diff options
Diffstat (limited to 'examples')
| -rw-r--r-- | examples/quantize-stats/quantize-stats.cpp | 8 | ||||
| -rw-r--r-- | examples/quantize/quantize.cpp | 1 |
2 files changed, 9 insertions, 0 deletions
diff --git a/examples/quantize-stats/quantize-stats.cpp b/examples/quantize-stats/quantize-stats.cpp index 746df844..4eb8f953 100644 --- a/examples/quantize-stats/quantize-stats.cpp +++ b/examples/quantize-stats/quantize-stats.cpp @@ -341,6 +341,10 @@ int main(int argc, char ** argv) { if (!layer_included(params, kv_tensor.first)) { continue; } + if (kv_tensor.second->ne[0] == 1 || kv_tensor.second->ne[1] == 1) { + // we never quantize those + continue; + } if (params.verbose) { printf("%s: type %s, size %" PRId64 "\n", kv_tensor.first.c_str(), ggml_type_name(kv_tensor.second->type), ggml_nelements(kv_tensor.second)); } @@ -386,6 +390,10 @@ int main(int argc, char ** argv) { if (!layer_included(params, kv_tensor.first)) { continue; } + if (kv_tensor.second->ne[0] == 1 || kv_tensor.second->ne[1] == 1) { + // we never quantize those + continue; + } if (params.verbose) { printf(" %s ...\n", kv_tensor.first.c_str()); } diff --git a/examples/quantize/quantize.cpp b/examples/quantize/quantize.cpp index 06927890..a5ffb2b2 100644 --- a/examples/quantize/quantize.cpp +++ b/examples/quantize/quantize.cpp @@ -27,6 +27,7 @@ static const std::vector<struct quant_option> QUANT_OPTIONS = { { "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)", }, + { "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", }, { "IQ3_XXS",LLAMA_FTYPE_MOSTLY_IQ3_XXS," 3.06 bpw quantization", }, |