From 69a32b6f508a4d0d38f52cf91cc8cd5b42a4bf62 Mon Sep 17 00:00:00 2001 From: Ed Addario Date: Sat, 29 Nov 2025 10:28:43 +0000 Subject: [PATCH] Relax target bpw range --- tools/quantize/quantize.cpp | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/tools/quantize/quantize.cpp b/tools/quantize/quantize.cpp index dd4b860e1b..ebeea65336 100644 --- a/tools/quantize/quantize.cpp +++ b/tools/quantize/quantize.cpp @@ -132,7 +132,7 @@ static void usage(const char * executable) { printf(" Advanced option to selectively quantize tensors. May be specified multiple times.\n"); printf(" --prune-layers L0,L1,L2...comma-separated list of layer numbers to prune from the model\n"); printf(" Advanced option to remove all tensors from the given layers\n"); - printf(" --target-bpw: target bits per weight (bpw). Must be a positive number between 0.0 and 8.0\n"); + printf(" --target-bpw: target bits per weight (bpw). Must be a positive number between 0.0 and 16.0\n"); printf(" Advanced option to automatically select quantization types to achieve a total bits per weight (bpw) target\n"); printf(" --no-importance: distribute bpw budget equitably across all tensors\n"); printf(" Advanced option to disable assigning more bpw budget to important tensors. It may increase quality for some models\n"); @@ -485,13 +485,13 @@ static bool parse_target_bpw(const char * data, float & target_bpw) { try { target_bpw = std::stof(data); - if (target_bpw < 0.0f || target_bpw > 8.0f) { - printf("\n%s: target bits per weight (bpw) must be a positive number between 0.0 and 8.0\n\n", __func__); + if (target_bpw < 0.0f || target_bpw > 16.0f) { + printf("\n%s: target bits per weight (bpw) must be a positive number between 0.0 and 16.0\n\n", __func__); return false; } } catch (const std::exception & e) { - printf("\n%s: '%s' is not valid. Target bits per weight (bpw) must be a positive number between 0.0 and 8.0\n\n", __func__, data); + printf("\n%s: '%s' is not valid. Target bits per weight (bpw) must be a positive number between 0.0 and 16.0\n\n", __func__, data); return false; }