Model card for MORAD-RIDE

README.md

@@ -2,7 +2,7 @@
 license: apache-2.0
 library_name: pytorch
 datasets:
-  - Gabriel166/MORAD-data
+  - MORAD-RNA/MORAD-Targets
 tags:
   - rna
   - rna-design
@@ -11,62 +11,57 @@ tags:
   - reinforcement-learning
 ---
 
-# MORAD
+# MORAD-RIDE
 
-Checkpoint for **MORAD: Reinforcement Learning with Multi-Objective Rewards for
-RNA Inverse Design**: the pretrained RIDE backbone-conditioned RNA diffusion
-model after 390 updates of online reinforcement learning with the six-component
-MORAD reward. A single shared policy designs nucleotide sequences for any target
-3D backbone, with no per-target optimization.
+The released model of **MORAD: Reinforcement Learning with Multi-Objective
+Rewards for RNA Inverse Design**: the pretrained RIDE backbone-conditioned RNA
+diffusion model, post-trained online for 390 updates with the six-objective
+MORAD reward. One shared policy designs nucleotide sequences for any target 3D
+backbone, with no per-target optimization.
 
-Code: [github.com/Gabrile166/MORAD](https://github.com/Gabrile166/MORAD) ·
-Data: [Gabriel166/MORAD-data](https://huggingface.co/datasets/Gabriel166/MORAD-data)
-
-## File
-
-| File | Content |
-|---|---|
-| `policy_step000390.pt` | MORAD policy after 390 updates, `{"model": state_dict}` (10,214,213 parameters) |
-
-SHA-256 checksums are listed in `SHA256SUMS`.
+[Project page](https://gabrile166.github.io/MORAD/) ·
+[Code](https://github.com/Gabrile166/MORAD) ·
+[Data](https://huggingface.co/datasets/MORAD-RNA/MORAD-Targets)
 
 ## Results on the 153 test targets
 
-Structural and pairing metrics use candidate 0 of each target (structures
-predicted with RhoFold+, pairs with EternaFold); NED is from the thermodynamic
-evaluation (ViennaRNA).
+| Model | GDT-TS ↑ | TM-score ↑ | RMSD (Å) ↓ | Pairing MCC ↑ | NED ↓ |
+|---|:---:|:---:|:---:|:---:|:---:|
+| RiboDiffusion | 0.3132 | 0.2839 | 10.9215 | 0.3267 | 0.6056 |
+| gRNAde | 0.3591 | 0.3179 | 8.8374 | 0.6734 | 0.3458 |
+| RDesign | 0.3291 | 0.2902 | 10.3441 | 0.6566 | 0.3349 |
+| RIDE | 0.3300 | 0.2841 | 9.7199 | 0.6110 | 0.3906 |
+| **MORAD-RIDE** | **0.3827** | **0.3400** | **7.9196** | **0.7295** | **0.2830** |
 
-| Model | GDT-TS ↑ | TM (C4′) ↑ | RMSD (Å) ↓ | TM (C1′) ↑ | MCC ↑ | NED ↓ |
-|---|---:|---:|---:|---:|---:|---:|
-| RiboDiffusion | 0.3132 | 0.2839 | 10.9215 | 0.3613 | 0.3267 | 0.6056 |
-| gRNAde (T=0.1) | 0.3591 | 0.3179 | 8.8374 | 0.3642 | 0.6734 | 0.3458 |
-| RDesign | 0.3291 | 0.2902 | 10.3441 | 0.3392 | 0.6566 | 0.3349 |
-| RIDE | 0.3300 | 0.2841 | 9.7199 | 0.3538 | 0.6110 | 0.3906 |
-| **MORAD** | **0.3827** | **0.3400** | **7.9196** | **0.3855** | **0.7295** | **0.2830** |
+<sub>Structures predicted with RhoFold+ (TM-score on C4′), base pairs with
+EternaFold, normalized ensemble defect (NED) with ViennaRNA. gRNAde at T = 0.1.</sub>
 
-All tables of the paper are in the
-[repository README](https://github.com/Gabrile166/MORAD#results).
+## Files
 
-## Usage
+| File | Content |
+|---|---|
+| `morad_ride.pt` | policy weights, `{"model": state_dict}`, 10.2 M parameters |
+| `SHA256SUMS` | checksum |
+
+## Quick start
 
 ```bash
 git clone https://github.com/Gabrile166/MORAD.git && cd MORAD
-# install as described in docs/INSTALL.md, then
-bash scripts/download_assets.sh            # model weights, data and this checkpoint
-bash scripts/evaluate_test153.sh morad     # full evaluation on the 153 test targets
+# set up the environments as described in the README, then
+bash scripts/download_assets.sh            # RIDE + RhoFold+ weights, data, this model
+bash scripts/evaluate_test153.sh morad     # evaluate on the 153 test targets
 ```
 
-`download_assets.sh` stores this file as `artifacts/models/morad/morad_update390.pt`.
-Any MORAD checkpoint can be evaluated with
-`RIDE_CHECKPOINT=path/to/policy.pt python evaluate.py --config configs/eval/test153.yaml`.
-
 ```python
 import torch
 
-checkpoint = torch.load("policy_step000390.pt", map_location="cpu")
-state_dict = checkpoint["model"]  # RIDE network, defined in configs/ride_policy.yaml
+state_dict = torch.load("morad_ride.pt", map_location="cpu")["model"]
+# RIDE network definition: configs/ride_policy.yaml in the code repository
 ```
 
+Training this model from pretrained RIDE takes about 2 hours on 4 × RTX 3090
+(`bash scripts/train_morad.sh`).
+
 ## Citation
 
 ```bibtex