sheaf-serve 0.5.1

Unified serving layer for non-text foundation models

Sheaf

Unified serving layer for non-text foundation models.

vLLM solved inference for text LLMs by defining a standard compute contract and optimizing behind it. The same problem exists for every other class of foundation model — time series, tabular, molecular, geospatial, diffusion, audio — and nobody has solved it. Sheaf is that solution.

Each model type gets a typed request/response contract. Batching, caching, and scheduling are optimized per model type. Ray Serve is the substrate. Feast is a first-class input primitive.

In mathematics, a sheaf tracks locally-defined data that glues consistently across a space. Each model type defines its own local contract; Sheaf ensures they cohere into a unified serving layer.

---

Install

pip install sheaf-serve                           # core only
pip install "sheaf-serve[time-series]"            # + Chronos2 / TimesFM / Moirai
pip install "sheaf-serve[tabular]"                # + TabPFN
pip install "sheaf-serve[molecular]"              # + ESM-3  (Python 3.12+)
pip install "sheaf-serve[genomics]"               # + Nucleotide Transformer
pip install "sheaf-serve[small-molecule]"         # + MolFormer
pip install "sheaf-serve[materials]"              # + MACE-MP
pip install "sheaf-serve[audio]"                  # + Whisper / faster-whisper
pip install "sheaf-serve[audio-generation]"       # + MusicGen
pip install "sheaf-serve[tts]"                    # + Bark
pip install "sheaf-serve[vision]"                 # + DINOv2 / OpenCLIP / SAM2 / Depth Anything / DETR
pip install "sheaf-serve[earth-observation]"      # + Prithvi
pip install "sheaf-serve[weather]"                # + GraphCast
pip install "sheaf-serve[feast]"                  # + Feast feature store integration
pip install "sheaf-serve[modal]"                  # + Modal serverless deployment
pip install "sheaf-serve[all]"                    # everything

Quickstart

Direct backend inference:

from sheaf.api.time_series import Frequency, OutputMode, TimeSeriesRequest
from sheaf.backends.chronos import Chronos2Backend

backend = Chronos2Backend(model_id="amazon/chronos-bolt-tiny", device_map="cpu")
backend.load()

req = TimeSeriesRequest(
    model_name="chronos-bolt-tiny",
    history=[312, 298, 275, 260, 255, 263, 285, 320,
             368, 402, 421, 435, 442, 438, 430, 425],
    horizon=12,
    frequency=Frequency.HOURLY,
    output_mode=OutputMode.QUANTILES,
    quantile_levels=[0.1, 0.5, 0.9],
)

response = backend.predict(req)
# response.mean, response.quantiles

Ray Serve (production, autoscaling):

from sheaf import ModelServer
from sheaf.spec import ModelSpec, ResourceConfig
from sheaf.api.base import ModelType

server = ModelServer(models=[
    ModelSpec(
        name="chronos",
        model_type=ModelType.TIME_SERIES,
        backend="chronos2",
        backend_kwargs={"model_id": "amazon/chronos-bolt-small"},
        resources=ResourceConfig(num_gpus=1),
    ),
])
server.run()  # POST /chronos/predict, GET /chronos/health

Feast feature store (resolve features at request time):

# ModelSpec wires Feast — no history needed in the request
spec = ModelSpec(
    name="chronos",
    model_type=ModelType.TIME_SERIES,
    backend="chronos2",
    feast_repo_path="/feast/feature_repo",
)

# Client sends feature_ref instead of raw history
{
    "model_type": "time_series",
    "model_name": "chronos",
    "feature_ref": {
        "feature_view": "asset_prices",
        "feature_name": "close_history_30d",
        "entity_key": "ticker",
        "entity_value": "AAPL"
    },
    "horizon": 7,
    "frequency": "1d"
}

Modal (serverless, zero-infra):

from sheaf import ModalServer

server = ModalServer(models=[spec], app_name="my-sheaf", gpu="A10G")
app = server.app  # modal deploy my_server.py

See examples/ for time series comparison, tabular, audio, vision, and the Feast feature store quickstart.

---

Supported model types

Type	Status	Backends
Time series	✅ v0.1	Chronos2, Chronos-Bolt, TimesFM, Moirai
Tabular	✅ v0.1	TabPFN v2
Audio transcription	✅ v0.3	Whisper, faster-whisper
Audio generation	✅ v0.3	MusicGen
Text-to-speech	✅ v0.3	Bark
Vision embeddings	✅ v0.3	OpenCLIP, DINOv2
Segmentation	✅ v0.3	SAM2
Depth estimation	✅ v0.3	Depth Anything v2
Object detection	✅ v0.3	DETR / RT-DETR
Protein / molecular	✅ v0.3	ESM-3 (Python 3.12+)
Genomics	✅ v0.3	Nucleotide Transformer
Small molecule	✅ v0.3	MolFormer-XL
Materials science	✅ v0.3	MACE-MP-0
Earth observation	✅ v0.3	Prithvi (IBM/NASA)
Weather forecasting	✅ v0.3	GraphCast
Cross-modal embeddings	✅ v0.3	ImageBind (text, vision, audio, depth, thermal)
Feast feature store	✅ v0.3	Any Feast online store (SQLite, Redis, DynamoDB, …)
Modal serverless	✅ v0.3	ModalServer — zero-infra GPU deployment
Diffusion / image gen	🔜 v0.4	FLUX
Neural operators	🔜 v0.4	FNO, DeepONet
Video understanding	🔜 v0.4	VideoMAE, TimeSformer

