Skip to content
Other

kron-ppg-age

streaming age estimator from PPG signals

2 min readupdated

The kron-ppg-age model estimates a person's chronological age in real time using raw photoplethysmography (PPG) data (an optical pulse measurement method). The architecture is based on the causal KronLSTM encoder, which processes the signal in 0.2-second batches and carries memory across chunks. The model weighs just 24 MB in ONNX format with 5.85 million parameters and can run directly on smartphones or smartwatches. It will be useful for biometric application developers and pulse signal researchers.

Key features

  • Compactness: the ONNX graph weighs 24 MB, and the total parameter count is 5.85M.
  • Streaming: real-time PPG processing in 5-second chunks with state transfer.
  • Accuracy: mean absolute error (MAE) on the test dataset is 6.38 years.
  • Training: pre-training for 160k steps and an MLP head on a frozen encoder.

What you can do

  • Mobile apps: running streaming age analysis on smartphones or watches.
  • Biometric analysis: research of PPG data from a finger or wrist (with preliminary calibration).
  • Tracker creation: developing real-time pulse wave monitoring systems.

Example

import json, numpy as np, onnxruntime as ort

meta = json.load(open("onnx/stream.json"))
sess = ort.InferenceSession("onnx/stream.onnx")
feeds = {i.name: np.zeros([1 if isinstance(d, str) else d for d in i.shape],
                          np.bool_ if i.type == "tensor(bool)" else np.float32)
         for i in sess.get_inputs()}

def step(chunk):
    feeds["samples"] = chunk[None].astype(np.float32)
    feeds["valid"] = np.ones((1, 25), np.bool_)
    out = dict(zip([o.name for o in sess.get_outputs()], sess.run(None, feeds)))
    for k, v in out.items():
        if k.endswith("_next"):
            feeds[k[:-5]] = v
    return float(out["age"][0])
Read moreDetails
Metrics: Test MAE 6.38 years (r = 0.84), compared to 11.74 for age median.
Requirements: 24 MB ONNX graph, 25 Hz input (0.5–8 Hz filtering needed).
Limitations: estimates chronological age, not health; requires calibration on the wrist.
Discuss Telegram 0 16

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.0 GB · ready files0.5 GB
  • FP8 / INT8recommendedweights 0.0 GB0.5 GB
  • 4-bitweights 0.0 GB0.5 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # GlucoseDao/kron-ppg-age
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download GlucoseDao/kron-ppg-age --local-dir ~/models/kron-ppg-age
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to kron-ppg-age
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of kron-ppg-age (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

GlucoseDao/kron-ppg-age
Author
GlucoseDao
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

160

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments