diff options
Diffstat (limited to 'ggml/src/iqk/iqk_quantize.h')
-rw-r--r-- | ggml/src/iqk/iqk_quantize.h | 5 |
1 files changed, 3 insertions, 2 deletions
diff --git a/ggml/src/iqk/iqk_quantize.h b/ggml/src/iqk/iqk_quantize.h index 64860b4d..1a991787 100644 --- a/ggml/src/iqk/iqk_quantize.h +++ b/ggml/src/iqk/iqk_quantize.h @@ -67,10 +67,10 @@ size_t quantize_iq4_nl_r4(const float * GGML_RESTRICT src, void * GGML_RESTRICT void dequantize_row_iq4_nl_r4(const block_iq4_nl_r4 * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k); void vec_dot_iq4_nl_r4_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc); -void quantize_row_q4_0_r4_ref(const float * GGML_RESTRICT x, block_iq4_nl_r4 * GGML_RESTRICT y, int64_t k); +void quantize_row_q4_0_r4_ref(const float * GGML_RESTRICT x, block_iq4_nl_r8 * GGML_RESTRICT y, int64_t k); void quantize_row_q4_0_r4(const float * GGML_RESTRICT x, void * GGML_RESTRICT y, int64_t k); size_t quantize_q4_0_r4(const float * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row, const float * imatrix); -void dequantize_row_q4_0_r4(const block_iq4_nl_r4 * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k); +void dequantize_row_q4_0_r4(const block_iq4_nl_r8 * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k); void vec_dot_q4_0_r4_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const void * GGML_RESTRICT vx, size_t bx, const void * GGML_RESTRICT vy, size_t by, int nrc); void quantize_row_q8_0_r4_ref(const float * GGML_RESTRICT x, block_q8_0_r8 * GGML_RESTRICT y, int64_t k); @@ -218,6 +218,7 @@ void repack_f32_bf16_r16 (const void * GGML_RESTRICT src, void * GGML_RESTRICT d void repack_bf16_bf16_r16(const void * GGML_RESTRICT src, void * GGML_RESTRICT dst, int64_t nrows, int64_t n_per_row); void iqk_repack_tensor(struct ggml_tensor * tensor); +bool iqk_modify_tensor(struct ggml_tensor * tensor); // So we can re-pack Microsoft's BitNet I2_S quants void dequantize_row_ms_i2s(const void * GGML_RESTRICT x, float * GGML_RESTRICT y, int64_t k); |