Skip to content

Commit

Permalink
Speed profiling improvements (ultralytics#2648)
Browse files Browse the repository at this point in the history
* Speed profiling improvements

* Update torch_utils.py

deepcopy() required to avoid adding elements to model.

* Update torch_utils.py
  • Loading branch information
glenn-jocher authored Mar 29, 2021
1 parent 9325433 commit 6b7ecac
Show file tree
Hide file tree
Showing 2 changed files with 5 additions and 4 deletions.
7 changes: 4 additions & 3 deletions hubconf.py
Original file line number Diff line number Diff line change
Expand Up @@ -38,9 +38,10 @@ def create(name, pretrained, channels, classes, autoshape):
fname = f'{name}.pt' # checkpoint filename
attempt_download(fname) # download if not found locally
ckpt = torch.load(fname, map_location=torch.device('cpu')) # load
state_dict = ckpt['model'].float().state_dict() # to FP32
state_dict = {k: v for k, v in state_dict.items() if model.state_dict()[k].shape == v.shape} # filter
model.load_state_dict(state_dict, strict=False) # load
msd = model.state_dict() # model state_dict
csd = ckpt['model'].float().state_dict() # checkpoint state_dict as FP32
csd = {k: v for k, v in csd.items() if msd[k].shape == v.shape} # filter
model.load_state_dict(csd, strict=False) # load
if len(ckpt['model'].names) == classes:
model.names = ckpt['model'].names # set class names attribute
if autoshape:
Expand Down
2 changes: 1 addition & 1 deletion utils/torch_utils.py
Original file line number Diff line number Diff line change
Expand Up @@ -191,7 +191,7 @@ def fuse_conv_and_bn(conv, bn):
# prepare filters
w_conv = conv.weight.clone().view(conv.out_channels, -1)
w_bn = torch.diag(bn.weight.div(torch.sqrt(bn.eps + bn.running_var)))
fusedconv.weight.copy_(torch.mm(w_bn, w_conv).view(fusedconv.weight.size()))
fusedconv.weight.copy_(torch.mm(w_bn, w_conv).view(fusedconv.weight.shape))

# prepare spatial bias
b_conv = torch.zeros(conv.weight.size(0), device=conv.weight.device) if conv.bias is None else conv.bias
Expand Down

0 comments on commit 6b7ecac

Please sign in to comment.