Updated readme.md

README.md

@@ -3,53 +3,56 @@ license: apache-2.0
 language: en
 pipeline_tag: video-classification
 library_name: pytorch
+datasets:
+  - DrGil/k600_test_ds
 tags:
   - video-classification
   - movinet
   - kinetics-600
   - efficient-video
-  - regularization
   - temporal-smoothing
-datasets:
-  - DrGil/k600_test_ds
+  - regularization
 ---
 
-# GRW-smoothing (MoViNet) — NeurIPS 2025 checkpoints
-
-This repository provides **PyTorch checkpoint weights** for **MoViNet** video classification models trained with **GRW-smoothing**, introduced in:
-
-- Paper page: https://huggingface.co/papers/2511.20928
-- arXiv: https://arxiv.org/abs/2511.20928
-- Code / reproduction guide: https://github.com/cmusatyalab/grw-smoothing
+# GRW Smoothing (MoViNet) — Model Weights
 
-## What is GRW-smoothing?
+This repository contains **MoViNet** checkpoint weights trained with **GRW-smoothing** from the paper:
 
-**GRW-smoothing** is a regularization technique for video recognition models that encourages **temporal smoothness** in intermediate feature embeddings. Consecutive frame embeddings are modeled as a **Gaussian Random Walk (GRW)**, and training penalizes high “acceleration” in embedding space while preserving the correct temporal ordering.
+## Paper
+- https://huggingface.co/papers/2511.20928
+- https://arxiv.org/abs/2511.20928
 
-The method targets improved accuracy–efficiency trade-offs, particularly for lightweight video architectures.
-
-## Provided checkpoints
-
-This repo contains the following checkpoints:
+## Code (reproduction instructions)
+For **step-by-step instructions** to reproduce results (environment setup, data prep, evaluation commands, exact settings), please refer to:
+- https://github.com/cmusatyalab/grw-smoothing
 
+## Files (checkpoints)
+This repo provides the following checkpoint files:
 - `a0s_grw.pt`
 - `a1s_grw.pt`
 - `a2s_grw.pt`
 - `a3b_grw.pt`
 
-## Intended use
-
-These checkpoints are intended for:
-- Video classification research and benchmarking
-- Reproducing and extending results from the GRW-smoothing paper
-- Studying the effect of temporal smoothness regularization on compact video backbones
+## Associated dataset
+A frozen evaluation dataset used for reproducible testing is provided here:
+- https://huggingface.co/datasets/DrGil/k600_test_ds
 
-Not intended for:
-- Safety-critical deployments without extensive validation
-- Use cases requiring strict robustness guarantees under domain shift
-
-## How to reproduce paper results
-
-For **specific, step-by-step instructions** (environment setup, data layout, evaluation commands, and exact settings) to reproduce the paper’s results, please follow the upstream GitHub repository:
-
-https://github.com/cmusatyalab/grw-smoothing
+## Intended use
+These weights are intended for research and benchmarking in **video classification**, especially when studying efficiency/accuracy trade-offs and temporal smoothness regularization.
+
+## Limitations
+- Performance and behavior depend on the training/evaluation setup described in the paper and code.
+- Models may not generalize well outside the evaluation distribution.
+- If used on real-world or sensitive video content, apply appropriate privacy and governance practices.
+
+## Citation
+If you use this model, please cite:
+
+```bibtex
+@inproceedings{goldman2025grwsmoothing,
+  title     = {Smooth Regularization for Efficient Video Recognition},
+  author    = {Gil Goldman and Raja Giryes and Mahadev Satyanarayanan},
+  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
+  year      = {2025},
+  url       = {https://arxiv.org/abs/2511.20928}
+}