rlmesh.Session

The workflow edition this session runs at, in the spelling the peers agreed on.

On a served model it is the floor across the env, the model, and this runtime, settled when the session was opened; on a local model the env leg is the whole negotiation, so it is what that connection settled on. Named apart from the workflow_edition declaration surfaces: this is what was agreed, not what this side asked for.

Whether the current episode has terminated or truncated.

Begin a new episode: end the previous one, then reset the env and adapter.

Ending the previous episode fires the model’s on_episode_end (the local per-episode boundary), so a stateful model clears its state between episodes on the hand-driven path too, not only via run().

trial_index is the 0-based ordinal of this episode in a benchmark’s trial sweep, delivered as reset(options={"trial_index": ...}) – but only to an env that declared the key in EnvFactory.reset_options. Passing one to an env that did not warns and resets without it.

Map one env observation to an env-ready action (the model’s adapter applied).

Apply one action to the env; record reward and termination.

The connected env’s contract (connects on first use).

Lanes the connected env runs (a session drives one).

The env’s image roles and render() source, as the viewer sees them.

Build a read-only, role-addressed view over this env’s observations.

Each item is a role constant – kept in the env’s native encoding – or a model-input leaf declaring the encoding you want (Image(IMAGE_PRIMARY, layout="hwc"), State(EEF_POS)). The returned Reader maps a raw observation to {role: value} through the same adapter pipeline a model uses, so it is encoding-agnostic across envs and runs identically in the native core. Resolved once here, reused each step::

read = sess.reader(Image(IMAGE_PRIMARY, layout=“hwc”), EEF_POS) obs, _ = sess.reset() while not sess.done: screen.show(read(obs)[IMAGE_PRIMARY]) obs, *_ = sess.step(sess.predict(obs))

A bare role is desugared to the env-native leaf for that role (by the env’s own tag); pass an explicit leaf to override the encoding.

One-shot read of a single role from one observation.

The single-value convenience for reader() – item is a role constant or a model-input leaf. The reader is resolved once and cached per item, so calling this every step does not re-resolve::

ee = sess.read(obs, EEF_POS) img = sess.read(obs, Image(IMAGE_PRIMARY, layout=“hwc”))

The value is typed Any (like Reader’s): its concrete shape is the leaf’s declared encoding, which the caller owns.

The observation roles this session’s env declares, grouped by kind.

Connects if needed, reads the env contract’s published tags, and returns their observation_roles. An env that publishes no tags yields empty groups – “none declared” is an answer, not an error.

Drive whole episodes to completion and return a typed RunResult.

The single drive loop: pumps this session’s own reset / predict / step primitives, so a served model’s run routes through here. episodes is the exact number of episodes to run (see resolve_episode_budget(): one by default, the length of seeds when only seeds are given, and the two must agree when both are). max_episode_steps / max_episode_seconds cap each episode – hitting a cap marks it truncated, exactly like the built-in step bound (the wall-clock cap is checked at the top of the step loop). Episode i walks trial ordinal trial_index_base + i, delivered as reset(options={"trial_index": ...}) to an env that declared the key in :attr:EnvFactory.reset_options \<rlmesh.EnvFactory.reset_options> (an env that did not never sees it, and its EpisodeResult.trial stays None); the base lets a local eval walk the same states as a platform shard. hooks (RunHooks) observes the loop; hook exceptions propagate and abort the run, and RunHooks.on_run_end() always fires exactly once with the completed episodes, even on an error or interrupt.

Does not close the session: a caller-held session (from rlmesh.session() / Model.session()) stays connected – viewer and hooks included – so run can be called again or mixed with manual driving; close it via close() or the with block. (The one-shot rlmesh.run() / Model.run() own their internal session and close it for you.)

Close this session: served route (and owned source), connection, env.

For a served model, closes the model client and shuts down a managed source it started (e.g. a SandboxModel container). For the env, shuts it down only on the close_env opt-in, and always releases what this session made itself: a connection it dialed or an env it built from a factory. A local model instance is the caller’s and is never closed here (one the session built from a class is closed exactly once).

Idempotent: later calls are no-ops, and any other use of a closed session raises RuntimeError.