[CI/Build] Split pooling and generation extended language models tests in CI

.buildkite/test-pipeline.yaml

@@ -486,16 +486,25 @@ steps:
     - pip freeze | grep -E 'torch'
     - pytest -v -s models/language -m core_model
 
-- label: Language Models Test (Extended)
+- label: Language Models Test (Extended Generation) # 1hr20min
   mirror_hardwares: [amdexperimental]
   optional: true
   source_file_dependencies:
   - vllm/
-  - tests/models/language
+  - tests/models/language/generation
   commands:
     # Install causal-conv1d for plamo2 models here, as it is not compatible with pip-compile.
     - pip install 'git+https://github.com/Dao-AILab/[email protected]'
-    - pytest -v -s models/language -m 'not core_model'
+    - pytest -v -s models/language/generation -m 'not core_model'
+
+- label: Language Models Test (Extended Pooling)  # 36min
+  mirror_hardwares: [amdexperimental]
+  optional: true
+  source_file_dependencies:
+  - vllm/
+  - tests/models/language/pooling
+  commands:
+    - pytest -v -s models/language/pooling -m 'not core_model'
 
 - label: Multi-Modal Models Test (Standard)
   mirror_hardwares: [amdexperimental]