diff options
Diffstat (limited to 'ggml/src/ggml.c')
| -rw-r--r-- | ggml/src/ggml.c | 21 |
1 files changed, 21 insertions, 0 deletions
diff --git a/ggml/src/ggml.c b/ggml/src/ggml.c index 6bfeca1e..0881756d 100644 --- a/ggml/src/ggml.c +++ b/ggml/src/ggml.c @@ -992,6 +992,18 @@ static const ggml_type_traits_t type_traits[GGML_TYPE_COUNT] = { .vec_dot_type = GGML_TYPE_Q8_K, .nrows = 1, }, + [GGML_TYPE_IQ2_K] = { + .type_name = "iq2_k", + .blck_size = QK_K, + .type_size = sizeof(block_iq2_k), + .is_quantized = true, + .to_float = (ggml_to_float_t) dequantize_row_iq2_k, + .from_float = quantize_row_iq2_k, + .from_float_ref = (ggml_from_float_t)quantize_row_iq2_k_ref, + .vec_dot = vec_dot_iq2_k_q8_k, + .vec_dot_type = GGML_TYPE_Q8_K, + .nrows = 1, + }, }; // For internal test use @@ -3342,6 +3354,7 @@ enum ggml_type ggml_ftype_to_ggml_type(enum ggml_ftype ftype) { case GGML_FTYPE_MOSTLY_IQ4_NL: wtype = GGML_TYPE_IQ4_NL; break; case GGML_FTYPE_MOSTLY_IQ4_XS: wtype = GGML_TYPE_IQ4_XS; break; case GGML_FTYPE_MOSTLY_IQ4_K: wtype = GGML_TYPE_IQ4_K; break; + case GGML_FTYPE_MOSTLY_IQ2_K: wtype = GGML_TYPE_IQ2_K; break; case GGML_FTYPE_MOSTLY_IQ3_S: wtype = GGML_TYPE_IQ3_S; break; case GGML_FTYPE_MOSTLY_IQ2_S: wtype = GGML_TYPE_IQ2_S; break; case GGML_FTYPE_MOSTLY_Q4_0_4_4: wtype = GGML_TYPE_Q4_0_4_4; break; @@ -9592,6 +9605,7 @@ static void ggml_compute_forward_add( case GGML_TYPE_IQ4_NL: case GGML_TYPE_IQ4_XS: case GGML_TYPE_IQ4_K: + case GGML_TYPE_IQ2_K: case GGML_TYPE_IQ3_S: case GGML_TYPE_IQ2_S: case GGML_TYPE_Q4_0_4_4: @@ -9973,6 +9987,7 @@ static void ggml_compute_forward_add1( case GGML_TYPE_IQ4_NL: case GGML_TYPE_IQ4_XS: case GGML_TYPE_IQ4_K: + case GGML_TYPE_IQ2_K: case GGML_TYPE_IQ3_S: case GGML_TYPE_IQ2_S: case GGML_TYPE_Q4_0_4_4: @@ -10104,6 +10119,7 @@ static void ggml_compute_forward_acc( case GGML_TYPE_IQ4_NL: case GGML_TYPE_IQ4_XS: case GGML_TYPE_IQ4_K: + case GGML_TYPE_IQ2_K: case GGML_TYPE_IQ3_S: case GGML_TYPE_IQ2_S: case GGML_TYPE_Q4_0_4_4: @@ -13024,6 +13040,7 @@ static void ggml_compute_forward_out_prod( case GGML_TYPE_IQ4_NL: case GGML_TYPE_IQ4_XS: case GGML_TYPE_IQ4_K: + case GGML_TYPE_IQ2_K: case GGML_TYPE_IQ3_S: case GGML_TYPE_IQ2_S: case GGML_TYPE_Q4_0_4_4: @@ -13215,6 +13232,7 @@ static void ggml_compute_forward_set( case GGML_TYPE_IQ4_NL: case GGML_TYPE_IQ4_XS: case GGML_TYPE_IQ4_K: + case GGML_TYPE_IQ2_K: case GGML_TYPE_IQ3_S: case GGML_TYPE_IQ2_S: case GGML_TYPE_Q4_0_4_4: @@ -13480,6 +13498,7 @@ static void ggml_compute_forward_get_rows( case GGML_TYPE_IQ4_NL: case GGML_TYPE_IQ4_XS: case GGML_TYPE_IQ4_K: + case GGML_TYPE_IQ2_K: case GGML_TYPE_IQ3_S: case GGML_TYPE_IQ2_S: case GGML_TYPE_Q4_0_4_4: @@ -14072,6 +14091,7 @@ static void ggml_compute_forward_clamp( case GGML_TYPE_IQ4_NL: case GGML_TYPE_IQ4_XS: case GGML_TYPE_IQ4_K: + case GGML_TYPE_IQ2_K: case GGML_TYPE_IQ3_S: case GGML_TYPE_IQ2_S: case GGML_TYPE_Q8_K: @@ -20808,6 +20828,7 @@ size_t ggml_quantize_chunk( case GGML_TYPE_IQ4_NL: result = quantize_iq4_nl (src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break; case GGML_TYPE_IQ4_XS: result = quantize_iq4_xs (src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break; case GGML_TYPE_IQ4_K: result = quantize_iq4_k (src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break; + case GGML_TYPE_IQ2_K: result = quantize_iq2_k (src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break; case GGML_TYPE_Q4_0_4_4: result = quantize_q4_0_4x4(src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break; case GGML_TYPE_Q4_0_4_8: result = quantize_q4_0_4x8(src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break; case GGML_TYPE_Q4_0_8_8: result = quantize_q4_0_8x8(src + start, (char *) dst + start_row * row_size, nrows, n_per_row, imatrix); break; |