Home
last modified time | relevance | path

Searched refs:num_groups (Results 1 – 2 of 2) sorted by relevance

/packages/modules/Bluetooth/system/bta/groups/
Dgroups.cc176 auto num_groups = std::count_if( in SerializeGroups() local
180 if ((num_groups == 0) || (num_groups > std::numeric_limits<uint8_t>::max())) in SerializeGroups()
183 out.resize(GROUP_STORAGE_HEADER_SZ + (num_groups * GROUP_STORAGE_ENTRY_SZ)); in SerializeGroups()
188 UINT8_TO_STREAM(ptr, num_groups); in SerializeGroups()
214 uint8_t num_groups; in DeserializeGroups() local
215 STREAM_TO_UINT8(num_groups, ptr); in DeserializeGroups()
218 GROUP_STORAGE_HEADER_SZ + (num_groups * GROUP_STORAGE_ENTRY_SZ)) { in DeserializeGroups()
224 while (num_groups--) { in DeserializeGroups()
/packages/modules/NeuralNetworks/tools/api/
Dtypes.spec3872 * Given an input tensor and a integer value of num_groups, CHANNEL_SHUFFLE
3873 * divide the channel dimension into num_groups groups, and reorganize the
3878 * output_channel[k * num_groups + g] = input_channel[g * group_size + k]
3880 * where group_size = num_channels / num_groups
3882 * The number of channels must be divisible by num_groups.
4290 * Specifically, the input channels are divided into num_groups groups, each with
4291 * depth depth_group, i.e. depth_in = num_groups * depth_group. The convolutional
4292 * filters are also divided into num_groups groups, i.e. depth_out is divisible
4293 * by num_groups. GROUPED_CONV applies each group of filters to the corresponding
4308 * where channel_multiplier = depth_out / num_groups
[all …]