diff options
| author | Iwan Kawrakow <iwan.kawrakow@gmail.com> | 2024-06-24 16:42:30 +0200 |
|---|---|---|
| committer | Iwan Kawrakow <iwan.kawrakow@gmail.com> | 2024-06-24 16:42:30 +0200 |
| commit | 707d08792701de5a517c45c6c16bf4d816c00d88 (patch) | |
| tree | 1720ae21c0ec450707f7b74b8a67cc4a54aaba31 /examples/quantize/quantize.cpp | |
| parent | 49822f84a9ad633919c40915cde5cb78694c5390 (diff) | |
Bitnet: tiny bity faster 1.625 bpw variant on Metal
We get 70.7 t/s for TG-128 vs 69.5 t/s before.
Diffstat (limited to 'examples/quantize/quantize.cpp')
| -rw-r--r-- | examples/quantize/quantize.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/examples/quantize/quantize.cpp b/examples/quantize/quantize.cpp index a5ffb2b2..0a61a083 100644 --- a/examples/quantize/quantize.cpp +++ b/examples/quantize/quantize.cpp @@ -26,7 +26,7 @@ static const std::vector<struct quant_option> QUANT_OPTIONS = { { "IQ2_M", LLAMA_FTYPE_MOSTLY_IQ2_M, " 2.7 bpw quantization", }, { "IQ1_S", LLAMA_FTYPE_MOSTLY_IQ1_S, " 1.56 bpw quantization", }, { "IQ1_M", LLAMA_FTYPE_MOSTLY_IQ1_M, " 1.75 bpw quantization", }, - { "IQ1_BN", LLAMA_FTYPE_MOSTLY_IQ1_BN, " 1.75 bpw quantization (Bitnet)", }, + { "IQ1_BN", LLAMA_FTYPE_MOSTLY_IQ1_BN, " 1.62 bpw quantization (Bitnet)", }, { "IQ2_BN", LLAMA_FTYPE_MOSTLY_IQ2_BN, " 2.00 bpw quantization (Bitnet)", }, { "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", }, |