Searched refs:Float16To32 (Results 1 – 2 of 2) sorted by relevance
133 y_data[j] += Float16To32(*weight_ptr) * scale; in SparseReluProductPlusBias()207 multiplier *= Float16To32(embedding_matrix.quant_scales[feature_id]); in ConcatEmbeddings()219 multiplier *= Float16To32(embedding_matrix.quant_scales[feature_id]); in ConcatEmbeddings()
54 static inline float Float16To32(float16 f) { in Float16To32() function