HyperView docs

HyperView · Guides

Spaces

Export a workspace as a bundle and share it as a Static Space that runs in any browser, or a Live Space backed by a HyperView server.

hyperview export packages a workspace — samples, media, layouts, panels and the view you arranged — into one folder, the bundle. The bundle is what you share. It can be hosted two ways:

Static SpaceLive Space
RunsPlain files on any static hosthyperview serve --from <bundle> in a container
Visitors canBrowse samples, layouts, selections, lasso, filters, precomputed neighbours and collections, static-compatible panelsEverything a Static Space does, plus typed text queries, new embeddings and layouts, Python tools
CostFree static hosting, no cold startA running server

Choose a Static Space unless visitors need to bring their own input and compute a new result. Every Space in the HyperView Spaces gallery is a Static Space.

Export

hyperview export my-workspace --out dist/my-space

Add --similarity-k 25 to precompute 25 nearest neighbours per sample, so the neighbour view works without a server. It is off by default to keep bundles small.

The bundle holds a hyperview-static.json manifest, the frontend, data shards, media and thumbnails, layouts, collections and panel modules. Its paths are relative, so the same folder works at a domain root or under any sub-path of another site. If the exporter reports warnings about missing media or panel sources, fix them before publishing.

For a reproducible export, build the workspace with a fresh HYPERVIEW_HOME.

Publish

Always start with a dry run; it prints the plan and touches nothing:

hyperview publish dist/my-space --to hf:your-org/my-space --dry-run
TargetCommand
Hugging Face Static Spacehyperview publish dist/my-space --to hf:your-org/my-space
Hugging Face Live Spacehyperview publish dist/my-space --to hf:your-org/my-space-live --mode live
Cloudflare Workershyperview publish dist/my-space --to cloudflare
A folder in your own sitehyperview publish dist/my-space --to dir:site/spaces/my-space

Hugging Face publishing reads HF_TOKEN or a prior hf auth login. For a Live Space that needs extra packages (pinned with ==) or more than the free CPU:

hyperview publish dist/my-space --to hf:your-org/my-space-live --mode live \
  --extra-pip "hyper-models[ml]==0.4.0" \
  --hardware cpu-upgrade

The same calls are available from Python:

hv.publish("dist/my-space", to="hf:your-org/my-space", mode="static")

Run a bundle locally as a Live Space

hyperview serve --from dist/my-space

HyperView restores the dataset, embedding spaces, layouts, collections and extensions recorded in the bundle and opens exactly the exported view. --public serves it without a session token: visitors keep the viewer commands, while extension installs, tools and compute stay closed.

Panels in a Static Space

Panels work in a Static Space when they only read data through the panel SDK. A panel marked static_compatible = false stays visible with its reason, and its code is not published. See Extensions.

Start from an example

The source of every example lives in Hyper3Labs/hyperview-spaces. Copy a folder from demos/, change the constants at the top of its demo.py to point at your dataset and models, and export.

Edit this page on GitHub