Skip to content

Commit

Permalink
Fix FP32 TensorRT model export (ultralytics#8046)
Browse files Browse the repository at this point in the history
Fixed FP32 TRT  model export

Co-authored-by: Glenn Jocher <[email protected]>
  • Loading branch information
SahilChachra and glenn-jocher authored May 31, 2022
1 parent 487edd5 commit 6dd6aea
Showing 1 changed file with 2 additions and 2 deletions.
4 changes: 2 additions & 2 deletions export.py
Original file line number Diff line number Diff line change
Expand Up @@ -264,8 +264,8 @@ def export_engine(model, im, file, train, half, simplify, workspace=4, verbose=F
for out in outputs:
LOGGER.info(f'{prefix}\toutput "{out.name}" with shape {out.shape} and dtype {out.dtype}')

LOGGER.info(f'{prefix} building FP{16 if builder.platform_has_fast_fp16 else 32} engine in {f}')
if builder.platform_has_fast_fp16:
LOGGER.info(f'{prefix} building FP{16 if builder.platform_has_fast_fp16 and half else 32} engine in {f}')
if builder.platform_has_fast_fp16 and half:
config.set_flag(trt.BuilderFlag.FP16)
with builder.build_engine(network, config) as engine, open(f, 'wb') as t:
t.write(engine.serialize())
Expand Down

0 comments on commit 6dd6aea

Please sign in to comment.