Cell Type Classifier

🧮 Model Details Deep Learning

Internalcandidate

ARCHITECTURE

pytorch · transformer · n_layers: 6 · hidden_dim: 512

METRICS

layer_config

{'n_layers': 6, 'hidden_dim': 512}

benchmark_name

Cell Type Classification Benchmark — SEA-AD MTG

benchmark_notes

Transformer architecture (6 layers, 512 hidden dim); final training loss 0.023

training_metrics

{'final_loss': 0.023}

EVALUATION CONTEXT

Dataset:Allen Brain SEA-AD MTG 10x snRNA-seq (held-out test split, ~20% donor stratified)

Benchmark:Cell Type Classification Benchmark — SEA-AD MTG

Training data:Allen Brain Atlas single-cell RNA-seq — Seattle Alzheimer's Disease Brain Cell Atlas (SEA-AD)