[feat] Align multimodal OpenAI serving APIs (#1781)

2026-08-28 10:09:58 -07:00
parent c7da2f5d60
commit 3987b9ddcd
22 changed files with 1827 additions and 439 deletions
@@ -0,0 +1,43 @@
+# OpenAI-compatible serving examples
+
+The REST serving engine is model-agnostic. Any model supported by
+`VideoGenerator` can use the same `/v1/models`, `/v1/videos`, and `/v1/images`
+surface; the two configs here are FastH3 validation profiles.
+
+Launch the full FastH3 checkpoint:
+
+```bash
+fastvideo serve --config examples/serving/openai_fasth3.yaml
+```
+
+Launch the dense FastH3 LoRA on the base MiniMax-H3 checkpoint:
+
+```bash
+adapter_path="$(hf download \
+  FastVideo/FastVideo-FastH3-4-step-Preview-v1-LoRA \
+  dense-datafree/adapter_model.safetensors)"
+
+fastvideo serve --config examples/serving/openai_fasth3_lora.yaml \
+  --generator.pipeline.components.lora_path "$adapter_path"
+```
+
+FastH3 adapters are hybrid startup patches: alongside low-rank factors they
+may contain dense deltas and a VSA compression-gate replacement. They must be
+selected when the server starts. A request may carry the vLLM-Omni `lora`
+selector, but its name, path, and scale must match that startup adapter. A VSA
+adapter also needs `attention_backend: VIDEO_SPARSE_ATTN_H3`, `VSA_sparsity`,
+and `VSA_tile_size` like the full-checkpoint config.
+
+Submit and poll an asynchronous job:
+
+```bash
+job_id="$(curl -sS http://localhost:8000/v1/videos \
+  -H 'content-type: application/json' \
+  -d '{"model":"fasth3","prompt":"A fox runs through fresh snow."}' \
+  | jq -r .id)"
+
+curl -sS "http://localhost:8000/v1/videos/$job_id"
+curl -o result.mp4 "http://localhost:8000/v1/videos/$job_id/content"
+```
+
+For a blocking call, `POST /v1/videos/sync` returns the MP4 body directly.
@@ -0,0 +1,49 @@
+# OpenAI-compatible FastH3 full-checkpoint server.
+generator:
+  model_path: FastVideo/FastVideo-Minimax-FastH3-Preview-v0.2
+  engine:
+    num_gpus: 4
+    use_fsdp_inference: false
+    parallelism:
+      tp_size: 1
+      sp_size: 4
+    offload:
+      dit: false
+      dit_layerwise: false
+      text_encoder: true
+      image_encoder: true
+      vae: true
+      pin_cpu_memory: true
+    compile:
+      enabled: false
+      vae_enabled: false
+  pipeline:
+    workload_type: t2v
+    experimental:
+      attention_backend: VIDEO_SPARSE_ATTN_H3
+      VSA_sparsity: 0.9
+      VSA_tile_size: 64
+      inference_torch_compile: false
+      vae_parallel_decode: true
+      vae_parallel_decode_strategy: gather
+
+server:
+  host: 0.0.0.0
+  port: 8000
+  output_dir: outputs/openai_fasth3
+  served_model_name: fasth3
+
+default_request:
+  negative_prompt: ""
+  sampling:
+    height: 768
+    width: 1344
+    num_frames: 124
+    fps: 24
+    num_inference_steps: 5
+    guidance_scale: 1.0
+    batch_cfg: false
+    seed: 1000
+  output:
+    return_frames: false
+
@@ -0,0 +1,52 @@
+# OpenAI-compatible FastH3 startup-LoRA server. Override lora_path at launch
+# with the exact file returned by `hf download`; see README.md.
+generator:
+  model_path: MiniMaxAI/MiniMax-H3
+  engine:
+    num_gpus: 4
+    use_fsdp_inference: false
+    parallelism:
+      tp_size: 1
+      sp_size: 4
+    offload:
+      dit: false
+      dit_layerwise: false
+      text_encoder: true
+      image_encoder: true
+      vae: true
+      pin_cpu_memory: true
+    compile:
+      enabled: false
+      vae_enabled: false
+  pipeline:
+    workload_type: t2v
+    components:
+      lora_path: /path/to/adapter_model.safetensors
+      lora_nickname: fasth3-dense-datafree
+      lora_strength: 1.0
+    experimental:
+      attention_backend: FLASH_ATTN
+      inference_torch_compile: false
+      vae_parallel_decode: true
+      vae_parallel_decode_strategy: gather
+
+server:
+  host: 0.0.0.0
+  port: 8000
+  output_dir: outputs/openai_fasth3_lora
+  served_model_name: minimax-h3
+
+default_request:
+  negative_prompt: ""
+  sampling:
+    height: 768
+    width: 1344
+    num_frames: 124
+    fps: 24
+    num_inference_steps: 5
+    guidance_scale: 1.0
+    batch_cfg: false
+    seed: 1000
+  output:
+    return_frames: false
+