PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度
adjust 3 bits hyperparameters at auto-round-best (#1081)
Wenhua Chengef78c688dafa
diff --git a/auto_round/compressors/base.py b/auto_round/compressors/base.pyindex 17470e2ae..7abcbc511 100644--- a/auto_round/compressors/base.py+++ b/auto_round/compressors/base.py@@ -342,12 +342,19 @@ class BaseCompressor(object):self.lr = 5e-3else:if not lr:- # TODO need to check 3/4 bits lr setting for auto-round-best- self.lr = 2.0 / self.iters if (self.iters >= 1000 and self.bits == 2) else 1.0 / self.iters- if self.iters >= 1000 and self.bits == 2:+ # TODO need to check 4 bits lr setting for auto-round-best, 3bits only validate on small models+ if self.iters >= 1000 and self.bits <= 3:+ self.lr = 2.0 / self.iterslogger.info("set the lr to 2.0/iters for better accuracy")+ else:+ self.lr = 1.0 / self.iterselse:self.lr = lr+ if self.bits <= 2 and (self.iters < 1000 or not enable_alg_ext):+ logger.warning(+ "For bits <= 2, it is recommended to enable `auto-round-best` " "and set `--enable_alg_ext` "+ )+self.minmax_lr = minmax_lr or self.lrself.enable_alg_ext = enable_alg_extself.not_use_best_mse = not_use_best_mse@@ -740,6 +747,8 @@ class BaseCompressor(object):" Please refer to https://github.com/intel/auto-round/tree/main/docs/gguf_alg_ext_acc.md"" for the accuracy results.")+ elif self.bits >= 8 and self.iters != 0:+ logger.warning("`iters=0` is recommended for bits>=8")if (self.seqlen is not None