[CI] Ensure documentation build is checked in CI

.buildkite/test-pipeline.yaml

@@ -49,3 +49,10 @@ steps:
   commands:
   - pip install aiohttp
   - bash run-benchmarks.sh
+
+- label: Documentation Build
+  working_dir: "/vllm-workspace/docs"
+  no_gpu: True
+  commands:
+  - pip install -r requirements-docs.txt
+  - SPHINXOPTS=\"-W\" make html

.buildkite/test-template.j2

@@ -35,13 +35,15 @@ steps:
               - image: "{{ docker_image }}"
                 command: ["bash"]
                 args:
-                - "-c"
+                - '-c'
                 - "'cd {{ (step.working_dir or default_working_dir) | safe  }} && {{ step.command  or (step.commands | join(' && ')) | safe }}'"
+                {% if not step.no_gpu %}
                 resources:
                   requests:
                     nvidia.com/gpu: "{{ step.num_gpus or default_num_gpu }}"
                   limits:
                     nvidia.com/gpu: "{{ step.num_gpus or default_num_gpu }}"
+                {% endif %}
                 env:
                   - name: HF_TOKEN
                     valueFrom:

docs/source/conf.py

@@ -94,3 +94,5 @@ def add_line(self, line: str, source: str, *lineno: int) -> None:
 
 
 autodoc.ClassDocumenter = MockedClassDocumenter
+
+navigation_with_keys = False

docs/source/index.rst

@@ -89,6 +89,7 @@ Documentation
    :caption: Quantization
 
    quantization/auto_awq
+   quantization/fp8_e5m2_kv_cache
 
 .. toctree::
    :maxdepth: 2

docs/source/quantization/fp8_e5m2_kv_cache.rst

@@ -9,6 +9,7 @@ The FP8 data format retains 2~3 mantissa bits and can convert float/fp16/bflaot1
 Here is an example of how to enable this feature:
 
 .. code-block:: python
+
     from vllm import LLM, SamplingParams
     # Sample prompts.
     prompts = [