From f437b3667882b6cfd1f23a8e048f7c181949b704 Mon Sep 17 00:00:00 2001 From: Guancheng Fu <110874468+gc-fu@users.noreply.github.com> Date: Wed, 26 Mar 2025 09:25:32 +0800 Subject: [PATCH] Fix vllm glm edge model (#13007) * fix done * fix --- python/llm/src/ipex_llm/transformers/convert.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/python/llm/src/ipex_llm/transformers/convert.py b/python/llm/src/ipex_llm/transformers/convert.py index 4a025081..305e69e7 100644 --- a/python/llm/src/ipex_llm/transformers/convert.py +++ b/python/llm/src/ipex_llm/transformers/convert.py @@ -192,7 +192,8 @@ def is_linear_module(module): and hasattr(module.quant_method, "quant_config") and module.quant_method.quant_config.get_name() == "gptq"): _USE_VLLM_GPTQ = True - invalidInputError(module.skip_bias_add is not True, "Currently, ipex-vllm does not" + invalidInputError(module.skip_bias_add is not True or module.bias is None, + "Currently, ipex-vllm does not" " support linear layers with skip_bias_add argument") if isinstance(module, RowParallelLinear) and tp_size >= 2: mp_group = get_tensor_model_parallel_group()