File size: 8,957 Bytes
368be8b 19cd814 f76e395 19cd814 8a0afd9 19cd814 94b0679 19cd814 a295347 19cd814 8146fe3 675b9a7 19cd814 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 |
import torch
import torch.nn as nn
from transformers import PreTrainedModel, PretrainedConfig
class MultiheadAttention(nn.Module):
def __init__(self, d_in, d_out, context_length, dropout, num_heads, qkv_bias=False):
super().__init__()
self.d_out = d_out
self.num_heads = num_heads
self.head_dim = d_out // num_heads
#step 3
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)
self.dropout = nn.Dropout(dropout)
self.register_buffer("mask",torch.triu(torch.ones(context_length, context_length), diagonal=1))
def forward(self, x):
b, num_tokens, d_in = x.shape
#step 4
keys = self.W_key(x)
queries = self.W_query(x)
values = self.W_value(x)
#step 5
keys = keys.view(b, num_tokens, self.num_heads, self.head_dim)
queries = queries.view(b, num_tokens, self.num_heads, self.head_dim)
values = values.view(b, num_tokens, self.num_heads, self.head_dim)
#step 6
keys = keys.transpose(1,2)
queries = queries.transpose(1,2)
values = values.transpose(1,2)
#step 7
attn_scores = queries @ keys.transpose(2,3)
#step 8
mask_bool = self.mask.bool()[:num_tokens, :num_tokens]
attn_scores.masked_fill_(mask_bool, -torch.inf)
attn_weights = torch.softmax(attn_scores / keys.shape[-1]**0.5, dim=-1)
attn_weights = self.dropout(attn_weights)
#step 9 - 11
ctx_vec = (attn_weights @ values).transpose(1, 2)
#step 12
ctx_vec = ctx_vec.contiguous().view(b, num_tokens, self.d_out)
ctx_vec = self.out_proj(ctx_vec)
return ctx_vec
#==========================================================================
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):
def __init__(self, cfg):
super().__init__()
self.att = MultiheadAttention(
d_in = cfg.emb_dim,
d_out = cfg.emb_dim,
context_length = cfg.context_length,
dropout = cfg.drop_rate,
num_heads = cfg.n_heads,
qkv_bias = cfg.qkv_bias
)
self.ff = FeedForward(cfg)
self.norm1 = LayerNorm(cfg.emb_dim)
self.norm2 = LayerNorm(cfg.emb_dim)
self.drop_shortcut = nn.Dropout(cfg.drop_rate)
def forward(self, x):
shortcut = x
x = self.norm1(x)
x = self.att(x)
x = self.drop_shortcut(x)
x = x + shortcut
shortcut = x
x = self.norm2(x)
x = self.ff(x)
x = self.drop_shortcut(x)
x = x + shortcut
return x
#==========================================================================
class CustomGPTConfig(PretrainedConfig):
model_type = "custom_gpt" # Unique identifier for the AutoClass
def __init__(self, context_length=1024, drop_rate=0.0, emb_dim=1024, n_heads=16, n_layers=24, qkv_bias=True, vocab_size=50257, **kwargs):
super().__init__(**kwargs)
self.context_length = context_length
self.drop_rate = drop_rate
self.emb_dim = emb_dim
self.n_heads = n_heads
self.n_layers = n_layers
self.qkv_bias = qkv_bias
self.vocab_size = vocab_size
#==========================================================================
class CustomGPT(
PreTrainedModel,
):
config_class = CustomGPTConfig
def __init__(self, config):
super().__init__(config)
self.tok_emb = nn.Embedding(config.vocab_size, config.emb_dim)
self.pos_emb = nn.Embedding(config.context_length, config.emb_dim)
self.drop_emb = nn.Dropout(config.drop_rate)
self.trf_blocks = nn.Sequential(
*[TransformerBlock(config) for _ in range(config.n_layers)]
)
self.final_norm = LayerNorm(config.emb_dim)
self.out_head = nn.Linear(config.emb_dim, config.vocab_size, bias=False)
def forward(self, x):
batch_size, seq_len = x.shape
tok_embeddings = self.tok_emb(x) #[2,4,768]
pos_embeddings = self.pos_emb(torch.arange(seq_len, device=x.device)) #[2,4,768]
x = tok_embeddings + pos_embeddings #[2,4,768]
x = self.drop_emb(x)
x = self.trf_blocks(x)
x = self.final_norm(x)
logits = self.out_head(x) #[2,4,50257]
return logits
def format_input(self, entry):
instruction_text = (
f"Below is an instruction that describes a task. "
f"Write a response that appropriately completes the request."
f"\n\n### Instruction:\n{entry['instruction']}"
)
# input_text = f"\n\n### Input:\n{entry['input']}" if entry["input"] else ""
input_text = f"\n\n### Input:\n{entry.get('input', '')}"
return instruction_text + input_text
def text_to_token_ids(self, text, tokenizer):
encoded = tokenizer.encode(text, allowed_special={'<|endoftext|>'})
encoded_tensor = torch.tensor(encoded).unsqueeze(0) # add batch dimension
return encoded_tensor
def token_ids_to_text(self, token_ids, tokenizer):
flat = token_ids.squeeze(0) # remove batch dimension
return tokenizer.decode(flat.tolist())
def generate(self, idx, max_new_tokens, context_size, temperature=0.0, top_k=None, eos_id=None):
for _ in range(max_new_tokens):
idx_cond = idx[:, -context_size:]
with torch.no_grad():
logits = self(idx_cond)
logits = logits[:, -1, :]
if top_k is not None:
# Keep only top_k values
top_logits, _ = torch.topk(logits, top_k)
min_val = top_logits[:, -1] # select the last element i.e., the smallest from each batch's output
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)
if idx_next == eos_id: # Stop generating early if end-of-sequence token is encountered and eos_id is specified
break
# 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
def generate_response(self, input_entry, tokenizer, temperature=0.0, topk=None):
current_device = next(self.parameters()).device
self.eval()
input_text = self.format_input(input_entry)
token_ids = self.generate(
idx=self.text_to_token_ids(input_text, tokenizer).to(current_device),
max_new_tokens=256,
context_size=1024,
temperature=temperature,
top_k=topk,
eos_id=50256
)
generated_text = self.token_ids_to_text(token_ids, tokenizer)
response_text = (
generated_text[len(input_text):]
.replace("### Response:", "")
.strip()
)
return response_text.strip()
|