Deploy PR #23 preview

Author: github-actions[bot]

pr-23/_sources/api/algorithms.rst.txt

@@ -119,13 +119,13 @@ Logged metrics include:
 Checkpointing
 -------------
 
-Checkpoints are saved automatically:
+Checkpoints are saved automatically in safetensors format:
 
-- ``checkpoint_last.pt``: Most recent checkpoint (frequency controlled by ``logging.checkpoint_freq``)
-- ``checkpoint_best.pt``: Best performing checkpoint
+- ``checkpoint_last.safetensors``: Most recent checkpoint (frequency controlled by ``logging.checkpoint_freq``)
+- ``checkpoint_best.safetensors``: Best performing checkpoint
 
 Load a checkpoint:
 
 .. code-block:: bash
 
-   python -m twisterl.train --config config.json --load_checkpoint_path runs/my_run/checkpoint_best.pt
+   python -m twisterl.train --config config.json --load_checkpoint_path runs/my_run/checkpoint_best.safetensors

pr-23/_sources/api/environments.rst.txt

@@ -96,6 +96,9 @@ All environments must provide these methods (called from Rust or Python):
 - ``step(action)``: Take action, return (obs, reward, done, info)
 - ``obs_shape()``: Return observation dimensions
 - ``num_actions()``: Return number of valid actions
+- ``is_final()``: Return True if current state is terminal
+- ``success()``: Return True if the goal was achieved (episode ended successfully)
+- ``reward()``: Return the reward value for the current state
 - ``twists()``: Return permutation symmetries (optional)
 - ``set_state(state)``: Set environment to specific state (for inference)
 - ``difficulty``: Property to get/set difficulty level

pr-23/_sources/api/twisterl.rst.txt

@@ -44,6 +44,7 @@ Key functions:
 
 - ``prepare_algorithm(config, run_path, load_checkpoint_path)``: Prepares an algorithm instance from config
 - ``load_config(config_path)``: Loads a JSON config file
+- ``load_checkpoint(checkpoint_path)``: Loads a checkpoint from either safetensors or pt format
 - ``pull_hub_algorithm(repo_id, model_path, revision, validate)``: Downloads models from HuggingFace Hub
 
 twisterl.defaults module

pr-23/api/algorithms.html

@@ -192,13 +192,13 @@ <h2>Metrics and Logging<a class="headerlink" href="#metrics-and-logging" title="
 </section>
 <section id="checkpointing">
 <h2>Checkpointing<a class="headerlink" href="#checkpointing" title="Link to this heading"></a></h2>
-<p>Checkpoints are saved automatically:</p>
+<p>Checkpoints are saved automatically in safetensors format:</p>
 <ul class="simple">
-<li><p><code class="docutils literal notranslate"><span class="pre">checkpoint_last.pt</span></code>: Most recent checkpoint (frequency controlled by <code class="docutils literal notranslate"><span class="pre">logging.checkpoint_freq</span></code>)</p></li>
-<li><p><code class="docutils literal notranslate"><span class="pre">checkpoint_best.pt</span></code>: Best performing checkpoint</p></li>
+<li><p><code class="docutils literal notranslate"><span class="pre">checkpoint_last.safetensors</span></code>: Most recent checkpoint (frequency controlled by <code class="docutils literal notranslate"><span class="pre">logging.checkpoint_freq</span></code>)</p></li>
+<li><p><code class="docutils literal notranslate"><span class="pre">checkpoint_best.safetensors</span></code>: Best performing checkpoint</p></li>
 </ul>
 <p>Load a checkpoint:</p>
-<div class="highlight-bash notranslate"><div class="highlight"><pre><span></span>python<span class="w"> </span>-m<span class="w"> </span>twisterl.train<span class="w"> </span>--config<span class="w"> </span>config.json<span class="w"> </span>--load_checkpoint_path<span class="w"> </span>runs/my_run/checkpoint_best.pt
+<div class="highlight-bash notranslate"><div class="highlight"><pre><span></span>python<span class="w"> </span>-m<span class="w"> </span>twisterl.train<span class="w"> </span>--config<span class="w"> </span>config.json<span class="w"> </span>--load_checkpoint_path<span class="w"> </span>runs/my_run/checkpoint_best.safetensors
 </pre></div>
 </div>
 </section>

pr-23/api/environments.html

@@ -182,6 +182,9 @@ <h2>Environment Interface<a class="headerlink" href="#environment-interface" tit
 <li><p><code class="docutils literal notranslate"><span class="pre">step(action)</span></code>: Take action, return (obs, reward, done, info)</p></li>
 <li><p><code class="docutils literal notranslate"><span class="pre">obs_shape()</span></code>: Return observation dimensions</p></li>
 <li><p><code class="docutils literal notranslate"><span class="pre">num_actions()</span></code>: Return number of valid actions</p></li>
+<li><p><code class="docutils literal notranslate"><span class="pre">is_final()</span></code>: Return True if current state is terminal</p></li>
+<li><p><code class="docutils literal notranslate"><span class="pre">success()</span></code>: Return True if the goal was achieved (episode ended successfully)</p></li>
+<li><p><code class="docutils literal notranslate"><span class="pre">reward()</span></code>: Return the reward value for the current state</p></li>
 <li><p><code class="docutils literal notranslate"><span class="pre">twists()</span></code>: Return permutation symmetries (optional)</p></li>
 <li><p><code class="docutils literal notranslate"><span class="pre">set_state(state)</span></code>: Set environment to specific state (for inference)</p></li>
 <li><p><code class="docutils literal notranslate"><span class="pre">difficulty</span></code>: Property to get/set difficulty level</p></li>

pr-23/api/twisterl.html

@@ -133,6 +133,7 @@ <h2>twisterl.utils module<a class="headerlink" href="#twisterl-utils-module" tit
 <ul class="simple">
 <li><p><code class="docutils literal notranslate"><span class="pre">prepare_algorithm(config,</span> <span class="pre">run_path,</span> <span class="pre">load_checkpoint_path)</span></code>: Prepares an algorithm instance from config</p></li>
 <li><p><code class="docutils literal notranslate"><span class="pre">load_config(config_path)</span></code>: Loads a JSON config file</p></li>
+<li><p><code class="docutils literal notranslate"><span class="pre">load_checkpoint(checkpoint_path)</span></code>: Loads a checkpoint from either safetensors or pt format</p></li>
 <li><p><code class="docutils literal notranslate"><span class="pre">pull_hub_algorithm(repo_id,</span> <span class="pre">model_path,</span> <span class="pre">revision,</span> <span class="pre">validate)</span></code>: Downloads models from HuggingFace Hub</p></li>
 </ul>
 </section>