raif-vllm 0.1.0__tar.gz

raif_vllm-0.1.0/.github/workflows/ci.yml

@@ -0,0 +1,60 @@
+name: CI
+
+# Lint + test for the `raif-vllm` plugin. GPU-free: the 3 vLLM-shim tests
+# `importorskip` and skip cleanly on the runner (no vLLM/CUDA), while the pure
+# decode/inject/route/structured logic runs in full. raif-format (>=0.6, with
+# schema_bridge/stream) installs from PyPI as a normal dependency.
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+
+# Default-deny at the top level. The single job re-grants only read access.
+permissions:
+  contents: read
+
+concurrency:
+  group: ci-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  test:
+    name: Lint & test (py${{ matrix.python-version }})
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: ["3.11", "3.12", "3.13"]
+
+    steps:
+      - name: Harden runner
+        # pin to SHA
+        uses: step-security/harden-runner@9af89fc71515a100421586dfdb3dc9c984fbf411 # v2.19.4
+        with:
+          egress-policy: audit
+
+      - name: Checkout code
+        # pin to SHA
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          persist-credentials: false
+
+      - name: Setup Python ${{ matrix.python-version }}
+        # pin to SHA
+        uses: actions/setup-python@e9d6f990972a57673cdb72ec29e19d42ba28880f # v6.0.0
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install the plugin (pulls raif-format>=0.6 from PyPI) + dev deps
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install . ruff pytest
+
+      - name: Lint (ruff)
+        run: ruff check
+
+      - name: Test (pytest)
+        run: PYTHONPATH=. pytest tests -q

raif_vllm-0.1.0/.github/workflows/release.yml

@@ -0,0 +1,139 @@
+name: Release
+
+# Release pipeline for the `raif-vllm` package via PyPI Trusted Publishing
+# (OIDC; no long-lived API token). PEP 740 attestations are attached
+# automatically by the publish action under OIDC.
+#
+# Tags `vllm-v*`            -> build + publish to PyPI.
+# Branch `release/*` push   -> build + publish to TestPyPI (rehearsal).
+#
+# ── ONE-TIME MANUAL SETUP (required before this workflow can publish) ──────────
+#   1. Register the Trusted Publisher on PyPI for project `raif-vllm`:
+#        PyPI -> "Publishing" / project settings -> add a GitHub Actions
+#        trusted publisher with:
+#          Owner:               skrrt-sh
+#          Repository:          raif-vllm
+#          Workflow filename:   release.yml
+#          Environment:         pypi
+#      Do the same on test.pypi.org with environment `testpypi` (optional).
+#   2. Create the `pypi` (and `testpypi`) GitHub Environments on the repo
+#      (Settings -> Environments) and add any desired protection rules.
+#   3. NAME AVAILABILITY: PyPI cannot attach a trusted publisher to a project
+#      that does not yet exist. For this brand-new name, use PyPI's "pending
+#      publisher" flow (register the publisher before the first upload) so the
+#      very first release works over OIDC with no manual token publish.
+#
+#   Then release with:  git tag vllm-v0.1.0 && git push origin vllm-v0.1.0
+# ──────────────────────────────────────────────────────────────────────────────
+
+on:
+  push:
+    tags:
+      - "vllm-v*"
+    branches:
