From 39944e1cbb296f1ad906db0f85e0d2e8bfe7c455 Mon Sep 17 00:00:00 2001
From: roc <xupeng@patsnap.com>
Date: Mon, 8 Jul 2024 14:47:51 +0800
Subject: [PATCH] fix dbrx & opt model prefix bug

---
 .../models/custom_modeling/flash_dbrx_modeling.py               | 2 +-
 .../models/custom_modeling/opt_modeling.py                      | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/server/text_generation_server/models/custom_modeling/flash_dbrx_modeling.py b/server/text_generation_server/models/custom_modeling/flash_dbrx_modeling.py
index e469495f..41aa5859 100644
--- a/server/text_generation_server/models/custom_modeling/flash_dbrx_modeling.py
+++ b/server/text_generation_server/models/custom_modeling/flash_dbrx_modeling.py
@@ -711,7 +711,7 @@ class FlashDbrxForCausalLM(torch.nn.Module):
         else:
             prefix = f"{prefix}.transformer"
 
-        self.model = DbrxModel(config, weights)
+        self.model = DbrxModel(prefix, config, weights)
         self.lm_head = SpeculativeHead.load(
             config,
             prefix="lm_head",
diff --git a/server/text_generation_server/models/custom_modeling/opt_modeling.py b/server/text_generation_server/models/custom_modeling/opt_modeling.py
index 5ab02959..84a1c069 100644
--- a/server/text_generation_server/models/custom_modeling/opt_modeling.py
+++ b/server/text_generation_server/models/custom_modeling/opt_modeling.py
@@ -757,7 +757,7 @@ class OPTForCausalLM(OPTPreTrainedModel):
         else:
             prefix = f"{prefix}.model"
 
-        self.model = OPTModel(config, weights)
+        self.model = OPTModel(prefix, config, weights)
 
         self.lm_head = SpeculativeHead.load(
             config, prefix=f"{prefix}.decoder.embed_tokens", weights=weights