From 0e018fe008eacebdbcfa2d61b6c988c245c961cd Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Sat, 22 Apr 2023 16:31:56 +0300 Subject: ggml : fix Q4_3 cuBLAS --- ggml.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'ggml.c') diff --git a/ggml.c b/ggml.c index d9a95af8..281b2028 100644 --- a/ggml.c +++ b/ggml.c @@ -7992,6 +7992,9 @@ static void ggml_compute_forward_mul_mat_q_f32( else if (type == GGML_TYPE_Q4_2) { dequantize_row_q_cuda = dequantize_row_q4_2_cuda; } + else if (type == GGML_TYPE_Q4_3) { + dequantize_row_q_cuda = dequantize_row_q4_3_cuda; + } else { GGML_ASSERT(false); } -- cgit v1.2.3