Crate eegdino_rs 

eegdino-rs

Rust inference crate for the EEG-DINO foundation model, built on the Burn ML framework.

EEG-DINO learns robust EEG representations via hierarchical self-distillation on 9 000+ hours of EEG data. This crate provides a faithful port of the encoder architecture with verified numerical parity (NRMSE < 1e-6) against the original PyTorch implementation.

Model sizes

Variant	Params	d_model	Heads	Layers	FFN dim
Small	4.6 M	200	8	12	512
Medium	33 M	512	16	16	1 024
Large	201 M	1 024	16	24	2 048

Quick start (builder)

use eegdino_rs::prelude::*;
use burn::backend::NdArray;

type B = NdArray;

let encoder = EegDinoEncoder::<B>::builder()
    .weights("weights/eeg_dino_small.safetensors")
    .size(ModelSize::Small)
    .device(Default::default())
    .build()?;

let signal = vec![0.0f32; 19 * 2000];
let result = encoder.encode_raw(&signal, 1, 19, 2000)?;
// result.shape == [1, 191, 200]

Batch encoding

let signals: Vec<Vec<f32>> = load_recordings();
// Single batched forward pass (fastest):
let result = encoder.encode_batch(&signals, 19, 2000)?;
// Or one-by-one:
let results = encoder.encode_many(&signals, 19, 2000);

Backends

Feature	Backend	Notes
ndarray (default)	CPU	Multi-threaded via Rayon + SIMD
blas-accelerate	CPU + Accelerate	Recommended on Apple Silicon
wgpu	GPU	Metal (macOS) / Vulkan (Linux)
wgpu-f16	GPU f16	Half-precision, 2x less memory

Re-exports

pub use config::ModelConfig;

pub use config::ModelSize;

pub use error::EegDinoError;

pub use error::Result;

pub use inference::EegDinoEncoder;

pub use inference::EegDinoEncoderBuilder;

pub use inference::EncodingResult;

pub use inference::EegDinoClassifier;

pub use inference::ClassificationResult;

pub use inference::detect_model_size;

pub use model::encoder::EEGEncoder;

pub use model::classifier::ClassificationModel;

pub use model::embedding::EmbeddingCache;

pub use model::embedding::PatchEmbedding;

Modules

config

error

Typed error type for the eegdino-rs public API.

inference

model

prelude

Convenience re-exports for common usage patterns.

Functions

init_threads

Configure the Rayon thread pool. Call once before model use.