The repo contains the exact checkpoints used to produce Tables 1–8 of the paper for both the paper's main contributions (CGM-JEPA, X-CGM-JEPA) and the two re-pretrained baselines (GluFormer, TS2Vec).

Code: github.com/cruiseresearchgroup/CGM-JEPA
Pretraining Dataset: CRUISEResearchGroup/CGM-JEPA-Pretraining
Downstream Splits: CRUISEResearchGroup/CGM-JEPA-Downstream

Quick start

huggingface-cli download CRUISEResearchGroup/CGM-JEPA --local-dir Output

Then from the code repository:

bash

# Reproduce paper Tables 1–6
python scripts/run_all_eval.py

The downstream eval will load all four checkpoints automatically from the subdirectories below.

Layout

.
├── cgm_jepa/
│   ├── model.safetensors
│   └── config.json
├── x_cgm_jepa/
│   ├── model.safetensors
│   └── config.json
└── baselines/
    ├── gluformer.pt
    └── ts2vec.pkl

cgm_jepa/ and x_cgm_jepa/ use the standard PyTorchModelHubMixin layout — model.safetensors for weights, config.json for architecture hyperparameters — so they load via the standard from_pretrained one-liner.

Loading examples

CGM-JEPA / X-CGM-JEPA — `from_pretrained` one-liner

Encoder is a PyTorchModelHubMixin subclass, so the architecture hyperparameters and weights load in a single call directly from this repo:

python

from models.encoder import Encoder

encoder = Encoder.from_pretrained("CRUISEResearchGroup/CGM-JEPA", subfolder="cgm_jepa")
encoder.eval()

# X-CGM-JEPA: same call, different subfolder
encoder_x = Encoder.from_pretrained("CRUISEResearchGroup/CGM-JEPA", subfolder="x_cgm_jepa")

Standalone PyTorch — GluFormer

python

import torch
import torch.nn as nn
from models.gluformer.gluformer import GluFormer

vocab_size = 278
gluformer = GluFormer(
    vocab_size=vocab_size,
    embed_dim=96,
    nhead=6,
    num_layers=3,
    dim_feedforward=192,
    max_seq_length=25000,
    dropout=0.0,
    pad_token=vocab_size,
)
gluformer.load_state_dict(
    torch.load("Output/baselines/gluformer.pt", map_location="cpu")["encoder"]
)
gluformer.output_head = nn.Identity()   # discard the LM head for embedding extraction
gluformer.eval()

Architectures

`cgm_jepa` and `x_cgm_jepa`

Both use the same models.encoder.Encoder class with identical hyperparameters; only the pretraining objective differs. At downstream / inference time only the temporal encoder is used.

Field	Value
`patch_size`	12
`encoder_kernel_size`	3
`encoder_embed_dim`	96
`encoder_nhead`	6
`encoder_num_layers`	3

Input: a tensor of shape (B, num_patches, patch_size) (raw glucose values, z-scored). Output: per-patch embedding of shape (B, num_patches, embed_dim).

Intended use

Frozen feature extraction from raw CGM windows (24-hour, 5-min sampled, 288 timesteps).
Linear-probe evaluation, especially for the metabolic subphenotyping tasks (IR / β-cell dysfunction) described in the paper.
Comparison baseline for new CGM representation methods.

Citation

bibtex

@article{muhammad2026cgm,
  title   = {CGM-JEPA: Learning Consistent Continuous Glucose Monitor Representations via Predictive Self-Supervised Pretraining},
  author  = {Muhammad, Hada Melino and Li, Zechen and Salim, Flora and Metwally, Ahmed A},
  journal = {arXiv preprint arXiv:2605.00933},
  year    = {2026}
}

License

Released under the MIT license.

CRUISEResearchGroup/CGM-JEPA

CGM-JEPA Pretrained Encoders