diff options
| author | Kawrakow <48489457+ikawrakow@users.noreply.github.com> | 2024-02-18 18:16:55 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-02-18 18:16:55 +0200 |
| commit | bd2d4e393b2b7d2a1b2e201058e26017c9728ead (patch) | |
| tree | 5c51109459cf1a25fc92fdb11d420895e16785ac /examples/quantize/quantize.cpp | |
| parent | c8e0d7efeb7634ecc2e9832e879ab9fca4510e71 (diff) | |
1.5 bit quantization (#5453)
* iq1_s: WIP basics
* iq1_s: CUDA is working
* iq1_s: scalar CPU dot product
* iq1_s: WIP AVX2 dot product - something is not right
* Fix tests
* Fix shadow warnings
* Fix after merge with latest master
* iq1_s: AVX2 finally works
* iq1_s: ARM_NEON dot product. Works, but not very fast
* iq1_s: better grid
* iq1_s: use IQ2_XXS for attn_output
At a cost of 0.04 extra bpw this gives a big improvement in PPL.
* iq1_s: Metal basics
Dequantize works, but not dot product
* iq1_s: Metal works, but quite slow
As usual, Apple Silicon does not like the code I write.
* iq1_s: Tests
* iq1_s: slightly faster dot product
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'examples/quantize/quantize.cpp')
| -rw-r--r-- | examples/quantize/quantize.cpp | 6 |
1 files changed, 4 insertions, 2 deletions
diff --git a/examples/quantize/quantize.cpp b/examples/quantize/quantize.cpp index 4a5c504e..ea7ba50c 100644 --- a/examples/quantize/quantize.cpp +++ b/examples/quantize/quantize.cpp @@ -23,6 +23,7 @@ static const std::vector<struct quant_option> QUANT_OPTIONS = { { "Q5_1", LLAMA_FTYPE_MOSTLY_Q5_1, " 4.70G, +0.0349 ppl @ LLaMA-v1-7B", }, { "IQ2_XXS",LLAMA_FTYPE_MOSTLY_IQ2_XXS," 2.06 bpw quantization", }, { "IQ2_XS", LLAMA_FTYPE_MOSTLY_IQ2_XS, " 2.31 bpw quantization", }, + { "IQ1_S", LLAMA_FTYPE_MOSTLY_IQ1_S, " 1.56 bpw quantization", }, { "Q2_K", LLAMA_FTYPE_MOSTLY_Q2_K, " 2.63G, +0.6717 ppl @ LLaMA-v1-7B", }, { "Q2_K_S", LLAMA_FTYPE_MOSTLY_Q2_K_S, " 2.16G, +9.0634 ppl @ LLaMA-v1-7B", }, { "IQ3_XXS",LLAMA_FTYPE_MOSTLY_IQ3_XXS," 3.06 bpw quantization", }, @@ -287,9 +288,10 @@ int main(int argc, char ** argv) { } } - if ((params.ftype == LLAMA_FTYPE_MOSTLY_IQ2_XS || params.ftype == LLAMA_FTYPE_MOSTLY_IQ2_XXS || params.ftype == LLAMA_FTYPE_MOSTLY_Q2_K_S) && imatrix_data.empty()) { + if ((params.ftype == LLAMA_FTYPE_MOSTLY_IQ2_XS || params.ftype == LLAMA_FTYPE_MOSTLY_IQ2_XXS || + params.ftype == LLAMA_FTYPE_MOSTLY_Q2_K_S || params.ftype == LLAMA_FTYPE_MOSTLY_IQ1_S) && imatrix_data.empty()) { fprintf(stderr, "\n===============================================================================================\n"); - fprintf(stderr, "Please do not use IQ2_XXS, IQ2_XS or Q2_K_S quantization without an importance matrix\n"); + fprintf(stderr, "Please do not use IQ1_S, IQ2_XXS, IQ2_XS or Q2_K_S quantization without an importance matrix\n"); fprintf(stderr, "===============================================================================================\n\n\n"); return 1; } |