# vLLM Service — Full Reference

Detail split out of `../AGENTS.md` (2026-07-08, to keep that file under the
Agents Settings Kit's 256-line threshold once the `e`/`n` workflow was appended
to it). `AGENTS.md` keeps the short version — architecture, quick start, the
env var table, file map — and links here for everything else.

## Issue recording — naming and template

```
issues/{NN}-{slug}.md
```

| Part | Rule | Example |
|------|------|---------|
| `{NN}` | Two-digit running number (`01`, `02`, …). Increment from the highest existing file. | `03` |
| `{slug}` | Lowercase kebab-case summary of the problem | `gpu-memory-startup-failure` |

Full example: `issues/04-gpu-memory-startup-failure.md`

### File template

```markdown
# Issue {NN}: {Short title}

## Problem
What failed, with exact error message or symptom.

## Context
Environment, command run, relevant config (`.env`, `config/vllm_config.yaml`).

## Solution
What fixed it, or current workaround / open status.

## References
Links, related issue files, or AGENTS.md sections.
```

Check `issues/` for the next number:

```bash
ls issues/*.md 2>/dev/null | sort
```

## Client usage

After the server is running:

```bash
# CLI
uv run paddleocr doc_parser \
  --input https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/paddleocr_vl_demo.png \
  --vl_rec_backend vllm-server \
  --vl_rec_server_url http://localhost:8118/v1
```

```python
from paddleocr import PaddleOCRVL

pipeline = PaddleOCRVL(
    vl_rec_backend="vllm-server",
    vl_rec_server_url="http://127.0.0.1:8118/v1",
)
output = pipeline.predict("path/to/image.png")
```

Note: The full PaddleOCR-VL client should run in a **separate** environment if it needs PaddlePaddle GPU + Transformers. This repo is the isolated vLLM server only.

## Tuning vLLM

Edit `config/vllm_config.yaml`:

```yaml
gpu-memory-utilization: 0.8
max-num-seqs: 128
```

Reference: [PaddleOCR-VL vLLM parameter tuning](https://www.paddleocr.ai/latest/en/version3.x/pipeline_usage/PaddleOCR-VL.html#331-server-side-parameter-adjustment)

## Troubleshooting

See `issues/` for full write-ups. Quick pointers:

| Symptom | Issue file |
|---------|------------|
| `paddleocr install_genai_server_deps` / `No module named pip` | [01-genai-server-deps-pip-in-uv-venv.md](../issues/01-genai-server-deps-pip-in-uv-venv.md) |
| flash-attn wheel incompatible with Python version | [02-flash-attn-wheel-python-version-mismatch.md](../issues/02-flash-attn-wheel-python-version-mismatch.md) |
| `uv pip` targets wrong venv from another project | [03-active-virtual-env-from-other-project.md](../issues/03-active-virtual-env-from-other-project.md) |
| Free memory below `gpu-memory-utilization` on startup | [04-gpu-memory-startup-failure.md](../issues/04-gpu-memory-startup-failure.md) |
| `TokenizersBackend has no attribute all_special_tokens_extended` | [05-transformers-tokenizers-incompatibility.md](../issues/05-transformers-tokenizers-incompatibility.md) |
| Extracted images not shown in Gradio demo (raw base64 in markdown) | [06-extracted-images-raw-base64-not-displayed.md](../issues/06-extracted-images-raw-base64-not-displayed.md) |

### flash-attn build failures

Install the prebuilt wheel after `uv sync` (see `scripts/install.sh`):

```bash
FLASH_ATTN_WHEEL="https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.3.14/flash_attn-2.8.2+cu128torch2.8-cp312-cp312-linux_x86_64.whl" \
  ./scripts/install.sh
```

Pick the wheel matching your Python and CUDA versions from [flash-attention prebuild wheels](https://mjunya.com/flash-attention-prebuild-wheels/). Details: [02-flash-attn-wheel-python-version-mismatch.md](../issues/02-flash-attn-wheel-python-version-mismatch.md).

Note: `paddleocr install_genai_server_deps` uses `pip` internally and is incompatible with uv-managed venvs. See [01-genai-server-deps-pip-in-uv-venv.md](../issues/01-genai-server-deps-pip-in-uv-venv.md). This repo installs the vLLM stack via `uv sync` + `uv pip`.

### `TokenizersBackend has no attribute all_special_tokens_extended`

Pin transformers (already in `pyproject.toml`):

```bash
uv pip install "transformers==4.57.6"
```

See [05-transformers-tokenizers-incompatibility.md](../issues/05-transformers-tokenizers-incompatibility.md).

### Do not install `paddlepaddle-gpu` in this venv

vLLM and PaddlePaddle GPU conflict. This server env uses `paddleocr[doc-parser]` without Paddle GPU.

### GPU memory on startup

If vLLM reports free memory below `gpu-memory-utilization`, either:

- Set `CUDA_VISIBLE_DEVICES` to a less-busy GPU
- Lower `gpu-memory-utilization` in `config/vllm_config.yaml` (e.g. `0.75` or `0.7`)

See [04-gpu-memory-startup-failure.md](../issues/04-gpu-memory-startup-failure.md).

### Health check

```bash
curl -s http://localhost:8118/v1/models | jq .
```

## References

- [PaddleOCR-VL usage tutorial](https://www.paddleocr.ai/latest/en/version3.x/pipeline_usage/PaddleOCR-VL.html)
- [PaddleOCR genai_server FAQ](https://github.com/PaddlePaddle/PaddleOCR/discussions/16822)