mcmonkey4eva
commented
3 weeks ago Can you post a workflow that replicates this bug?
Open rking2981 opened 3 weeks ago
mcmonkey4eva
commented
3 weeks ago Can you post a workflow that replicates this bug?
Seems the model merge is giving the WARNING SHAPE MISMATCH error again. I looked into the merge_patcher file seen from other errors from back in March and see that the typo was corrected, so seems there is another issue somewhere.
Please look into this. Thanks.
WARNING SHAPE MISMATCH diffusion_model.output_blocks.8.0.out_layers.3.weight WEIGHT NOT MERGED torch.Size([320, 320, 3, 3]) != torch.Size([640, 640, 3, 3]) WARNING SHAPE MISMATCH diffusion_model.output_blocks.8.0.out_layers.3.bias WEIGHT NOT MERGED torch.Size([320]) != torch.Size([640]) WARNING SHAPE MISMATCH diffusion_model.output_blocks.8.0.skip_connection.weight WEIGHT NOT MERGED torch.Size([320, 640, 1, 1]) != torch.Size([640, 960, 1, 1]) WARNING SHAPE MISMATCH diffusion_model.output_blocks.8.0.skip_connection.bias WEIGHT NOT MERGED torch.Size([320]) != torch.Size([640]) 0%| | 0/20 [00:00<?, ?it/s]D:\ComfyUI\ComfyUI\comfy\ldm\modules\attention.py:407: UserWarning: 1Torch was not compiled with flash attention. (Triggered internally at ..\aten\src\ATen\native\transformers\cuda\sdp_utils.cpp:455.) out = torch.nn.functional.scaled_dot_product_attention(q, k, v, attn_mask=mask, dropout_p=0.0, is_causal=False) C:\Users\ryank\AppData\Local\Programs\Python\Python310\lib\site-packages\torchsde\_brownian\brownian_interval.py:608: UserWarning: Should have tb<=t1 but got tb=14.614644050598145 and t1=14.614643. warnings.warn(f"Should have {tb_name}<=t1 but got {tb_name}={tb} and t1={self._end}.")