+      - "release/*"
+
+# Default-deny at the top level. Each job re-grants the minimum it needs.
+permissions: {}
+
+concurrency:
+  group: release-${{ github.ref }}
+  cancel-in-progress: false
+
+jobs:
+  build:
+    name: Build sdist + wheel
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+
+    steps:
+      - name: Harden runner
+        # pin to SHA
+        uses: step-security/harden-runner@9af89fc71515a100421586dfdb3dc9c984fbf411 # v2.19.4
+        with:
+          egress-policy: audit
+
+      - name: Checkout code
+        # pin to SHA
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          persist-credentials: false
+
+      - name: Setup uv
+        # pin to SHA
+        uses: astral-sh/setup-uv@e92bafb6253dcd438e0484186d7669ea7a8ca1cc # v6.4.3
+        with:
+          version: "0.11.14"
+          # No action caching in the release pipeline — keep the supply chain lean.
+          enable-cache: false
+
+      - name: Build distributions
+        run: uv build
+
+      - name: Upload distributions
+        # pin to SHA
+        uses: actions/upload-artifact@330a01c490aca151604b8cf639adc76d48f6c5d4 # v5.0.0
+        with:
+          name: python-package-distributions
+          path: dist/
+
+  publish-to-testpypi:
+    name: Publish to TestPyPI
+    needs: build
+    if: startsWith(github.ref, 'refs/heads/release/')
+    runs-on: ubuntu-latest
+    environment: testpypi
+    permissions:
+      id-token: write # PyPI Trusted Publishing (OIDC)
+      contents: read
+
+    steps:
+      - name: Harden runner
+        # pin to SHA
+        uses: step-security/harden-runner@9af89fc71515a100421586dfdb3dc9c984fbf411 # v2.19.4
+        with:
+          egress-policy: audit
+
+      - name: Download distributions
+        # pin to SHA
+        uses: actions/download-artifact@018cc2cf5baa6db3ef3c5f8a56943fffe632ef53 # v6.0.0
+        with:
+          name: python-package-distributions
+          path: dist/
+
+      - name: Publish to TestPyPI
+        # pin to SHA — Trusted Publishing: no username/password.
+        uses: pypa/gh-action-pypi-publish@76f52bc884231f62b9a034ebfe128415bbaabdfc # v1.12.4
+        with:
+          repository-url: https://test.pypi.org/legacy/
+          skip-existing: true
+
+  publish-to-pypi:
+    name: Publish to PyPI
+    needs: build
+    if: startsWith(github.ref, 'refs/tags/vllm-v')
+    runs-on: ubuntu-latest
+    environment: pypi
+    permissions:
+      id-token: write # PyPI Trusted Publishing (OIDC)
+      contents: read
+
+    steps:
+      - name: Harden runner
+        # pin to SHA
+        uses: step-security/harden-runner@9af89fc71515a100421586dfdb3dc9c984fbf411 # v2.19.4
+        with:
+          egress-policy: audit
+
+      - name: Download distributions
+        # pin to SHA
+        uses: actions/download-artifact@018cc2cf5baa6db3ef3c5f8a56943fffe632ef53 # v6.0.0
+        with:
+          name: python-package-distributions
+          path: dist/
+
+      - name: Publish to PyPI
+        # pin to SHA — Trusted Publishing: no username/password. PEP 740
+        # attestations are generated and attached automatically under OIDC.
+        uses: pypa/gh-action-pypi-publish@76f52bc884231f62b9a034ebfe128415bbaabdfc # v1.12.4

