From ea65a3f19c16e9c30eba121e3e91327dc1b8e975 Mon Sep 17 00:00:00 2001 From: Daniel Han Date: Wed, 17 Dec 2025 01:51:29 -0800 Subject: [PATCH] Update loader.py --- unsloth/models/loader.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/unsloth/models/loader.py b/unsloth/models/loader.py index e1c13315f7..bfa94d86d7 100644 --- a/unsloth/models/loader.py +++ b/unsloth/models/loader.py @@ -739,6 +739,8 @@ class FastModel(FastBaseModel): "compatible with `full_finetuning=True`. If you wish to use QAT with LoRA, " "please pass in `qat_scheme` in `FastLanguageModel.get_peft_model(...)` instead." ) + if qat_scheme == "phone-deployment": + qat_scheme = "int8-int4" # Check if 4bit is allowed specifically for AMD if not ALLOW_BITSANDBYTES and not use_exact_model_name: if load_in_4bit or load_in_8bit or model_name.lower().endswith("-bnb-4bit"):