Disable LLama 3b since it ooms on GH instances

huggingface · Dec 18, 2024 · 8f60d66 · 8f60d66
1 parent 09965ae
commit 8f60d66
Show file tree

Hide file tree

Showing 3 changed files with 5 additions and 4 deletions.
diff --git a/tests/executorch/runtime/test_modeling_gemma.py b/tests/executorch/runtime/test_modeling_gemma.py
@@ -32,7 +32,7 @@ def __init__(self, *args, **kwargs):
     @slow
     @pytest.mark.run_slow
     def test_gemma_text_generation_with_xnnpack(self):
-        # TODO: Swithc to use google/gemma-2b once https://github.com/huggingface/optimum/issues/2127 is fixed
+        # TODO: Switch to use google/gemma-2b once https://github.com/huggingface/optimum/issues/2127 is fixed
         # model_id = "google/gemma-2b"
         model_id = "weqweasdas/RM-Gemma-2B"
         model = ExecuTorchModelForCausalLM.from_pretrained(

diff --git a/tests/executorch/runtime/test_modeling_gemma2.py b/tests/executorch/runtime/test_modeling_gemma2.py
@@ -32,7 +32,7 @@ def __init__(self, *args, **kwargs):
     @slow
     @pytest.mark.run_slow
     def test_gemma2_text_generation_with_xnnpack(self):
-        # TODO: Swithc to use google/gemma-2-2b once https://github.com/huggingface/optimum/issues/2127 is fixed
+        # TODO: Switch to use google/gemma-2-2b once https://github.com/huggingface/optimum/issues/2127 is fixed
         # model_id = "google/gemma-2-2b"
         model_id = "unsloth/gemma-2-2b-it"
         model = ExecuTorchModelForCausalLM.from_pretrained(

diff --git a/tests/executorch/runtime/test_modeling_llama.py b/tests/executorch/runtime/test_modeling_llama.py
@@ -32,7 +32,7 @@ def __init__(self, *args, **kwargs):
     @slow
     @pytest.mark.run_slow
     def test_llama3_2_1b_text_generation_with_xnnpack(self):
-        # TODO: Swithc to use meta-llama/Llama-3.2-1B once https://github.com/huggingface/optimum/issues/2127 is fixed
+        # TODO: Switch to use meta-llama/Llama-3.2-1B once https://github.com/huggingface/optimum/issues/2127 is fixed
         # model_id = "lama/Llama-3.2-1B"
         model_id = "NousResearch/Llama-3.2-1B"
         model = ExecuTorchModelForCausalLM.from_pretrained(
@@ -55,8 +55,9 @@ def test_llama3_2_1b_text_generation_with_xnnpack(self):
 
     @slow
     @pytest.mark.run_slow
+    @pytest.mark.skip(reason="OOMs with macos-15 CI instances on GH.")
     def test_llama3_2_3b_text_generation_with_xnnpack(self):
-        # TODO: Swithc to use meta-llama/Llama-3.2-3B once https://github.com/huggingface/optimum/issues/2127 is fixed
+        # TODO: Switch to use meta-llama/Llama-3.2-3B once https://github.com/huggingface/optimum/issues/2127 is fixed
         # model_id = "lama/Llama-3.2-3B"
         model_id = "NousResearch/Hermes-3-Llama-3.2-3B"
         model = ExecuTorchModelForCausalLM.from_pretrained(