Skip to content

Commit 8d7cbc5

Browse files
authored
QNN Whisper Runner transcribe documentation (#13665)
Summary: Clarify the format of input to the `transcribe` function of the runner Reviewed By: cccclai Differential Revision: D80957217
1 parent 7cf8a5e commit 8d7cbc5

File tree

1 file changed

+5
-1
lines changed
  • examples/qualcomm/oss_scripts/whisper/runner

1 file changed

+5
-1
lines changed

examples/qualcomm/oss_scripts/whisper/runner/runner.cpp

Lines changed: 5 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -112,7 +112,11 @@ uint64_t Runner::logits_to_token(
112112
const executorch::aten::Tensor& logits_tensor) {
113113
return sampler_->sample(logits_tensor.data_ptr<float>());
114114
}
115-
115+
/**
116+
* @param inputs: A vector containing one element: a vector of bytes that
117+
* encodes a float tensor in little-endian byte order.
118+
*
119+
*/
116120
Error Runner::transcribe(
117121
int32_t seq_len,
118122
std::vector<std::vector<char>>& inputs,

0 commit comments

Comments
 (0)