Skip to content

Updated CREF to handle bias=null for FC,Conv 8x16 CREF code #3492

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
unmeshna017 wants to merge 5 commits into
base: main
Choose a base branch
from
Open

Updated CREF to handle bias=null for FC,Conv 8x16 CREF code #3492

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
73c2e8e
Handle bias=null for FC,Conv 8x16 CREF code
unmeshna017 Feb 5, 2026
452cca9
Uncommented input and output precision check in fully connected operator
unmeshna017 Mar 23, 2026
89b2545
Revert "Uncommented input and output precision check in fully connect…
unmeshna017 Mar 23, 2026
fe32354
Comment out tensor type equality check
rameshkunasi Mar 24, 2026
7bcb776
Uncommented input and output precision check in fully connected operator
unmeshna017 Mar 24, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 7 additions & 7 deletions tensorflow/lite/micro/kernels/conv.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -80,7 +80,7 @@ TfLiteStatus ConvEval(TfLiteContext* context, TfLiteNode* node) {
break;
}
case kTfLiteInt16: {
if (bias == nullptr || bias->type == kTfLiteInt32) {
if (bias != nullptr && bias->type == kTfLiteInt32) {
reference_integer_ops::ConvPerChannel(
ConvParamsQuantized(params, data),
data.per_channel_output_multiplier, data.per_channel_output_shift,
Expand All @@ -92,16 +92,16 @@ TfLiteStatus ConvEval(TfLiteContext* context, TfLiteNode* node) {
weights_comp_td,
data.weights_scratch_index),
tflite::micro::GetTensorShape(bias),
tflite::micro::GetOptionalTensorData<int32_t>(
tflite::micro::GetTensorData<int32_t>(
micro_context, bias, bias_comp_td, data.bias_scratch_index),
#else // USE_TFLM_COMPRESSION
tflite::micro::GetTensorData<int8_t>(filter),
tflite::micro::GetTensorShape(bias),
tflite::micro::GetOptionalTensorData<std::int32_t>(bias),
tflite::micro::GetTensorData<std::int32_t>(bias),
#endif // USE_TFLM_COMPRESSION
tflite::micro::GetTensorShape(output),
tflite::micro::GetTensorData<int16_t>(output));
} else if (bias->type == kTfLiteInt64) {
} else if (bias == nullptr || bias->type == kTfLiteInt64) {
reference_integer_ops::ConvPerChannel(
ConvParamsQuantized(params, data),
data.per_channel_output_multiplier, data.per_channel_output_shift,
Expand All @@ -113,12 +113,12 @@ TfLiteStatus ConvEval(TfLiteContext* context, TfLiteNode* node) {
weights_comp_td,
data.weights_scratch_index),
tflite::micro::GetTensorShape(bias),
tflite::micro::GetTensorData<int64_t>(
tflite::micro::GetOptionalTensorData<int64_t>(
micro_context, bias, bias_comp_td, data.bias_scratch_index),
#else // USE_TFLM_COMPRESSION
tflite::micro::GetTensorData<int8_t>(filter),
tflite::micro::GetTensorShape(bias),
tflite::micro::GetTensorData<std::int64_t>(bias),
tflite::micro::GetOptionalTensorData<std::int64_t>(bias),
#endif // USE_TFLM_COMPRESSION
tflite::micro::GetTensorShape(output),
tflite::micro::GetTensorData<int16_t>(output));
Expand Down Expand Up @@ -194,4 +194,4 @@ TFLMRegistration Register_CONV_2D() {
return tflite::micro::RegisterOp(ConvInit, ConvPrepare, ConvEval);
}

} // namespace tflite
} // namespace tflite
12 changes: 6 additions & 6 deletions tensorflow/lite/micro/kernels/fully_connected.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -238,7 +238,7 @@ TfLiteStatus FullyConnectedEval(TfLiteContext* context, TfLiteNode* node) {
case kTfLiteInt16: {
switch (filter->type) {
case kTfLiteInt8: {
if (bias == nullptr || bias->type == kTfLiteInt32) {
if (bias != nullptr && bias->type == kTfLiteInt32) {
data.is_per_channel
? tflite::reference_integer_ops::FullyConnectedPerChannel(
FullyConnectedParamsQuantized(data),
Expand All @@ -253,13 +253,13 @@ TfLiteStatus FullyConnectedEval(TfLiteContext* context, TfLiteNode* node) {
micro_context, filter, weights_comp_td,
data.weights_scratch_index),
tflite::micro::GetTensorShape(bias),
tflite::micro::GetOptionalTensorData<int32_t>(
tflite::micro::GetTensorData<int32_t>(
micro_context, bias, bias_comp_td,
data.bias_scratch_index),
#else // USE_TFLM_COMPRESSION
tflite::micro::GetTensorData<int8_t>(filter),
tflite::micro::GetTensorShape(bias),
tflite::micro::GetOptionalTensorData<int32_t>(bias),
tflite::micro::GetTensorData<int32_t>(bias),
#endif // USE_TFLM_COMPRESSION
tflite::micro::GetTensorShape(output),
tflite::micro::GetTensorData<int16_t>(output))
Expand All @@ -273,17 +273,17 @@ TfLiteStatus FullyConnectedEval(TfLiteContext* context, TfLiteNode* node) {
micro_context, filter, weights_comp_td,
data.weights_scratch_index),
tflite::micro::GetTensorShape(bias),
tflite::micro::GetOptionalTensorData<int32_t>(
tflite::micro::GetTensorData<int32_t>(
micro_context, bias, bias_comp_td,
data.bias_scratch_index),
#else // USE_TFLM_COMPRESSION
tflite::micro::GetTensorData<int8_t>(filter),
tflite::micro::GetTensorShape(bias),
tflite::micro::GetOptionalTensorData<int32_t>(bias),
tflite::micro::GetTensorData<int32_t>(bias),
#endif // USE_TFLM_COMPRESSION
tflite::micro::GetTensorShape(output),
tflite::micro::GetTensorData<int16_t>(output));
} else if (bias->type == kTfLiteInt64) {
} else if (bias == nullptr || bias->type == kTfLiteInt64) {
data.is_per_channel
? tflite::reference_integer_ops::FullyConnectedPerChannel(
FullyConnectedParamsQuantized(data),
Expand Down
Loading