WaveCut/DeepSeek-V4-Flash-0731-REAM144-163B-DS4-GGUF

DeepSeek V4 Flash — REAM144 (163B) · DS4 Q2

A 2-bit build of DeepSeek-V4-Flash-0731-REAM144-163B — DeepSeek-V4-Flash with 144 of the original 256 experts per layer, sized to run fully resident on a 64 GiB Mac with room for 8k context. Quantized with the standard DS4 recipe (2-bit experts, 8-bit attention) and a fresh importance matrix.

[!IMPORTANT] This is a DS4-specific GGUF. Run it with the DS4 fork — the 144-expert topology needs its variable expert count support. Generic llama.cpp will not load this file.

[!WARNING] Live smoke testing passed 7/10 scenarios on the first run. Independent reruns show the failures (Tool calling (DSML), Code refactoring, Tool call → code chain) are intermittent, not absolute — see the Stability column below for per-scenario pass rates. Multilingual chat, reasoning and long dialogs are consistently healthy. The full-precision native checkpoint may behave better — 2-bit quantization hits agentic behavior hardest.

Files

Run

Plain:

./ds4 -m ream144.gguf -c 8192

With DSpark speculative decoding (faster generation, more memory):

./ds4 -m ream144.gguf --mtp dspark.gguf --dspark -c 8192

Adding DSpark pushes the total past a 64 GiB budget — measured on a 64 GiB Mac it slows prefill ~10× and can thrash generation; use it on larger hosts only.

How it was made

One pruning step, straight from the original — no cascading. Expert importance was measured by running `deepseek-ai/DeepSeek-V4-Flash-0731` over a ~5-million-token calibration mix (multi-turn dialogs, thinking and direct modes, rendered with the model's own chat encoder). The strongest experts of every domain were protected from pruning, the survivors were carried over byte-identical, and the router was re-balanced to keep the original selection behavior.

This line replaces the earlier cascaded REAM builds (now archived under -exp names), which degraded badly in multi-turn use.

Smoke results

Every scenario is a live multi-turn conversation run end-to-end on the DS4 runtime (raw evidence ships in SMOKE_REPORT.json).

Stability = pass rate over independent reruns of the scenarios that failed the first run; passing scenarios were not re-run.

Limitations

• —Needs the DS4 fork; not a generic llama.cpp file.
• —2-bit quantization is aggressive: expect the native checkpoint to be smarter than this build, especially on agentic tool use.
• —Memory use grows with context length and DSpark; the sizes above are the files alone.