update test_models.py to conform to new quantization config

2025-04-21 11:34:37 -04:00
parent ffd4ef1ece
commit 9be971d47c
1 changed files with 4 additions and 2 deletions
--- a/tests/utils/test_models.py
+++ b/tests/utils/test_models.py
@@ -21,8 +21,10 @@ class TestModelsUtils:
                "base_model": "JackFram/llama-68m",
                "model_type": "LlamaForCausalLM",
                "tokenizer_type": "LlamaTokenizer",
-                "load_in_8bit": True,
-                "load_in_4bit": False,
+                "quantization": {
+                    "backend": "bnb",
+                    "bits": 8,
+                },
                "adapter": "lora",
                "flash_attention": False,
                "sample_packing": True,