Roadmap to production

v0.2 — serving layer (complete)

• Ray Serve integration tested end-to-end
• Async predict() handlers
• HTTP API with proper request validation (422 on bad input)
• Health check and readiness probe endpoints
• Batching scheduler (BatchPolicy wired into @serve.batch per deployment)
• Error handling at the service boundary (backend exceptions → structured HTTP 500)
• Model hot-swap without restart (ModelServer.update())
• Container-friendly auth for TabPFN v2 (TABPFN_TOKEN env var)

v0.3 — model types + integrations (complete)

• ESM-3 protein embeddings
• Nucleotide Transformer genomics embeddings
• MolFormer-XL small molecule embeddings
• MACE-MP-0 materials (energy, forces, stress)
• Whisper / faster-whisper audio transcription
• MusicGen audio generation
• Bark text-to-speech
• OpenCLIP image/text embeddings
• DINOv2 image embeddings
• SAM2 segmentation
• Depth Anything v2 depth estimation
• DETR / RT-DETR object detection
• Prithvi earth observation embeddings
• GraphCast weather forecasting
• ImageBind cross-modal embeddings (text, vision, audio, depth, thermal)
• Feast feature store integration (feature_ref in requests, FeastResolver, feast_repo_path on ModelSpec)
• Modal serverless deployment (ModalServer — zero-infra alternative to Ray Serve)

v0.4 — generation + video (complete)

• FLUX diffusion / image generation
• VideoMAE / TimeSformer video understanding

v0.5 — observability + new modalities

Ops / DX:

• PyPI publish (v0.4.0)
• Prometheus metrics endpoint per deployment
• Structured logging with request IDs end-to-end
• OpenTelemetry traces through the request path

Serving / infra:

• Streaming responses (POST /{name}/stream → SSE; FLUX emits per-step progress events)
• Request caching (CacheConfig on ModelSpec — in-process LRU, optional TTL)
• bucket_by batching — group requests by field value before @serve.batch

New model types:

• LiDAR / 3D point cloud (PointNet — pure-PyTorch, no torch-geometric; embed + ModelNet40 classify; install with pip install 'sheaf-serve[lidar]')
• Pose estimation (ViTPose — COCO 17-keypoint skeleton, optional person bboxes; install with pip install 'sheaf-serve[pose]')
• Optical flow (RAFT — raft_large/raft_small via torchvision; (H, W, 2) float32 flow field; install with pip install 'sheaf-serve[optical-flow]')
• Multimodal generation — text+image-conditioned (SDXL img2img + inpainting; install with pip install 'sheaf-serve[multimodal-generation]')
• Speech synthesis with fine-grained control (Kokoro — voice + speed per request; install with pip install 'sheaf-serve[kokoro]')

v0.6 — batch inference + async jobs

The goal: cover every shape of production inference, not just synchronous HTTP.

Offline / batch:

• BatchRunner — same backend, same typed contract, offline batch mode; Ray Data substrate for distributed execution, checkpointing, and output sinks (S3, Delta Lake, BigQuery)
• BatchSpec — mirrors ModelSpec with source/sink config; reuses all existing backends without modification

Async job queue:

• SheafWorker — queue-consumer pattern for long-running inference (Redis Streams, SQS, Kafka); decouples clients from compute for jobs where HTTP request/response is the wrong shape (FLUX 50-step, GraphCast multi-day rollouts)
• Job lifecycle: enqueue → processing → result / dead-letter; webhook on completion
• Priority lanes + per-tenant fair queuing

v0.7 — adapter multiplexing + client SDK

Adapter multiplexing:

• LoRA / adapter hot-swap per request — one GPU deployment serves many fine-tunes; adapters dict on ModelSpec, adapter_id field in requests
• Adapter registry: load on demand, LRU eviction when VRAM is tight
• First targets: FLUX (style LoRAs), Whisper (language adapters), ESM-3 (task heads)

Client SDK:

• pip install sheaf-client — typed Python client generated from request/response schemas
• Async client (httpx-backed); retry + timeout; streams SSE natively
• Language-agnostic: publish OpenAPI spec so teams can generate clients in any language

---

Architecture

┌─────────────────────────────────────────┐
│           API Layer                      │  typed contracts per model type
│  TimeSeriesRequest  TabularRequest  ...  │
├─────────────────────────────────────────┤
│         Scheduling Layer                 │  model-type-aware batching
│  BatchPolicy  RequestQueue               │
├─────────────────────────────────────────┤
│          Backend Layer                   │  pluggable execution + Ray Serve
│  ModelBackend  CacheManager  Feast       │
└─────────────────────────────────────────┘

Adding a new backend takes one class:

from sheaf.backends.base import ModelBackend
from sheaf.registry import register_backend

@register_backend("my-model")
class MyModelBackend(ModelBackend):
    def load(self) -> None:
        self._model = load_my_model()

    def predict(self, request):
        ...

    @property
    def model_type(self):
        return "time_series"

---

Contributing

Issues and PRs welcome. See CONTRIBUTING.md for development setup.

License

Apache 2.0