raif_vllm-0.1.0/.gitignore

@@ -0,0 +1,7 @@
+__pycache__/
+*.pyc
+.ruff_cache/
+.pytest_cache/
+.venv/
+dist/
+*.egg-info/

raif_vllm-0.1.0/LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2026 skrrt-sh
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

raif_vllm-0.1.0/PKG-INFO

@@ -0,0 +1,107 @@
+Metadata-Version: 2.4
+Name: raif-vllm
+Version: 0.1.0
+Summary: One vLLM plugin for transparent RAIF token savings — install it and existing OpenAI clients get RAIF on tools & response_format with no proxy and no client changes.
+Project-URL: Repository, https://github.com/skrrt-sh/raif-vllm
+Project-URL: Homepage, https://github.com/skrrt-sh/raif-vllm
+Author-email: truehazker <40111175+truehazker@users.noreply.github.com>
+License-Expression: Apache-2.0
+License-File: LICENSE
+Keywords: llm,plugin,raif,response-format,structured-output,tool-call,vllm
+Classifier: Development Status :: 3 - Alpha
+Classifier: Intended Audience :: Developers
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Software Development :: Libraries :: Python Modules
+Classifier: Typing :: Typed
+Requires-Python: >=3.11
+Requires-Dist: raif-format>=0.6
+Provides-Extra: vllm
+Requires-Dist: vllm<0.20,>=0.19; extra == 'vllm'
+Description-Content-Type: text/markdown
+
+# raif-vllm
+
+One vLLM plugin for transparent RAIF token savings. Install it and existing
+OpenAI clients get RAIF on `tools` and `response_format` — **no proxy, no client
+changes, no vLLM fork**. The fine-tuned model emits compact RAIF-G; the plugin
+decodes it to JSON at the request/response boundary.
+
+## Install
+
+`raif-vllm` is not yet on PyPI; install it from the repo (it pulls
+[`raif-format`](https://pypi.org/project/raif-format/) `>=0.6` from PyPI
+automatically):
+
+```sh
+pip install "raif-vllm @ git+https://github.com/skrrt-sh/raif-vllm.git"
+# or, from a checkout:  pip install -e .
+```
+
+vLLM itself is provided by the serving host (it pins CUDA/torch); target
+**`vllm>=0.19,<0.20`** — v0.19 is the last CUDA-12 vLLM and carries the hooks the
+plugin needs. `pip install "raif-vllm[vllm]"` pulls a compatible engine for local
+experiments.
+
+## Serve
+
+```sh
+VLLM_PLUGINS=raif vllm serve unsloth/Llama-3.2-3B-Instruct \
+  --enable-lora --lora-modules raif=skrrt-sh/raif-llama-3.2-3b-lora \
+  --max-lora-rank 32 --max-model-len 8192 \
+  --chat-template raif_llama32.jinja \
+  --reasoning-parser raif \
+  --enable-auto-tool-choice --tool-call-parser raif
+```
+
+- `VLLM_PLUGINS=raif` runs the entry point, which registers the `raif` reasoning +
+  tool parsers **and** installs the `render_chat` inject hook (the seam that adds
+  the compact `<schema>` cue before chat-templating).
+- `--tool-call-parser raif` decodes the tools path into `tool_calls`;
+  `--reasoning-parser raif` decodes the `response_format` path into
+  `message.content`.
+- `--chat-template raif_llama32.jinja` (in `chat_templates/`) is load-bearing: it
+  renders messages only and ignores the `tools` variable, so the served prompt
+  matches training. Without it the LoRA echoes the verbose OpenAI tool-def JSON.
+
+## What a plain OpenAI client gets
+
+```python
+from openai import OpenAI
+client = OpenAI(base_url="http://localhost:8000/v1", api_key="EMPTY")
+
+# tools -> JSON tool_calls
+client.chat.completions.create(model="raif", tools=[...], tool_choice="auto",
+    messages=[{"role": "user", "content": "Weather in Oslo?"}])
+
+# response_format -> JSON content (use non-streaming — see below)
+client.chat.completions.create(model="raif",
+    response_format={"type": "json_schema", "json_schema": {...}},
+    messages=[{"role": "user", "content": "..."}])
+```
+
+| OpenAI path | Behavior |
+|---|---|
+| plain chat | passthrough, untouched |
+| `tools` | RAIF-G → JSON `tool_calls` (streaming + non-streaming) |
+| `response_format` (`json_schema` / `json_object`) | RAIF-G → JSON `message.content` |
+| plain chat **streaming** | passthrough |
+
+### Known limitation: streaming `response_format`
+
+Streaming a `response_format` request is **not decoded** — the client receives raw
+RAIF-G. (vLLM's streaming seam passes the parser no schema, and the shared
+`is_reasoning_end` flag must stay `True` so the *tools* streaming path keeps
+working.) **Use non-streaming `response_format` for structured output** — it
+decodes fully. Tool-call streaming is unaffected. See
+[`docs/vllm_e2e_results.md`](docs/vllm_e2e_results.md).
+
+## More
+
+- End-to-end GPU smoke: [`scripts/serve_smoke.sh`](scripts/serve_smoke.sh) + [`examples/smoke_plugin.py`](examples/smoke_plugin.py).
+- Serving guide + the chat-template fix: [`docs/vllm_tool_calling.md`](docs/vllm_tool_calling.md).
+- RunPod runbook: [`docs/runpod_testing.md`](docs/runpod_testing.md).
+- The model: the [`skrrt-sh/raif-llama-3.2-3b-lora`](https://huggingface.co/skrrt-sh/raif-llama-3.2-3b-lora) adapter, trained in [`skrrt-sh/raif-lora`](https://github.com/skrrt-sh/raif-lora). The codec: [`raif-format`](https://github.com/skrrt-sh/raif-standard).

