Anonandksanislo 74945954fe ple: read Q8_0 and Q5_1 n-gram tables (#651) ...

Ordinary GGUFs ship per_layer_token_embd.weight in other formats than IQ4_NL: Unsloth's UD-Q6_K_XL and
Swift-1.5 Q4_K_L as Q8_0 (170-byte rows, ~54 GB), a Q5_K_M finetune as Q5_1 (120-byte rows). The engine refused
both at start.

- dequantize_q5_1 in strata/artifact/dequant.hpp (ggml's dequantize_row_q5_1: unsigned 5-bit codes times d plus the
  block minimum m); Q8_0 rows use the existing dequantize_q8_0
- two entries in the PLE format list (the table-driven commit before this one): Q5_1 and Q8_0, on both readers;
  PLE_ROW_BYTES_MAX grows to 170
- tests: dequant_q5_1_test (synthetic blocks against ggml, no model, no GPU), dequant_bf16_test knows Q5_1,
  ple_q5_parity takes Q5_0, Q5_1 or Q8_0 tables and checks both the mapped and the direct reader

Ported onto the format table by the maintainer from anon761's PR #651 (the --ple-io ram fault-in is the next
commit). Quality of the Q8_0 table: ksanislo measured 0.53% error against the BF16 table, FP8 2.66%, IQ4_NL 7.59%.

Co-authored-by: ksanislo <[email protected]>

2026-10-05 12:18:47 +02:00

..

core

ple: read Q8_0 and Q5_1 n-gram tables (#651)

2026-10-05 12:18:47 +02:00

cuda

Merge the 0.1.36 speed work: fused Q2_0 prompt experts on by default (STRATA_PF_FUSED=0 off), the native IQ packs' fused kernels opt-in (=1), cluster decode kernels; the fused test waits for its uploads

2026-10-02 16:31:42 +02:00

data/native_experts

Split GGUF artifacts: every shard found (a missing one is an error), tensors from their owning shard, per-role expert files (native_experts.txt v4)

2026-09-30 22:30:10 +02:00

hip

hip: preserve Q2_0 signed zero on gfx1012 with legacy HIP

2026-10-03 23:33:46 +02:00

platform

HIP tests: main's batched indexer append and I/O pool (C-2, C-4) replace the PR's duplicates

2026-09-29 14:36:32 +02:00