tests : add bit-level DT3 tests and Rust parity driver

test-dt3 checks the layout against an independent packer written from
the format spec: single-trit position mapping for all 256 (plane, pos)
pairs, structural byte-position checks at the region boundaries
(79/80, 119/120), exact round-trips with negative scales, byte parity
of the in-tree quantizer on already-ternary inputs, and the vec_dot
against a hand-made sum over the known trits (catches any path that
reads the padding 5th trit of the qh bytes).

test-dt3-rust-parity.py packs known trits with ternaria's pack_dt3 and
verifies that dequantize_row_dt3 (via test-dt3 --dequant) reproduces
d1*t1 + d2*t2 bit-exactly.

Also wires DT3 into the test-quantize-fns thresholds (ternary class).
This commit is contained in:
Millaguie committed 2026-08-10 23:33:14 +02:00
1 parent 985b0ecba2
commit b300ba053d
4 files changed
+498 -1

No files matched your search

+2 -1
View File
@@ -158,6 +158,7 @@ static int test_vec_dot_q(bool verbose) {
type == GGML_TYPE_Q1_0 ? MAX_QUANTIZATION_TOTAL_ERROR_BINARY :
type == GGML_TYPE_TQ1_0 ? MAX_QUANTIZATION_TOTAL_ERROR_TERNARY :
type == GGML_TYPE_TQ2_0 ? MAX_QUANTIZATION_TOTAL_ERROR_TERNARY :
type == GGML_TYPE_DT3 ? MAX_QUANTIZATION_TOTAL_ERROR_TERNARY :
type == GGML_TYPE_Q2_0 ? MAX_QUANTIZATION_TOTAL_ERROR_TERNARY :
type == GGML_TYPE_Q2_K ? MAX_QUANTIZATION_TOTAL_ERROR_2BITS :
type == GGML_TYPE_IQ2_S ? MAX_QUANTIZATION_TOTAL_ERROR_2BITS :
@@ -184,7 +185,7 @@ static int test_vec_dot_q(bool verbose) {
? MAX_DOT_PRODUCT_ERROR_LOWBIT
: type == GGML_TYPE_Q1_0
? MAX_DOT_PRODUCT_ERROR_BINARY
: type == GGML_TYPE_TQ1_0 || type == GGML_TYPE_TQ2_0 || type == GGML_TYPE_Q2_0
: type == GGML_TYPE_TQ1_0 || type == GGML_TYPE_TQ2_0 || type == GGML_TYPE_Q2_0 || type == GGML_TYPE_DT3
? MAX_DOT_PRODUCT_ERROR_TERNARY
: type == GGML_TYPE_NVFP4
? MAX_DOT_PRODUCT_ERROR_FP4