diff --git a/vllm/model_executor/models/starcoder2.py b/vllm/model_executor/models/starcoder2.py index e5003361bdf2a..e72c5cf1544f7 100644 --- a/vllm/model_executor/models/starcoder2.py +++ b/vllm/model_executor/models/starcoder2.py @@ -141,8 +141,9 @@ def __init__(self, bias=config.use_bias, linear_method=linear_method, ) - self.act = get_act_fn(config.hidden_act, - intermediate_size=config.intermediate_size) + quant_config = getattr(linear_method, "quant_config", None) + self.act = get_act_fn(config.hidden_act, quant_config, + config.intermediate_size) def forward(self, hidden_states: torch.Tensor) -> torch.Tensor: hidden_states, _ = self.c_fc(hidden_states)