disable neonpixel optimizations on M1 hardware (#414)

* disable neonpixel optimizations on M1 hardware

* fix typo that was causing random noise images on m1

ldm/modules/attention.py

@@ -1,9 +1,10 @@
-from inspect import isfunction
 import math
+from inspect import isfunction
+
 import torch
 import torch.nn.functional as F
-from torch import nn, einsum
 from einops import rearrange, repeat
+from torch import nn, einsum
 
 from ldm.modules.diffusionmodules.util import checkpoint
 
@@ -174,6 +175,7 @@ class CrossAttention(nn.Module):
         context = default(context, x)
         k = self.to_k(context)
         v = self.to_v(context)
+        device_type = x.device.type
         del context, x
 
         q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q, k, v))
@@ -188,7 +190,9 @@ class CrossAttention(nn.Module):
             sim.masked_fill_(~mask, max_neg_value)
             del mask
 
-        # attention, what we cannot get enough of, by halves
+        if device_type == 'mps':      #special case for M1 - disable neonsecret optimization
+            sim = sim.softmax(dim=-1)
+        else:
             sim[4:] = sim[4:].softmax(dim=-1)
             sim[:4] = sim[:4].softmax(dim=-1)
 
@@ -200,7 +204,8 @@ class CrossAttention(nn.Module):
 class BasicTransformerBlock(nn.Module):
     def __init__(self, dim, n_heads, d_head, dropout=0., context_dim=None, gated_ff=True, checkpoint=True):
         super().__init__()
-        self.attn1 = CrossAttention(query_dim=dim, heads=n_heads, dim_head=d_head, dropout=dropout)  # is a self-attention
+        self.attn1 = CrossAttention(query_dim=dim, heads=n_heads, dim_head=d_head,
+                                    dropout=dropout)  # is a self-attention
         self.ff = FeedForward(dim, dropout=dropout, glu=gated_ff)
         self.attn2 = CrossAttention(query_dim=dim, context_dim=context_dim,
                                     heads=n_heads, dim_head=d_head, dropout=dropout)  # is self-attn if context is none
@@ -228,6 +233,7 @@ class SpatialTransformer(nn.Module):
     Then apply standard transformer action.
     Finally, reshape to image
     """
+
     def __init__(self, in_channels, n_heads, d_head,
                  depth=1, dropout=0., context_dim=None):
         super().__init__()