 
					
						
							|  |  |  |     def forward(self, x): | 
					
						
							|  |  |  |         b, num_tokens, d_in = x.shape | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         keys = self.W_key(x)  # Shape: (b, num_tokens, d_out) | 
					
						
							|  |  |  |         queries = self.W_query(x) | 
					
						
							|  |  |  |         values = self.W_value(x) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # We implicitly split the matrix by adding a `num_heads` dimension | 
					
						
							|  |  |  |         # Unroll last dim: (b, num_tokens, d_out) -> (b, num_tokens, num_heads, head_dim) | 
					
						
							|  |  |  |         keys = keys.view(b, num_tokens, self.num_heads, self.head_dim) | 
					
						
							|  |  |  |         values = values.view(b, num_tokens, self.num_heads, self.head_dim) | 
					
						
							|  |  |  |         queries = queries.view(b, num_tokens, self.num_heads, self.head_dim) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Transpose: (b, num_tokens, num_heads, head_dim) -> (b, num_heads, num_tokens, head_dim) | 
					
						
							|  |  |  |         keys = keys.transpose(1, 2) | 
					
						
							|  |  |  |         queries = queries.transpose(1, 2) | 
					
						
							|  |  |  |         values = values.transpose(1, 2) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Compute scaled dot-product attention (aka self-attention) with a causal mask | 
					
						
							|  |  |  |         attn_scores = queries @ keys.transpose(2, 3)  # Dot product for each head | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2024-05-18 17:03:36 -05:00
										 |  |  |         if not self.disable_causal_mask: | 
					
						
							|  |  |  |             # Original mask truncated to the number of tokens and converted to boolean | 
					
						
							|  |  |  |             mask_bool = self.mask.bool()[:num_tokens, :num_tokens] | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2024-05-18 17:03:36 -05:00
										 |  |  |             # Use the mask to fill attention scores | 
					
						
							|  |  |  |             attn_scores.masked_fill_(mask_bool, -torch.inf) | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  | 
 | 
					
						
							|  |  |  |         attn_weights = torch.softmax(attn_scores / keys.shape[-1]**0.5, dim=-1) | 
					
						
							|  |  |  |         attn_weights = self.dropout(attn_weights) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Shape: (b, num_tokens, num_heads, head_dim) | 
					
						
							|  |  |  |         context_vec = (attn_weights @ values).transpose(1, 2) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Combine heads, where self.d_out = self.num_heads * self.head_dim | 
					
						
							|  |  |  |         context_vec = context_vec.reshape(b, num_tokens, self.d_out) | 
					
						
							|  |  |  |         context_vec = self.out_proj(context_vec)  # optional projection | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         return context_vec | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | ##################################### | 
					
						
							|  |  |  | # Chapter 4 | 
					
						
							|  |  |  | ##################################### | 
					
						
							|  |  |  | class LayerNorm(nn.Module): | 
					
						
							|  |  |  |     def __init__(self, emb_dim): | 
					
						
							|  |  |  |         super().__init__() | 
					
						
							|  |  |  |         self.eps = 1e-5 | 
					
						
							|  |  |  |         self.scale = nn.Parameter(torch.ones(emb_dim)) | 
					
						
							|  |  |  |         self.shift = nn.Parameter(torch.zeros(emb_dim)) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     def forward(self, x): | 
					
						
							|  |  |  |         mean = x.mean(dim=-1, keepdim=True) | 
					
						
							|  |  |  |         var = x.var(dim=-1, keepdim=True, unbiased=False) | 
					
						
							|  |  |  |         norm_x = (x - mean) / torch.sqrt(var + self.eps) | 
					
						
							|  |  |  |         return self.scale * norm_x + self.shift | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | class GELU(nn.Module): | 
					
						
							|  |  |  |     def __init__(self): | 
					
						
							|  |  |  |         super().__init__() | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     def forward(self, x): | 
					
						
							|  |  |  |         return 0.5 * x * (1 + torch.tanh( | 
					
						
							|  |  |  |             torch.sqrt(torch.tensor(2.0 / torch.pi)) * | 
					
						
							|  |  |  |             (x + 0.044715 * torch.pow(x, 3)) | 
					
						
							|  |  |  |         )) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | class FeedForward(nn.Module): | 
					
