omit the prefix sections of the sequence undergoing full attention from the cross entropy loss

Merge pull request #20 from lucidrains/prefix-full-attention

omit the prefix sections of the sequence undergoing full attention fr…

Author: lucidrains

dalle_pytorch/dalle_pytorch.py

@@ -256,6 +256,7 @@ def __init__(
         ff_dropout = 0,
         sparse_attn = False,
         noncausal_attn_len = 0,
+        ignore_index = -100
     ):
         super().__init__()
         assert isinstance(vae, DiscreteVAE), 'vae must be an instance of DiscreteVAE'
@@ -279,7 +280,9 @@ def __init__(
         seq_len = text_seq_len + image_seq_len
         total_tokens = num_text_tokens + num_image_tokens + 1 # extra for EOS
         self.total_tokens = total_tokens
-        
+
+        self.noncausal_attn_len = noncausal_attn_len
+
         self.vae = vae
         if exists(self.vae):
             self.vae = vae
@@ -319,6 +322,8 @@ def __init__(
 
         self.register_buffer('logits_mask', logits_mask)
 
+        self.ignore_index = ignore_index
+
     @torch.no_grad()
     @eval_decorator
     def generate_images(
@@ -404,9 +409,15 @@ def forward(
             return logits
 
         assert exists(image), 'when training, image must be supplied'
-
+        noncausal_attn_len = self.noncausal_attn_len
         offsetted_image = image + self.num_text_tokens
         labels = torch.cat((text, offsetted_image), dim = 1)
+
+        if noncausal_attn_len > 0:
+            mask = torch.arange(seq_len, device = device)
+            mask = mask < noncausal_attn_len
+            labels.masked_fill_(mask[None, :], -100) # -100 is the ignore index for cross entropy loss
+
         labels = F.pad(labels, (0, 1), value = eos_token_id) # last token predicts EOS
         loss = F.cross_entropy(rearrange(logits, 'b n c -> b c n'), labels[:, 1:])
         return loss

setup.py

@@ -3,7 +3,7 @@
 setup(
   name = 'dalle-pytorch',
   packages = find_packages(),
-  version = '0.0.41',
+  version = '0.0.43',
   license='MIT',
   description = 'DALL-E - Pytorch',
   author = 'Phil Wang',