mirror of https://github.com/zenlm/zen5-engine.git synced 2026-10-30 14:57:31 +00:00

Files

T

antirez d0400a15bd Add DS4 imatrix and GGUF quantization tools

Add routed-MoE imatrix collection to the DS4 graph path, including dataset generation docs and the tracked calibration prompt corpus.

Add a standalone DeepSeek V4 Flash HF-to-GGUF quantizer that reuses GGUF template metadata, emits DS4 Q2/Q4 recipes, supports imatrix-aware routed expert quantization, and internalizes the required Q8_0, Q4_K, Q2_K, and IQ2_XXS quantizers instead of vendoring GGML.

Document quantizer provenance, output formats, and the DS4-specific imatrix mapping used by the generated GGUF files.

2026-05-12 18:34:08 +02:00

5.2 KiB

Raw Permalink Blame History

DS4 Imatrix Pipeline

This directory contains the calibration dataset and instructions used to build activation importance matrices for DeepSeek V4 Flash GGUF quantization.

The current imatrix target is the routed MoE path. DS4 has 43 layers, 256 routed experts per layer, and three routed expert tensors per layer:

blk.N.ffn_gate_exps.weight
blk.N.ffn_up_exps.weight
blk.N.ffn_down_exps.weight

For gate/up tensors, the collector records the squared FFN-normalized input activation. For down tensors, it records the squared routed SwiGLU row after route weighting. The result tells the quantizer which input columns are used more heavily by the actual DS4 inference graph.

1. Build The Calibration Dataset

The tracked dataset is in gguf-tools/imatrix/dataset/. Regenerate it from the repository root with:

python3 gguf-tools/imatrix/dataset/build_ds4_imatrix_dataset.py

The important output is:

gguf-tools/imatrix/dataset/rendered_prompts.txt

It contains DS4-rendered chat prompts, separated by visible DS4_IMATRIX_PROMPT markers. The prompts include:

C/Metal source-review prompts from this repository.
Long-context snippets.
Agent/tool-call prompts using DS4's DSML syntax.
English and Italian prompts.
Both thinking and non-thinking assistant prefixes.

The current tracked dataset has 1952 rendered prompts and roughly 1.44M tokens by the coarse bytes/4 estimate. Check gguf-tools/imatrix/dataset/manifest.json for the exact generated-file summary.

2. Collect The Imatrix

Use the DS4 runtime itself to collect routed MoE activation statistics:

./ds4 \
  -m ../deepseek-v4-quants/gguf/DeepSeek-V4-Flash-Q4KExperts-F16HC-F16Compressor-F16Indexer-Q8Attn-Q8Shared-Q8Out-chat-v2.gguf \
  --imatrix-dataset gguf-tools/imatrix/dataset/rendered_prompts.txt \
  --imatrix-out ../deepseek-v4-quants/imatrix/DeepSeek-V4-Flash-chat-v2-routed-moe-ds4-1p5m.dat \
  --ctx 32768

Useful smoke-test limits:

./ds4 \
  -m MODEL.gguf \
  --imatrix-dataset gguf-tools/imatrix/dataset/rendered_prompts.txt \
  --imatrix-out /tmp/ds4-test.imatrix.dat \
  --imatrix-max-prompts 1 \
  --imatrix-max-tokens 4096

The collector is Metal-only because it hooks the layer-major Metal prefill graph. It does not change inference math; it reads the already materialized MoE inputs and accumulates sum(x[column]^2) per routed expert.

The output format is llama.cpp's legacy binary .dat imatrix format. DS4 packs per-expert vectors into one entry per routed expert tensor:

entry length = n_expert * n_columns

The quantizer slices the right expert's segment when quantizing each expert.

3. Generate GGUF Files With The Imatrix

The local C quantization tool in gguf-tools/ supports:

--imatrix FILE
--imatrix-strict

Example Q4 routed-expert regeneration:

gguf-tools/deepseek4-quantize \
  --hf ../deepseek-v4-quants/hf/DeepSeek-V4-Flash \
  --template ../deepseek-v4-quants/gguf/DeepSeek-V4-Flash-Q4KExperts-F16HC-F16Compressor-F16Indexer-Q8Attn-Q8Shared-Q8Out-chat-v2.gguf \
  --out ../deepseek-v4-quants/gguf/DeepSeek-V4-Flash-Q4KExperts-F16HC-F16Compressor-F16Indexer-Q8Attn-Q8Shared-Q8Out-chat-v2-imatrix.gguf \
  --imatrix ../deepseek-v4-quants/imatrix/DeepSeek-V4-Flash-chat-v2-routed-moe-ds4-1p5m.dat

Example Q2 regeneration:

gguf-tools/deepseek4-quantize \
  --hf ../deepseek-v4-quants/hf/DeepSeek-V4-Flash \
  --template ../deepseek-v4-quants/gguf/DeepSeek-V4-Flash-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ8-chat-v2.gguf \
  --out ../deepseek-v4-quants/gguf/DeepSeek-V4-Flash-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ8-chat-v2-imatrix.gguf \
  --imatrix ../deepseek-v4-quants/imatrix/DeepSeek-V4-Flash-chat-v2-routed-moe-ds4-1p5m.dat

For Q4, the imatrix does not change the runtime tensor type: routed experts remain Q4_K. It changes how quantization error is weighted while choosing scales and codes. For Q2, it replaces the previous synthetic weight-energy fallback used for IQ2_XXS gate/up experts with real activation statistics.

4. Evaluate

Useful local tools:

misc/quant_eval.c
gguf-tools/quality-testing/

misc/quant_eval.c compares local GGUF variants by greedy/top-logit behavior. gguf-tools/quality-testing/ can score local GGUFs against official DeepSeek API continuations by target-token negative log likelihood.

The Q4 imatrix file uploaded to Hugging Face was tested on 100 official DeepSeek V4 Flash continuations:

old Q4 avg NLL:         0.177357819
Q4 imatrix avg NLL:     0.173895148
relative NLL change:   -1.95%
case wins:              54 imatrix / 46 old
first-token matches:    83 imatrix / 81 old
avg greedy LCP:         12.21 imatrix / 11.94 old

Compatibility

The .dat file is intentionally in llama.cpp's legacy imatrix format, so the data is not conceptually tied to DS4. In practice, it is immediately useful only with a quantizer that understands DS4's tensor names and packed per-expert entries. The current deepseek4-quantize tooling does that.

Other GGUF creation tools can use the same imatrix if they implement the same name mapping and per-expert slicing convention. Without that DS4-specific mapping, a generic imatrix loader will see valid data but will not know how to apply the packed routed-expert vectors correctly.

5.2 KiB Raw Permalink Blame History