more sdpa updates

models/layers.py

@@ -12,27 +12,17 @@ ops = comfy.ops.manual_cast
 
 logpy = logging.getLogger(__name__)
 
+backends = []
+
 if version.parse(torch.__version__) >= version.parse("2.0.0"):
     SDP_IS_AVAILABLE = True
-    from torch.backends.cuda import SDPBackend, sdp_kernel
+    from torch.nn.attention import SDPBackend, sdpa_kernel
 
     BACKEND_MAP = {
-        SDPBackend.MATH: {
-            "enable_math": True,
-            "enable_flash": False,
-            "enable_mem_efficient": False,
-        },
-        SDPBackend.FLASH_ATTENTION: {
-            "enable_math": False,
-            "enable_flash": True,
-            "enable_mem_efficient": False,
-        },
-        SDPBackend.EFFICIENT_ATTENTION: {
-            "enable_math": False,
-            "enable_flash": False,
-            "enable_mem_efficient": True,
-        },
-        None: {"enable_math": True, "enable_flash": True, "enable_mem_efficient": True},
+        SDPBackend.MATH: [SDPBackend.MATH],
+        SDPBackend.FLASH_ATTENTION: [SDPBackend.FLASH_ATTENTION],
+        SDPBackend.EFFICIENT_ATTENTION: [SDPBackend.EFFICIENT_ATTENTION],
+        None: [SDPBackend.MATH, SDPBackend.FLASH_ATTENTION, SDPBackend.EFFICIENT_ATTENTION]
     }
 else:
     from contextlib import nullcontext
@@ -143,7 +133,9 @@ class TemporalAttention_Masked(nn.Module):
 
         q, k, v = map(lambda t: rearrange(t, "b n (h d) -> b h n d", h=h), (q, k, v))
 
-        with sdp_kernel(**BACKEND_MAP[self.backend]):
+        backends.extend(BACKEND_MAP[self.backend])
+
+        with sdpa_kernel(backends):
             # print("dispatching into backend", self.backend, "q/k/v shape: ", q.shape, k.shape, v.shape)
             out = F.scaled_dot_product_attention(
                 q, k, v, attn_mask=mask
@@ -244,7 +236,9 @@ class ReferenceAttention(nn.Module):
 
         q, k, v = map(lambda t: rearrange(t, "b n (h d) -> b h n d", h=h), (q, k, v))
 
-        with sdp_kernel(**BACKEND_MAP[self.backend]):
+        backends.extend(BACKEND_MAP[self.backend])
+
+        with sdpa_kernel(backends):
             # print("dispatching into backend", self.backend, "q/k/v shape: ", q.shape, k.shape, v.shape)
             out = F.scaled_dot_product_attention(
                 q, k, v, attn_mask=mask

nodes.py

@@ -37,7 +37,6 @@ class LoadLVCDModel:
     def loadmodel(self, model, precision, use_xformers):
         
         device = mm.get_torch_device()
-        print(device)
         offload_device = mm.unet_offload_device()
         dtype = {"bf16": torch.bfloat16, "fp16": torch.float16, "fp32": torch.float32}[precision]
         mm.soft_empty_cache()
@@ -60,7 +59,6 @@ class LoadLVCDModel:
         config = OmegaConf.load(config_path)
         config.model.params.drop_first_stage_model = False
         config.model.params.init_from_unet = False
-        print(config.model.params.conditioner_config.params.emb_models[0])
 
         if use_xformers:
             config.model.params.network_config.params.spatial_transformer_attn_type = 'softmax-xformers'
@@ -97,7 +95,7 @@ class LVCDSampler:
                 "LVCD_pipe": ("LVCDPIPE",),
                 "ref_images": ("IMAGE",),
                 "sketch_images": ("IMAGE",),
-                "num_frames": ("INT", {"default": 19, "min": 1, "max": 100, "step": 1}),
+                "num_frames": ("INT", {"default": 19, "min": 15, "max": 100, "step": 1}),
                 "num_steps": ("INT", {"default": 25, "min": 1, "max": 100, "step": 1}),
                 "fps_id": ("INT", {"default": 6, "min": 1, "max": 100, "step": 1}),
                 "motion_bucket_id": ("INT", {"default": 160, "min": 0, "max": 1000, "step": 1}),