diff --git a/colossalai/inference/modeling/models/glide_llama.py b/colossalai/inference/modeling/models/glide_llama.py
index 0ee78a303004..10f27d23d372 100644
--- a/colossalai/inference/modeling/models/glide_llama.py
+++ b/colossalai/inference/modeling/models/glide_llama.py
@@ -12,9 +12,7 @@
     LlamaAttention,
     LlamaConfig,
     LlamaDecoderLayer,
-    LlamaDynamicNTKScalingRotaryEmbedding,
     LlamaForCausalLM,
-    LlamaLinearScalingRotaryEmbedding,
     LlamaMLP,
     LlamaModel,
     LlamaRMSNorm,
diff --git a/requirements/requirements-test.txt b/requirements/requirements-test.txt
index 3fcf53e1858e..b60a65eca74a 100644
--- a/requirements/requirements-test.txt
+++ b/requirements/requirements-test.txt
@@ -18,3 +18,4 @@ datasets
 pydantic
 ray
 peft>=0.7.1
+transformers==4.51.3
diff --git a/tests/test_infer/test_drafter.py b/tests/test_infer/test_drafter.py
index 3c5dda1578a2..bb330786bd49 100644
--- a/tests/test_infer/test_drafter.py
+++ b/tests/test_infer/test_drafter.py
@@ -2,7 +2,7 @@
 import torch
 from transformers import AutoTokenizer, LlamaConfig, LlamaForCausalLM
 
-from colossalai.inference.modeling.models.glide_llama import GlideLlamaConfig, GlideLlamaForCausalLM
+# from colossalai.inference.modeling.models.glide_llama import GlideLlamaConfig, GlideLlamaForCausalLM
 from colossalai.inference.spec.drafter import Drafter
 from colossalai.utils import get_current_device