Upload folder using huggingface_hub

README.md

@@ -22,20 +22,26 @@ generation, following [openpi_subtask](https://github.com/Ke-Wang1017/openpi_sub
 
 - `config.json`: policy `type` is `pi05_subtask` instead of `pi05`.
 - `policy_preprocessor.json`: pi05's `pi05_prepare_state_tokenizer_processor_step` +
-  `tokenizer_processor` steps are replaced by `pi05_subtask_tokenize_processor_step`, which builds
-  the two-segment prompt
+  `tokenizer_processor` steps are replaced by `pi05_subtask_context_tokenize_processor_step`,
+  which tokenizes the context segment of the two-segment prompt
 
   ```
   context (no loss):   "Task: {task}.; State: {256-bin ints}; Subtask: "
   predicted (CE loss): "{subtask}.;\nAction: " + EOS
   ```
 
-  and emits a subtask region mask alongside the language tokens.
+  into `observation.language.context_tokens`/`.context_mask`. The predicted segment is model
+  territory: the policy appends it at training time (computing the CE region mask next to the
+  loss) and decodes it autoregressively at inference.
 
 This exists because LeRobot instantiates processor pipelines from the checkpoint's
 `policy_preprocessor.json` (by processor-step registry name), so fine-tuning the subtask policy
 directly from `lerobot/pi05_base` would silently install plain pi05's tokenizer steps.
 
+The same artifact is the base for the `framepick` policy type (`lerobot_policy_framepick`):
+framepick has no processor steps of its own — frame selection, chunk re-expression and the
+assignment prefix all live in the policy, configured from the policy config.
+
 ## Usage
 
 Requires the `lerobot_policy_pi05_subtask` plugin installed (LeRobot auto-imports

policy_preprocessor.json

@@ -34,7 +34,7 @@
       }
     },
     {
-      "registry_name": "pi05_subtask_tokenize_processor_step",
+      "registry_name": "pi05_subtask_context_tokenize_processor_step",
       "config": {
         "tokenizer_name": "google/paligemma-3b-pt-224",
         "max_length": 200,