Rewrite model card: clearer docs, dataset links, no checkpoint iters

README.md

@@ -7,38 +7,35 @@ tags:
   - modulated-convolution
   - fastmri
   - conditional
+pipeline_tag: image-to-image
 ---
 
 # DIRECT — Modulated Convolution for Conditional MRI Reconstruction
 
-Pretrained checkpoints and matching YAML configs for
-[Conditional Learned Reconstruction for Medical Imaging](https://proceedings.mlr.press/v315/moriakov26a.html)
-(Moriakov et al., MIDL 2026).
+Pretrained vSHARP models with **modulated convolutions** (and baselines) that condition reconstruction on acceleration \(R\) and ACS fraction at inference time.
 
-## Dataset
+- Paper (MIDL 2026, PMLR 315): [Conditional Learned Reconstruction for Medical Imaging](https://proceedings.mlr.press/v315/moriakov26a.html)
+- OpenReview: [PDF](https://openreview.net/pdf?id=qNjleGZJis)
+- Project configs: [`projects/modulated_convolution`](https://github.com/NKI-AI/direct/tree/main/projects/modulated_convolution)
 
-| Anatomy | Dataset | Notes |
-|---|---|---|
-| **Knee** | [fastMRI](https://fastmri.med.nyu.edu/) knee | Multi-coil; conditioned on acceleration \(R\) and ACS fraction |
-| **Prostate** | [fastMRI](https://fastmri.med.nyu.edu/) prostate | Same conditioning |
+## Datasets
 
-## Paper
+| Anatomy | Dataset | Link |
+|---------|---------|------|
+| Knee | fastMRI knee (multi-coil) | [fastmri.med.nyu.edu](https://fastmri.med.nyu.edu/) |
+| Prostate | fastMRI prostate (multi-coil) | [fastmri.med.nyu.edu](https://fastmri.med.nyu.edu/) |
 
-- **MIDL 2026 (PMLR 315:754–780):** [proceedings](https://proceedings.mlr.press/v315/moriakov26a.html) · [OpenReview PDF](https://openreview.net/pdf?id=qNjleGZJis)
-
-## Files
+## Models
 
 ```text
-knee/<name>.yaml
-knee/<name>.pt
-prostate/<name>.yaml
-prostate/<name>.pt
+knee/<name>.yaml + knee/<name>.pt
+prostate/<name>.yaml + prostate/<name>.pt
 ```
 
 ### Knee
 
 | Name | Description |
-|---|---|
+|------|-------------|
 | `vsharp_triang` | Non-modulated vSHARP baseline (triangular \(R\) sampling) |
 | `vsharp_modconv_features_triang` | ModConv features, triangular |
 | `vsharp_modconv_features_triang_32_16` | ModConv features, MLP 32→16 |
@@ -49,41 +46,47 @@ prostate/<name>.pt
 ### Prostate
 
 | Name | Description |
-|---|---|
+|------|-------------|
 | `vsharp_triang` | Non-modulated baseline |
 | `vsharp_modconv_features_triang` | ModConv features |
 | `vsharp_modconv_features_triang_32_16` | ModConv features, MLP 32→16 |
 | `vsharp_modconv_features_triang_32_8` | ModConv features, MLP 32→8 |
 
-## Training protocol & inference masking
+## Training protocol
 
-Trained on **fastMRI** knee / prostate multi-coil data. Training samples
-acceleration in a triangular / range schedule (typically `R ∈ [4, 16]`) with
-paired ACS `center_fractions` (e.g. `[0.08, 0.02]`) under
-`FastMRIEquispaced`.
+Training samples acceleration in a triangular / range schedule (typically \(R \in [4, 16]\)) with paired ACS `center_fractions` (e.g. `[0.08, 0.02]`) under `FastMRIEquispaced`. Models are conditioned on the realized \(R\) / ACS so a single checkpoint covers the training range.
 
-Released YAMLs pin **exactly one** R and **one** ACS (default validation 4×:
-`accelerations: [4]`, `center_fractions: [0.08]`). `MaskFunc` randomly draws
-from lists, so multi-R lists are for training only.
+Released inference YAMLs pin **exactly one** \(R\) and **one** ACS (default validation 4×: `accelerations: [4]`, `center_fractions: [0.08]`). Keep lists of length 1 when changing rate:
 
-Examples for other rates (keep lists length 1): 8× → `[8]` / `[0.04]`;
-16× → `[16]` / `[0.02]`.
+| Target \(R\) | `accelerations` | `center_fractions` |
+|-------------|-----------------|--------------------|
+| 4× | `[4]` | `[0.08]` |
+| 8× | `[8]` | `[0.04]` |
+| 16× | `[16]` | `[0.02]` |
 
 ## Usage
 
 ```bash
+pip install direct-recon   # or: https://github.com/NKI-AI/direct
 hf download NKI-AI/direct-modulated-convolution --local-dir ./modconv
 
 direct predict ./predictions \
   --cfg ./modconv/knee/vsharp_modconv_features_triang_32_8.yaml \
   --checkpoint ./modconv/knee/vsharp_modconv_features_triang_32_8.pt \
-  --data-root /path/to/fastmri/knee/val \
+  --data-root /path/to/fastmri/knee/multicoil_val \
   --num-gpus 1
 ```
 
-The first argument is the **output directory** (predictions), not an experiment folder.
-
+The first argument is the **prediction output directory**.
 
-## Framework
+## Citation
 
-[DIRECT](https://github.com/NKI-AI/direct) · project docs in `projects/modulated_convolution/` · [NKI-AI](https://huggingface.co/NKI-AI)
+```bibtex
+@inproceedings{moriakov2026modconv,
+  title={Conditional Learned Reconstruction for Medical Imaging},
+  author={Moriakov, Nikita and Yiasemis, George and Sonke, Jan-Jakob and Teuwen, Jonas},
+  booktitle={Medical Imaging with Deep Learning},
+  year={2026},
+  url={https://proceedings.mlr.press/v315/moriakov26a.html}
+}
+```