1
0
mirror of https://github.com/osmarks/nanogpt-experiments.git synced 2024-12-18 14:10:28 +00:00

attempt a non-biased model, per few papers that cite this as working well

This commit is contained in:
Andrej Karpathy 2023-01-27 18:54:08 +00:00
parent f29a9ff5bf
commit 2892858ce7

View File

@ -22,15 +22,24 @@ def new_gelu(x):
"""
return 0.5 * x * (1.0 + torch.tanh(math.sqrt(2.0 / math.pi) * (x + 0.044715 * torch.pow(x, 3.0))))
class LayerNormNoBias(nn.Module):
def __init__(self, ndim):
super().__init__()
self.weight = nn.Parameter(torch.ones(ndim))
def forward(self, input):
return F.layer_norm(input, self.weight.shape, self.weight, None, 1e-5)
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)
self.c_attn = nn.Linear(config.n_embd, 3 * config.n_embd, bias=False)
# output projection
self.c_proj = nn.Linear(config.n_embd, config.n_embd)
self.c_proj = nn.Linear(config.n_embd, config.n_embd, bias=False)
# regularization
self.attn_dropout = nn.Dropout(config.dropout)
self.resid_dropout = nn.Dropout(config.dropout)
@ -65,8 +74,8 @@ class MLP(nn.Module):
def __init__(self, config):
super().__init__()
self.c_fc = nn.Linear(config.n_embd, 4 * config.n_embd)
self.c_proj = nn.Linear(4 * config.n_embd, config.n_embd)
self.c_fc = nn.Linear(config.n_embd, 4 * config.n_embd, bias=False)
self.c_proj = nn.Linear(4 * config.n_embd, config.n_embd, bias=False)
self.dropout = nn.Dropout(config.dropout)
def forward(self, x):
@ -80,9 +89,9 @@ class Block(nn.Module):
def __init__(self, config):
super().__init__()
self.ln_1 = nn.LayerNorm(config.n_embd)
self.ln_1 = LayerNormNoBias(config.n_embd)
self.attn = CausalSelfAttention(config)
self.ln_2 = nn.LayerNorm(config.n_embd)
self.ln_2 = LayerNormNoBias(config.n_embd)
self.mlp = MLP(config)
def forward(self, x):
@ -112,7 +121,7 @@ class GPT(nn.Module):
wpe = nn.Embedding(config.block_size, config.n_embd),
drop = nn.Dropout(config.dropout),
h = nn.ModuleList([Block(config) for _ in range(config.n_layer)]),
ln_f = nn.LayerNorm(config.n_embd),
ln_f = LayerNormNoBias(config.n_embd),
))
self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
# with weight tying when using torch.compile() some warnings get generated:
@ -242,7 +251,7 @@ class GPT(nn.Module):
decay = set()
no_decay = set()
whitelist_weight_modules = (torch.nn.Linear, )
blacklist_weight_modules = (torch.nn.LayerNorm, torch.nn.Embedding)
blacklist_weight_modules = (torch.nn.LayerNorm, LayerNormNoBias, torch.nn.Embedding)
for mn, m in self.named_modules():
for pn, p in m.named_parameters():
fpn = '%s.%s' % (mn, pn) if mn else pn # full param name