From 4f237d44f6d75afbb5cef39d4d6b0b35b2a517c7 Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Tue, 30 Jul 2024 16:11:25 +0300 Subject: iq3_k: Basics Quantize/dequantize, CUDA dequantize. PPL of LLaMA-3.1-8B is better than iq3_s and iq3_m. --- ggml/src/ggml-quants.c | 1 + 1 file changed, 1 insertion(+) (limited to 'ggml/src/ggml-quants.c') diff --git a/ggml/src/ggml-quants.c b/ggml/src/ggml-quants.c index 4b3bf361..c2c66f38 100644 --- a/ggml/src/ggml-quants.c +++ b/ggml/src/ggml-quants.c @@ -14948,6 +14948,7 @@ bool ggml_validate_row_data(enum ggml_type type, const void * data, size_t nbyte VALIDATE_ROW_DATA_D_F16_IMPL(block_iq4_nl, data, nb); } break; case GGML_TYPE_IQ2_K: break; + case GGML_TYPE_IQ3_K: break; case GGML_TYPE_IQ4_K: break; case GGML_TYPE_IQ5_K: break; case GGML_TYPE_Q4_0_4_4: -- cgit v1.2.3