minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 11 Packages Projects Releases Wiki Activity

[Diffusion] migrate the whole _register_configs from registry.py to the model own config file (#40475)

Browse Source
This commit is contained in:
ronnie_zheng
2026-09-21 20:45:53 +03:00
committed by GitHub
parent 7a6191c4b9
commit e6931ca889
32 changed files with 936 additions and 857 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/docs/sglang-diffusion/support_new_models.mdx
+25 -20
View File
@@ -12,12 +12,11 @@ though the runtime is split into separate folders.
The files are split by runtime responsibility. For a new model, read the
request path first:
1. `registry.py` chooses the model family, sampling params, and pipeline config.
2. `configs/pipeline_configs/{model}.py` defines model-specific denoising and
1. `configs/pipeline_configs/{model}.py` defines model-specific denoising and
decoding behavior.
3. `runtime/pipelines/{model}.py` wires modules into stages.
4. `runtime/pipelines_core/stages/` runs the shared stage logic.
5. `runtime/models/` contains native model components only when the architecture
2. `runtime/pipelines/{model}.py` wires modules into stages.
3. `runtime/pipelines_core/stages/` runs the shared stage logic.
4. `runtime/models/` contains native model components only when the architecture
cannot be reused.
That is the dependency direction. Avoid making a model PR that requires readers
@@ -124,7 +123,6 @@ behavior.
| Area | Add or edit when | Typical file |
| --- | --- | --- |
| Registry | Always, unless extending an already registered family | `python/sglang/multimodal_gen/registry.py` |
| Runtime parameters | The request schema differs from existing models | `configs/sample/{model}.py` |
| Pipeline config | Denoising, decoding, precision, position encoding, or CFG hooks differ | `configs/pipeline_configs/{model}.py` |
| Pipeline wiring | The model needs a new stage layout or module list | `runtime/pipelines/{model}.py` |
@@ -137,11 +135,10 @@ behavior.
For a new native architecture, the common minimum is:
1. `registry.py`
2. `configs/sample/{model}.py`
3. `configs/pipeline_configs/{model}.py`
4. `runtime/pipelines/{model}.py`
5. `runtime/models/dits/{model}.py`
1. `configs/sample/{model}.py`
2. `configs/pipeline_configs/{model}.py`
3. `runtime/pipelines/{model}.py`
4. `runtime/models/dits/{model}.py`
Every extra file should map to model behavior that existing code cannot express
clearly.
@@ -499,19 +496,27 @@ native integration contract.
### 6. Registry
Register the family once the sampling params and pipeline config exist.
Define a `register()` function in `configs/pipeline_configs/{model}.py`. The
runtime auto-discovers it on startup and calls it to register the sampling
params and pipeline config.
```python
register_configs(
model_family="my_model",
sampling_param_cls=MyModelSamplingParams,
pipeline_config_cls=MyModelPipelineConfig,
hf_model_paths=["org/my-model"],
)
# python/sglang/multimodal_gen/configs/pipeline_configs/my_model.py
def register():
from sglang.multimodal_gen.registry import register_configs
register_configs(
sampling_param_cls=MyModelSamplingParams,
pipeline_config_cls=MyModelPipelineConfig,
hf_model_paths=["org/my-model"],
model_detectors=[lambda hf_id: "my-model" in hf_id.lower()],
)
```
The pipeline file is discovered through its `EntryClass`; do not add a second
pipeline registry unless the existing registry requires it.
`model_detectors` matches a model path or `model_index.json` `_class_name` when
the Hugging Face path varies; see `wan.py` or `qwen_image21.py` for real
examples. The pipeline file is discovered through its `EntryClass`; do not add
a second pipeline registry unless the existing registry requires it.
## Verify the Port