						
							|  |  |  |     def __init__(self, cfg): | 
					
						
							|  |  |  |         super().__init__() | 
					
						
							|  |  |  |         self.layers = nn.Sequential( | 
					
						
							|  |  |  |             nn.Linear(cfg["emb_dim"], 4 * cfg["emb_dim"]), | 
					
						
							|  |  |  |             GELU(), | 
					
						
							|  |  |  |             nn.Linear(4 * cfg["emb_dim"], cfg["emb_dim"]), | 
					
						
							|  |  |  |         ) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     def forward(self, x): | 
					
						
							|  |  |  |         return self.layers(x) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | class TransformerBlock(nn.Module): | 
					
						
							| 
									
										
										
										
											2024-05-18 17:03:36 -05:00
										 |  |  |     def __init__(self, cfg, disable_causal_mask=False): | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  |         super().__init__() | 
					
						
							|  |  |  |         self.att = MultiHeadAttention( | 
					
						
							|  |  |  |             d_in=cfg["emb_dim"], | 
					
						
							|  |  |  |             d_out=cfg["emb_dim"], | 
					
						
							|  |  |  |             context_length=cfg["context_length"], | 
					
						
							|  |  |  |             num_heads=cfg["n_heads"], | 
					
						
							|  |  |  |             dropout=cfg["drop_rate"], | 
					
						
							| 
									
										
										
										
											2024-05-18 17:03:36 -05:00
										 |  |  |             qkv_bias=cfg["qkv_bias"], | 
					
						
							|  |  |  |             disable_causal_mask=disable_causal_mask | 
					
						
							|  |  |  |         ) | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  |         self.ff = FeedForward(cfg) | 
					
						
							|  |  |  |         self.norm1 = LayerNorm(cfg["emb_dim"]) | 
					
						
							|  |  |  |         self.norm2 = LayerNorm(cfg["emb_dim"]) | 
					
						
							| 
									
										
										
										
											2024-04-28 14:31:27 -05:00
										 |  |  |         self.drop_shortcut = nn.Dropout(cfg["drop_rate"]) | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  | 
 | 
					
						
							|  |  |  |     def forward(self, x): | 
					
						
							|  |  |  |         # Shortcut connection for attention block | 
					
						
							|  |  |  |         shortcut = x | 
					
						
							|  |  |  |         x = self.norm1(x) | 
					
						
							|  |  |  |         x = self.att(x)   # Shape [batch_size, num_tokens, emb_size] | 
					
						
							| 
									
										
										
										
											2024-04-28 14:31:27 -05:00
										 |  |  |         x = self.drop_shortcut(x) | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  |         x = x + shortcut  # Add the original input back | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Shortcut connection for feed-forward block | 
					
						
							|  |  |  |         shortcut = x | 
					
						
							|  |  |  |         x = self.norm2(x) | 
					
						
							|  |  |  |         x = self.ff(x) | 
					
						
							| 
									
										
										
										
											2024-04-28 14:31:27 -05:00
										 |  |  |         x = self.drop_shortcut(x) | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  |         x = x + shortcut  # Add the original input back | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         return x | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | class GPTModel(nn.Module): | 
					
						
							| 
									
										
										
										
											2024-05-18 17:03:36 -05:00
										 |  |  |     def __init__(self, cfg, disable_causal_mask=False): | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  |         super().__init__() | 
					
						
							|  |  |  |         self.tok_emb = nn.Embedding(cfg["vocab_size"], cfg["emb_dim"]) | 
					
						
							|  |  |  |         self.pos_emb = nn.Embedding(cfg["context_length"], cfg["emb_dim"]) | 
					
						
							|  |  |  |         self.drop_emb = nn.Dropout(cfg["drop_rate"]) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         self.trf_blocks = nn.Sequential( | 
					
						
							| 
									
										
										
