Update modeling_gpt2l.py

modeling_gpt2l.py

@@ -169,11 +169,11 @@ class Attention(nn.Module):
             query, key, value = self.c_attn(hidden_states).split(self.split_size, dim=2)
 
         query = self.split_heads(query)
-        key = self.split_heads(key, k=True)
+        key = self.split_heads(key) # Dude what? @ the k=True
         value = self.split_heads(value)
         if layer_past is not None:
             past_key, past_value = layer_past[0].transpose(-2, -1), layer_past[1]  # transpose back cf below
-            key = torch.cat((past_key, key), dim=-1)
+            key = torch.cat((past_key, key), dim=-2) # this was dim=-1??? I'm trying to patch in flash attention and this is giving me TROUBLE
             value = torch.cat((past_value, value), dim=-2)
 
         if use_cache is True: