Document exact runtime config paths

README.md

@@ -1,147 +0,0 @@
----
-library_name: memoryvla
-tags:
-- robotics
-- vision-language-action
-- memory
-- action-diffusion
-- piper
-pipeline_tag: robotics
-license: apache-2.0
----
-
-# MemoryVLA
-
-MemoryVLA checkpoint for real-robot action prediction. This repository is
-organized so that additional real-world task checkpoints can be added under
-`checkpoints/<task-name>/` without replacing the current model.
-
-## Latest checkpoint
-
-The current default checkpoint is:
-
-```text
-checkpoints/piper-color-sorting/frozen-ema-step-20000.pt
-```
-
-It is the EMA action-diffusion checkpoint from the Frozen+EMA training run.
-The matching action normalization statistics are stored at:
-
-```text
-configs/piper-color-sorting/dataset_statistics.json
-```
-
-The checkpoint is intended to be loaded with the MemoryVLA codebase and
-`use_ema=True`. The model uses the `custom_finetuning` normalization key.
-
-## Training configuration
-
-- Task: Piper color sorting
-- Data: 21 valid real-world episodes, 14,300 frames
-- Observation: front camera only in this dataset; no wrist-camera stream was
-  available in the training data
-- Base VLM: `prism-dinosiglip-224px+7b`
-- Vision backbone: frozen
-- LLM backbone: frozen
-- Last LLM layer: frozen
-- Trainable components: action diffusion model and MemoryVLA trainable
-  modules
-- Action dimension: 7 (`x, y, z, roll, pitch, yaw, gripper`)
-- Action model: `DiT-L`
-- Future action window: 15
-- Memory length: 256
-- Retrieval layers: 2
-- Fusion: `gate`
-- Consolidation: `tome`
-- Per-device batch size: 2
-- Global batch size: 32
-- Optimizer learning rate: `2e-5`
-- Scheduler: linear warmup + cosine decay
-- Warmup ratio: 0.03
-- EMA: enabled, decay `0.999`
-- Mixed precision: enabled
-- Training strategy: FSDP full shard
-- Maximum training steps: 20,000
-- Image augmentation: disabled
-- Random seed: 42
-
-## Open-loop evaluation
-
-On the same 21 training episodes, using frame-by-frame memory-aware inference:
-
-- Normalized overall action RMSE: `0.2124`
-- Gripper accuracy: `98.64%`
-- Memory reset: at the first frame of every episode
-- Cognitive and perception memory banks: capped at 256 entries
-
-These are training-set open-loop results and should not be interpreted as
-unseen-task generalization.
-
-## Loading outline
-
-The exact loader depends on the MemoryVLA code revision. The essential
-settings for this checkpoint are:
-
-```python
-model = load_vla(
-    model_id_or_path="checkpoints/piper-color-sorting/frozen-ema-step-20000.pt",
-    load_for_training=False,
-    action_dim=7,
-    future_action_window_size=15,
-    action_model_type="DiT-L",
-    mem_length=256,
-    retrieval_layers=2,
-    use_timestep_pe=True,
-    fusion_type="gate",
-    consolidate_type="tome",
-    update_fused=False,
-    use_ema=True,
-)
-```
-
-For real deployment, reset the episode memory before the first observation of
-each task and pass the matching `dataset_statistics.json` when unnormalizing
-actions. The gripper action is sign-encoded (`-1` / `+1`) by the current
-pipeline.
-
-## Repository layout
-
-```text
-checkpoints/
-  piper-color-sorting/
-    frozen-ema-step-20000.pt
-configs/
-  piper-color-sorting/
-    dataset_statistics.json
-```
-
-Future tasks should use a separate directory, for example:
-
-```text
-checkpoints/
-  piper-color-sorting/
-  drawer-opening/
-  peg-insertion/
-configs/
-  piper-color-sorting/
-  drawer-opening/
-  peg-insertion/
-```
-
-Keep the checkpoint, action statistics, task name, camera convention, and
-training configuration together for every task.
-
-## Code and reproducibility
-
-The Hugging Face repository stores model artifacts and deployment metadata.
-A separate GitHub repository is recommended for the MemoryVLA model code,
-real-robot wrapper, preprocessing, and evaluation scripts. It is not required
-to download the weights, but it makes future task training, deployment, and
-exact code-version tracking much safer. Record the Git commit or release tag
-used for each checkpoint in the corresponding task directory or Model Card.
-
-## Intended use
-
-Research and development for real-robot manipulation. Validate workspace
-limits, action scaling, emergency stop behavior, camera calibration, and
-gripper sign conventions before sending actions to hardware.