prain: int8/int4 experts quantized once on the GPU, rounds through vLLM's expert_map, [miss] min_weight skips experts not in VRAM; scripts: tune_moe, quality (answer NLL on identical token ids), bench --context, measure --quant

2026-10-28 22:07:25 +01:00
parent 3ae9bd73a7
commit 708e152549
15 changed files with 703 additions and 120 deletions
@@ -23,4 +23,7 @@ export PATH
 # libc++/compiler-rt defaults: everything links libstdc++/libgcc like the ROCm libraries.
 export CC=$PRAIN_ROOT/build/bin/clang CXX=$PRAIN_ROOT/build/bin/clang++
 export CFLAGS="-march=native" CXXFLAGS="-march=native"
+# Fused-MoE kernel configs tuned for this GPU by scripts/tune_moe.py (vLLM's defaults when
+# the files are missing).
+export VLLM_TUNED_CONFIG_FOLDER=$PRAIN_ROOT/build/moe-configs
 unset _site _dev _core