From dbc8a65a5d69c795f0798d01ad88f8ccbf45dd1f Mon Sep 17 00:00:00 2001 From: Nicolas Patry Date: Mon, 13 May 2024 10:32:29 +0000 Subject: [PATCH] Old config support ? --- .../models/custom_modeling/flash_llama_modeling.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/server/text_generation_server/models/custom_modeling/flash_llama_modeling.py b/server/text_generation_server/models/custom_modeling/flash_llama_modeling.py index 1869db9e..6a6b2e0a 100644 --- a/server/text_generation_server/models/custom_modeling/flash_llama_modeling.py +++ b/server/text_generation_server/models/custom_modeling/flash_llama_modeling.py @@ -193,7 +193,7 @@ class LlamaMLP(nn.Module): ) ) # Fuse gate and up proj - bias = config.mlp_bias + bias = getattr(config, "mlp_bias", False) if config.model_type == "phi3": self.gate_up_proj = TensorParallelColumnLinear.load_gate_up( config,