										
											2024-05-18 17:03:36 -05:00
										 |  |  |             *[TransformerBlock(cfg, disable_causal_mask) for _ in range(cfg["n_layers"])]) | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  | 
 | 
					
						
							|  |  |  |         self.final_norm = LayerNorm(cfg["emb_dim"]) | 
					
						
							|  |  |  |         self.out_head = nn.Linear(cfg["emb_dim"], cfg["vocab_size"], bias=False) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     def forward(self, in_idx): | 
					
						
							|  |  |  |         batch_size, seq_len = in_idx.shape | 
					
						
							|  |  |  |         tok_embeds = self.tok_emb(in_idx) | 
					
						
							|  |  |  |         pos_embeds = self.pos_emb(torch.arange(seq_len, device=in_idx.device)) | 
					
						
							|  |  |  |         x = tok_embeds + pos_embeds  # Shape [batch_size, num_tokens, emb_size] | 
					
						
							|  |  |  |         x = self.drop_emb(x) | 
					
						
							|  |  |  |         x = self.trf_blocks(x) | 
					
						
							|  |  |  |         x = self.final_norm(x) | 
					
						
							|  |  |  |         logits = self.out_head(x) | 
					
						
							|  |  |  |         return logits | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | def generate_text_simple(model, idx, max_new_tokens, context_size): | 
					
						
							|  |  |  |     # idx is (B, T) array of indices in the current context | 
					
						
							|  |  |  |     for _ in range(max_new_tokens): | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Crop current context if it exceeds the supported context size | 
					
						
							|  |  |  |         # E.g., if LLM supports only 5 tokens, and the context size is 10 | 
					
						
							|  |  |  |         # then only the last 5 tokens are used as context | 
					
						
							|  |  |  |         idx_cond = idx[:, -context_size:] | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Get the predictions | 
					
						
							|  |  |  |         with torch.no_grad(): | 
					
						
							|  |  |  |             logits = model(idx_cond) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Focus only on the last time step | 
					
						
							|  |  |  |         # (batch, n_token, vocab_size) becomes (batch, vocab_size) | 
					
						
							|  |  |  |         logits = logits[:, -1, :] | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Get the idx of the vocab entry with the highest logits value | 
					
						
							|  |  |  |         idx_next = torch.argmax(logits, dim=-1, keepdim=True)  # (batch, 1) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Append sampled index to the running sequence | 
					
						
							|  |  |  |         idx = torch.cat((idx, idx_next), dim=1)  # (batch, n_tokens+1) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return idx | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | ##################################### | 
					
						
							|  |  |  | # Chapter 5 | 
					
						
							|  |  |  | ##################################### | 
					
						
							|  |  |  | def assign(left, right): | 
					
						
							|  |  |  |     if left.shape != right.shape: | 
					
						
							|  |  |  |         raise ValueError(f"Shape mismatch. Left: {left.shape}, Right: {right.shape}") | 
					
						
							|  |  |  |     return torch.nn.Parameter(torch.tensor(right)) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | def load_weights_into_gpt(gpt, params): | 
					
						
							|  |  |  |     gpt.pos_emb.weight = assign(gpt.pos_emb.weight, params['wpe']) | 
					
						
							|  |  |  |     gpt.tok_emb.weight = assign(gpt.tok_emb.weight, params['wte']) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     for b in range(len(params["blocks"])): | 
					
						
							|  |  |  |         q_w, k_w, v_w = np.split( | 
					
						
							|  |  |  |             (params["blocks"][b]["attn"]["c_attn"])["w"], 3, axis=-1) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].att.W_query.weight = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].att.W_query.weight, q_w.T) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].att.W_key.weight = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].att.W_key.weight, k_w.T) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].att.W_value.weight = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].att.W_value.weight, v_w.T) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         q_b, k_b, v_b = np.split( | 
					
						
							|  |  |  |             (params["blocks"][b]["attn"]["c_attn"])["b"], 3, axis=-1) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].att.W_query.bias = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].att.W_query.bias, q_b) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].att.W_key.bias = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].att.W_key.bias, k_b) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].att.W_value.bias = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].att.W_value.bias, v_b) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         gpt.trf_blocks[b].att.out_proj.weight = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].att.out_proj.weight, | 
					
						
							|  |  |  |             params["blocks"][b]["attn"]["c_proj"]["w"].T) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].att.out_proj.bias = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].att.out_proj.bias, | 
					
