fix onnx
This commit is contained in:
parent
d2ea3f1930
commit
e65275cbc1
@ -4,7 +4,6 @@ from infer.lib.infer_pack.models_onnx import SynthesizerTrnMsNSFsidM
|
||||
|
||||
|
||||
def export_onnx(ModelPath, ExportedPath):
|
||||
global cpt
|
||||
cpt = torch.load(ModelPath, map_location="cpu")
|
||||
cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0]
|
||||
vec_channels = 256 if cpt.get("version", "v1") == "v1" else 768
|
||||
|
Loading…
Reference in New Issue
Block a user