[V1][Core] Support for Structured Outputs (vllm-project#12388)

Signed-off-by: Aaron Pham <[email protected]>
Signed-off-by: Russell Bryant <[email protected]>
Co-authored-by: Russell Bryant <[email protected]>
Co-authored-by: Michael Goin <[email protected]>
Co-authored-by: Nick Hill <[email protected]>

Author: 4 people

.buildkite/test-pipeline.yaml

@@ -204,6 +204,7 @@ steps:
     - VLLM_USE_V1=1 pytest -v -s v1/engine
     - VLLM_USE_V1=1 pytest -v -s v1/sample
     - VLLM_USE_V1=1 pytest -v -s v1/worker
+    - VLLM_USE_V1=1 pytest -v -s v1/structured_output
     - VLLM_USE_V1=1 pytest -v -s v1/test_stats.py
     - VLLM_USE_V1=1 pytest -v -s v1/test_utils.py
     # TODO: accuracy does not match, whether setting

.gitignore

@@ -197,7 +197,7 @@ _build/
 hip_compat.h
 
 # Benchmark dataset
-benchmarks/*.json
+benchmarks/**/*.json
 
 # Linting
 actionlint