add codestral 2b code

Signed-off-by: FredricZ-2007 <[email protected]>

Author: fredricz-20070104

tests/integration/defs/accuracy/references/cnn_dailymail.yaml

@@ -45,8 +45,10 @@ microsoft/Phi-3.5-mini-instruct:
   - accuracy: 31.354
 microsoft/Phi-4-mini-instruct:
   - accuracy: 32.921
-codestral/codestral-22b-v0.1:
-  - accuracy: 28.90
+mistralai/Codestral-22B-v0.1:
+  - accuracy: 30.316
+  - quant_algo: FP8
+    accuracy: 30.316
 state-spaces/mamba-130m-hf:
   - accuracy: 19.470
 lmsys/vicuna-7b-v1.3:

tests/integration/defs/accuracy/references/gsm8k.yaml

@@ -158,8 +158,8 @@ microsoft/Phi-4-multimodal-instruct-long-rope:
   - accuracy: 75.85
 microsoft/Phi-4-mini-instruct:
   - accuracy: 82.30
-codestral/codestral-22b-v0.1:
-  - accuracy: 28.90
+mistralai/Codestral-22B-v0.1:
+  - accuracy: 67.10
 GPT-OSS/BF16:
   - accuracy: 90.3
 GPT-OSS/MXFP4:

tests/integration/defs/accuracy/references/mmlu.yaml

@@ -229,8 +229,10 @@ nvidia/Nemotron-H-56B-Base-8K:
     accuracy: 83.82
 microsoft/Phi-4-mini-instruct:
   - accuracy: 68.98
-codestral/codestral-22b-v0.1:
-  - accuracy: 28.90
+mistralai/Codestral-22B-v0.1:
+  - accuracy: 61.72
+  - quant_algo: FP8
+    accuracy: 61.72
 # Created a dummy accuracy to track tp_size=2 for phi4-mini model.
 # TODO: update once https://nvbugs/5393849 is fixed.
 microsoft/Phi-4-mini-instruct-tp2:

tests/integration/defs/accuracy/test_llm_api.py

@@ -433,3 +433,29 @@ def test_auto_dtype(self):
                 speculative_config=self.speculative_config) as llm:
             task = CnnDailymail(self.MODEL_NAME)
             task.evaluate(llm)
+
+
+class TestCodestral_22B_V01(LlmapiAccuracyTestHarness):
+    MODEL_NAME = "mistralai/Codestral-22B-v0.1"
+    MODEL_PATH = f"{llm_models_root()}/Codestral-22B-v0.1"
+    kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.6)
+
+    @pytest.mark.skip_less_device_memory(80000)
+    def test_auto_dtype(self):
+        with LLM(self.MODEL_PATH, kv_cache_config=self.kv_cache_config) as llm:
+            task = CnnDailymail(self.MODEL_NAME)
+            task.evaluate(llm)
+            task = MMLU(self.MODEL_NAME)
+            task.evaluate(llm)
+
+    @skip_pre_ada
+    @pytest.mark.skip_less_device_memory(80000)
+    def test_fp8(self):
+        quant_config = QuantConfig(QuantAlgo.FP8)
+        with LLM(self.MODEL_PATH,
+                 quant_config=quant_config,
+                 kv_cache_config=self.kv_cache_config) as llm:
+            task = CnnDailymail(self.MODEL_NAME)
+            task.evaluate(llm)
+            task = MMLU(self.MODEL_NAME)
+            task.evaluate(llm)

tests/integration/defs/accuracy/test_llm_api_pytorch.py

@@ -2437,11 +2437,14 @@ def test_auto_dtype(self):
 
 
 class TestCodestral_22B_V01(LlmapiAccuracyTestHarness):
-    MODEL_NAME = "codestral/codestral-22b-v0.1"
-    MODEL_PATH = f"{llm_models_root()}/codestral-22b-v0.1"
+    MODEL_NAME = "mistralai/Codestral-22B-v0.1"
+    MODEL_PATH = f"{llm_models_root()}/Codestral-22B-v0.1"
 
+    @pytest.mark.timeout(2400)
+    @pytest.mark.skip_less_device_memory(80000)
     def test_auto_dtype(self):
-        with LLM(self.MODEL_PATH) as llm:
+        kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.6)
+        with LLM(self.MODEL_PATH, kv_cache_config=kv_cache_config) as llm:
             task = CnnDailymail(self.MODEL_NAME)
             task.evaluate(llm)
             task = MMLU(self.MODEL_NAME)