Rare Earth Recovery Predictor
Predicts element recovery across a solvent-extraction separation train from stage conditions and feed assay. Reference implementation trained on synthetic separation data.
This model is a reference implementation. Industrial use requires validation on plant-specific data. Customer-specific fine-tuning remains private unless explicitly approved for publication.
Input schema
| Field | Type | Description |
|---|---|---|
| stage_ph | float[] | pH by extraction / scrub / strip stage |
| org_aq_ratio | float | Organic-to-aqueous phase ratio |
| feed_assay | float[] · wt% | Feed composition by element |
| feed_flow | float · m³/h | Feed volumetric flow |
Output schema
| Field | Type | Description |
|---|---|---|
| recovery | float · % | Predicted element recovery |
| recovery_ci | float[] | Prediction interval |
Training data & evaluation
Training data type
Synthetic separation dataset
Evaluation metrics
Example usage
from matereal import load_model
m = load_model('matereal/ree-recovery-predictor')
m.predict({'stage_ph': [...], 'org_aq_ratio': 1.2, ...})Limitations
- Trained on synthetic data with idealized stage behaviour.
- Does not model reagent degradation over long campaigns.
Discussion
- s.Validation note
s.iyer
ML researcher · 3d
Reproduced the reported metric on the public split with a fixed seed. Numbers line up within noise, so it's a good baseline to build on.
18 - l.Improvement
l.zhang
Process engineer · 1w
Adding an uncertainty head would make this far more useful for prioritization. Happy to open a PR against the reference repo.
11
Validate on your data, keep the result private
Reference models are a starting point. We can fine-tune privately on plant-specific data, with customer-owned IP and no public disclosure.