						
							|  |  |  |             params["blocks"][b]["attn"]["c_proj"]["b"]) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         gpt.trf_blocks[b].ff.layers[0].weight = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].ff.layers[0].weight, | 
					
						
							|  |  |  |             params["blocks"][b]["mlp"]["c_fc"]["w"].T) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].ff.layers[0].bias = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].ff.layers[0].bias, | 
					
						
							|  |  |  |             params["blocks"][b]["mlp"]["c_fc"]["b"]) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].ff.layers[2].weight = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].ff.layers[2].weight, | 
					
						
							|  |  |  |             params["blocks"][b]["mlp"]["c_proj"]["w"].T) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].ff.layers[2].bias = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].ff.layers[2].bias, | 
					
						
							|  |  |  |             params["blocks"][b]["mlp"]["c_proj"]["b"]) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         gpt.trf_blocks[b].norm1.scale = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].norm1.scale, | 
					
						
							|  |  |  |             params["blocks"][b]["ln_1"]["g"]) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].norm1.shift = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].norm1.shift, | 
					
						
							|  |  |  |             params["blocks"][b]["ln_1"]["b"]) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].norm2.scale = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].norm2.scale, | 
					
						
							|  |  |  |             params["blocks"][b]["ln_2"]["g"]) | 
					
						
							|  |  |  |         gpt.trf_blocks[b].norm2.shift = assign( | 
					
						
							|  |  |  |             gpt.trf_blocks[b].norm2.shift, | 
					
						
							|  |  |  |             params["blocks"][b]["ln_2"]["b"]) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     gpt.final_norm.scale = assign(gpt.final_norm.scale, params["g"]) | 
					
						
							|  |  |  |     gpt.final_norm.shift = assign(gpt.final_norm.shift, params["b"]) | 
					
						
							|  |  |  |     gpt.out_head.weight = assign(gpt.out_head.weight, params["wte"]) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2024-05-19 09:04:49 -05:00
										 |  |  | def generate(model, idx, max_new_tokens, context_size, temperature=0.0, top_k=None, eos_id=None): | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  |     # For-loop is the same as before: Get logits, and only focus on last time step | 
					
						
							|  |  |  |     for _ in range(max_new_tokens): | 
					
						
							|  |  |  |         idx_cond = idx[:, -context_size:] | 
					
						
							|  |  |  |         with torch.no_grad(): | 
					
						
							|  |  |  |             logits = model(idx_cond) | 
					
						
							|  |  |  |         logits = logits[:, -1, :] | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # New: Filter logits with top_k sampling | 
					
						
							|  |  |  |         if top_k is not None: | 
					
						
							|  |  |  |             # Keep only top_k values | 
					
						
							|  |  |  |             top_logits, _ = torch.topk(logits, top_k) | 
					
						
							|  |  |  |             min_val = top_logits[:, -1] | 
					
						
							|  |  |  |             logits = torch.where(logits < min_val, torch.tensor(float('-inf')).to(logits.device), logits) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # New: Apply temperature scaling | 
					
						
							|  |  |  |         if temperature > 0.0: | 
					
						
							|  |  |  |             logits = logits / temperature | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |             # Apply softmax to get probabilities | 
					
						
							|  |  |  |             probs = torch.softmax(logits, dim=-1)  # (batch_size, context_len) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |             # Sample from the distribution | 
					
						
							|  |  |  |             idx_next = torch.multinomial(probs, num_samples=1)  # (batch_size, 1) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         # Otherwise same as before: get idx of the vocab entry with the highest logits value | 
					
						
							|  |  |  |         else: | 
					
						
							|  |  |  |             idx_next = torch.argmax(logits, dim=-1, keepdim=True)  # (batch_size, 1) | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2024-05-18 12:35:40 -05:00
										 |  |  |         if idx_next == eos_id:  # Stop generating early if end-of-sequence token is encountered and eos_id is specified | 
					
						
							|  |  |  |             break | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2024-04-23 09:51:52 -05:00
										 |  |  |         # Same as before: append sampled index to the running sequence | 
					
						
							|  |  |  |         idx = torch.cat((idx, idx_next), dim=1)  # (batch_size, num_tokens+1) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return idx |