vllm/.buildkite/test_areas/spec_decode.yaml

group: Spec Decode
depends_on:
  - image-build
steps:
- label: Spec Decode Eagle
  timeout_in_minutes: 30
  device: h200_18gb
  source_file_dependencies:
    - vllm/v1/spec_decode/
    - vllm/v1/worker/gpu/spec_decode/
    - tests/v1/e2e/spec_decode/
  commands:
    - pytest -v -s v1/e2e/spec_decode -k "eagle_correctness"

- label: Spec Decode Eagle Nightly B200
  timeout_in_minutes: 30
  device: b200
  optional: true
  source_file_dependencies:
    - vllm/v1/spec_decode/
    - vllm/v1/worker/gpu/spec_decode/
    - tests/v1/e2e/spec_decode/
  commands:
    - pytest -v -s v1/e2e/spec_decode -k "eagle_correctness"

- label: Spec Decode Speculators + MTP
  timeout_in_minutes: 30
  device: h200_18gb
  source_file_dependencies:
    - vllm/v1/spec_decode/
    - vllm/v1/worker/gpu/spec_decode/
    - vllm/transformers_utils/configs/speculators/
    - tests/v1/e2e/spec_decode/
  commands:
    - pytest -v -s v1/e2e/spec_decode -k "speculators or mtp_correctness"

- label: Spec Decode Speculators + MTP Nightly B200
  timeout_in_minutes: 30
  device: b200
  optional: true
  source_file_dependencies:
    - vllm/v1/spec_decode/
    - vllm/v1/worker/gpu/spec_decode/
    - vllm/transformers_utils/configs/speculators/
    - tests/v1/e2e/spec_decode/
  commands:
    - pytest -v -s v1/e2e/spec_decode -k "speculators or mtp_correctness"
  
- label: Spec Decode Ngram + Suffix
  timeout_in_minutes: 30
  device: h200_18gb
  source_file_dependencies:
    - vllm/v1/spec_decode/
    - vllm/v1/worker/gpu/spec_decode/
    - tests/v1/e2e/spec_decode/
  commands:
    - pytest -v -s v1/e2e/spec_decode -k "ngram or suffix"

- label: Spec Decode Draft Model
  timeout_in_minutes: 30
  device: h200_18gb
  source_file_dependencies:
    - vllm/v1/spec_decode/
    - vllm/v1/worker/gpu/spec_decode/
    - tests/v1/e2e/spec_decode/
  commands:
    - pytest -v -s v1/e2e/spec_decode -k "draft_model or no_sync or batch_inference"

- label: Spec Decode Draft Model Nightly B200
  timeout_in_minutes: 30
  device: b200
  optional: true
  source_file_dependencies:
    - vllm/v1/spec_decode/
    - vllm/v1/worker/gpu/spec_decode/
    - tests/v1/e2e/spec_decode/
  commands:
    - pytest -v -s v1/e2e/spec_decode -k "draft_model or no_sync or batch_inference"