[GH-ISSUE #967] Stuck at "size mismatch for dec.conv_post.weight: copying a param with shape torch. Not sure if anything's working. #625

New issue

Closed

opened 2026-10-20 10:01:22 +00:00 by bsakaguchi · 1 comment

bsakaguchi commented

2026-10-20 10:01:22 +00:00

Owner

Originally created by @MastermindR on GitHub (Aug 9, 2023).
Original GitHub issue: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/967

size mismatch for enc_p.emb_phone.weight: copying a param with shape torch.Size([192, 256]) from checkpoint, the shape in current model is torch.Size([192, 768]).
size mismatch for dec.noise_convs.0.weight: copying a param with shape torch.Size([256, 1, 96]) from checkpoint, the shape in current model is torch.Size([256, 1, 80]).
size mismatch for dec.noise_convs.1.weight: copying a param with shape torch.Size([128, 1, 16]) from checkpoint, the shape in current model is torch.Size([128, 1, 8]).
size mismatch for dec.noise_convs.2.weight: copying a param with shape torch.Size([64, 1, 8]) from checkpoint, the shape in current model is torch.Size([64, 1, 4]).
size mismatch for dec.noise_convs.3.weight: copying a param with shape torch.Size([32, 1, 4]) from checkpoint, the shape in current model is torch.Size([32, 1, 1]).
size mismatch for dec.ups.0.weight_v: copying a param with shape torch.Size([512, 256, 16]) from checkpoint, the shape in current model is torch.Size([512, 256, 24]).
size mismatch for dec.ups.1.weight_v: copying a param with shape torch.Size([256, 128, 16]) from checkpoint, the shape in current model is torch.Size([256, 128, 20]).
size mismatch for dec.conv_post.weight: copying a param with shape torch.Size([1, 16, 7]) from checkpoint, the shape in current model is torch.Size([1, 32, 7]).

I'm using a Mangio RVC fork to run RVC on Mac, and I'm stalled at this point. I have no idea if I'm supposed to wait for something else to load or if I'm stuck. I don't know if it's supposed to take a long time because I don't know how to check if a process is running. I see questions are answered sparingly on here so I don't know if I'll get a response but yeah

Originally created by @MastermindR on GitHub (Aug 9, 2023). Original GitHub issue: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/967 size mismatch for enc_p.emb_phone.weight: copying a param with shape torch.Size([192, 256]) from checkpoint, the shape in current model is torch.Size([192, 768]). size mismatch for dec.noise_convs.0.weight: copying a param with shape torch.Size([256, 1, 96]) from checkpoint, the shape in current model is torch.Size([256, 1, 80]). size mismatch for dec.noise_convs.1.weight: copying a param with shape torch.Size([128, 1, 16]) from checkpoint, the shape in current model is torch.Size([128, 1, 8]). size mismatch for dec.noise_convs.2.weight: copying a param with shape torch.Size([64, 1, 8]) from checkpoint, the shape in current model is torch.Size([64, 1, 4]). size mismatch for dec.noise_convs.3.weight: copying a param with shape torch.Size([32, 1, 4]) from checkpoint, the shape in current model is torch.Size([32, 1, 1]). size mismatch for dec.ups.0.weight_v: copying a param with shape torch.Size([512, 256, 16]) from checkpoint, the shape in current model is torch.Size([512, 256, 24]). size mismatch for dec.ups.1.weight_v: copying a param with shape torch.Size([256, 128, 16]) from checkpoint, the shape in current model is torch.Size([256, 128, 20]). size mismatch for dec.conv_post.weight: copying a param with shape torch.Size([1, 16, 7]) from checkpoint, the shape in current model is torch.Size([1, 32, 7]). I'm using a Mangio RVC fork to run RVC on Mac, and I'm stalled at this point. I have no idea if I'm supposed to wait for something else to load or if I'm stuck. I don't know if it's supposed to take a long time because I don't know how to check if a process is running. I see questions are answered sparingly on here so I don't know if I'll get a response but yeah

bsakaguchi

2026-10-20 10:01:22 +00:00

closed this issue
added the
following up
label

bsakaguchi commented

2026-10-20 10:01:24 +00:00

Author

Owner

@RVC-Boss commented on GitHub (Aug 10, 2023):

This issue means you want to train a v2 model but use a v1 model as pretrained model.

@RVC-Boss commented on GitHub (Aug 10, 2023): This issue means you want to train a v2 model but use a v1 model as pretrained model.