mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-08-12 22:29:39 +04:00

Files

T

NexesenexandGitHub 3e2f5696a1 help: document all previously undocumented CLI options across tools (#2180 )

examples/quantize/quantize.cpp:
  - add --per-layer-token-embedding-type to usage line and description

common/common.cpp (gpt_params_print_usage):
  - speculative: --spec-replace
  - sampling: --dry-multiplier, --dry-base, --dry-allowed-length,
    --dry-penalty-last-n, --dry-sequence-breaker
  - multi-modality: --audio, --mmproj-url, --no-mmproj-offload
  - main infill: --infill
  - backend: --offload-policy/-op, --no-offload-only-active-experts/-no-ooae,
    --gpu-fit-margin/-gfm
  - model: --override-tensor/-ot
  - imatrix: --output-tensor-name
  - bench: --n-repetitions/-nrep, --warmup-batch/-wb, --output-format
  - server: --send-done, --sql-save-file, --sqlite-zstd-ext-file

examples/imatrix/imatrix.cpp:
  - add --layer-similarity/-lsim under new imatrix-specific options section

examples/sweep-bench/sweep-bench.cpp:
  - replace stub usage with full help: delegates to gpt_params_print_usage
    and documents sweep-bench specific options (-nrep, -wb, --output-format)

2026-07-25 16:04:30 +03:00

CMakeLists.txt

build: rename main → llama-cli, server → llama-server, llava-cli → llama-llava-cli, etc... (#7809 )

2024-06-13 00:41:52 +01:00

imatrix.cpp

help: document all previously undocumented CLI options across tools (#2180 )

2026-07-25 16:04:30 +03:00

README.md

Merge mainline llama.cpp (#3 )

2024-07-27 07:55:01 +02:00

README.md

llama.cpp/examples/imatrix

Compute an importance matrix for a model and given text dataset. Can be used during quantization to enchance the quality of the quantized models. More information is available here: https://github.com/ggerganov/llama.cpp/pull/4861

Usage

./llama-imatrix \
    -m model.gguf -f some-text.txt [-o imatrix.dat] [--process-output] [--verbosity 1] \
    [--no-ppl] [--chunk 123] [--output-frequency 10] [--save-frequency 0] \
    [--in-file imatrix-prev-0.dat --in-file imatrix-prev-1.dat ...]

Here -m with a model name and -f with a file containing training data (such as e.g. wiki.train.raw) are mandatory. The parameters in square brackets are optional and have the following meaning:

-o (or --output-file) specifies the name of the file where the computed data will be stored. If missing imatrix.dat is used.
--verbosity specifies the verbosity level. If set to 0, no output other than the perplexity of the processed chunks will be generated. If set to 1, each time the results are saved a message is written to stderr. If >=2, a message is output each time data is collected for any tensor. Default verbosity level is 1.
--output-frequency specifies how often the so far computed result is saved to disk. Default is 10 (i.e., every 10 chunks)
--save-frequency specifies how often to save a copy of the imatrix in a separate file. Default is 0 (i.e., never)
--process-output specifies if data will be collected for the output.weight tensor. My experience is that it is better to not utilize the importance matrix when quantizing output.weight, so this is set to false by default.

For faster computation, make sure to use GPU offloading via the -ngl argument

Example

GGML_CUDA=1 make -j

# generate importance matrix (imatrix.dat)
./llama-imatrix -m ggml-model-f16.gguf -f train-data.txt -ngl 99

# use the imatrix to perform a Q4_K_M quantization
./llama-quantize --imatrix imatrix.dat ggml-model-f16.gguf ./ggml-model-q4_k_m.gguf q4_k_m