raif_vllm-0.1.0/README.md

@@ -0,0 +1,82 @@
+# raif-vllm
+
+One vLLM plugin for transparent RAIF token savings. Install it and existing
+OpenAI clients get RAIF on `tools` and `response_format` — **no proxy, no client
+changes, no vLLM fork**. The fine-tuned model emits compact RAIF-G; the plugin
+decodes it to JSON at the request/response boundary.
+
+## Install
+
+`raif-vllm` is not yet on PyPI; install it from the repo (it pulls
+[`raif-format`](https://pypi.org/project/raif-format/) `>=0.6` from PyPI
+automatically):
+
+```sh
+pip install "raif-vllm @ git+https://github.com/skrrt-sh/raif-vllm.git"
+# or, from a checkout:  pip install -e .
+```
+
+vLLM itself is provided by the serving host (it pins CUDA/torch); target
+**`vllm>=0.19,<0.20`** — v0.19 is the last CUDA-12 vLLM and carries the hooks the
+plugin needs. `pip install "raif-vllm[vllm]"` pulls a compatible engine for local
+experiments.
+
+## Serve
+
+```sh
+VLLM_PLUGINS=raif vllm serve unsloth/Llama-3.2-3B-Instruct \
+  --enable-lora --lora-modules raif=skrrt-sh/raif-llama-3.2-3b-lora \
+  --max-lora-rank 32 --max-model-len 8192 \
+  --chat-template raif_llama32.jinja \
+  --reasoning-parser raif \
+  --enable-auto-tool-choice --tool-call-parser raif
+```
+
+- `VLLM_PLUGINS=raif` runs the entry point, which registers the `raif` reasoning +
+  tool parsers **and** installs the `render_chat` inject hook (the seam that adds
+  the compact `<schema>` cue before chat-templating).
+- `--tool-call-parser raif` decodes the tools path into `tool_calls`;
+  `--reasoning-parser raif` decodes the `response_format` path into
+  `message.content`.
+- `--chat-template raif_llama32.jinja` (in `chat_templates/`) is load-bearing: it
+  renders messages only and ignores the `tools` variable, so the served prompt
+  matches training. Without it the LoRA echoes the verbose OpenAI tool-def JSON.
+
+## What a plain OpenAI client gets
+
+```python
+from openai import OpenAI
+client = OpenAI(base_url="http://localhost:8000/v1", api_key="EMPTY")
+
+# tools -> JSON tool_calls
+client.chat.completions.create(model="raif", tools=[...], tool_choice="auto",
+    messages=[{"role": "user", "content": "Weather in Oslo?"}])
+
+# response_format -> JSON content (use non-streaming — see below)
+client.chat.completions.create(model="raif",
+    response_format={"type": "json_schema", "json_schema": {...}},
+    messages=[{"role": "user", "content": "..."}])
+```
+
+| OpenAI path | Behavior |
+|---|---|
+| plain chat | passthrough, untouched |
+| `tools` | RAIF-G → JSON `tool_calls` (streaming + non-streaming) |
+| `response_format` (`json_schema` / `json_object`) | RAIF-G → JSON `message.content` |
+| plain chat **streaming** | passthrough |
+
+### Known limitation: streaming `response_format`
+
+Streaming a `response_format` request is **not decoded** — the client receives raw
+RAIF-G. (vLLM's streaming seam passes the parser no schema, and the shared
+`is_reasoning_end` flag must stay `True` so the *tools* streaming path keeps
+working.) **Use non-streaming `response_format` for structured output** — it
+decodes fully. Tool-call streaming is unaffected. See
+[`docs/vllm_e2e_results.md`](docs/vllm_e2e_results.md).
+
+## More
+
+- End-to-end GPU smoke: [`scripts/serve_smoke.sh`](scripts/serve_smoke.sh) + [`examples/smoke_plugin.py`](examples/smoke_plugin.py).
+- Serving guide + the chat-template fix: [`docs/vllm_tool_calling.md`](docs/vllm_tool_calling.md).
+- RunPod runbook: [`docs/runpod_testing.md`](docs/runpod_testing.md).
+- The model: the [`skrrt-sh/raif-llama-3.2-3b-lora`](https://huggingface.co/skrrt-sh/raif-llama-3.2-3b-lora) adapter, trained in [`skrrt-sh/raif-lora`](https://github.com/skrrt-sh/raif-lora). The codec: [`raif-format`](https://github.com/skrrt-sh/raif-standard).