skills.rest
NVIDIA-BioNeMo
NVIDIA-BioNeMoOFFICIAL
Explore all skills by this author →
WebsiteGitHub

kermt-embed

Extract per-molecule embeddings from KERMT checkpoints for SMILES processing.

413|62|Updated Jun 23, 2026
Search Tags:#embeddings#bioinformatics#chemistry#nvidia#molecular-analysis#kermt
One-click install
npx skills add https://github.com/NVIDIA-BioNeMo/bionemo-agent-toolkit --skill kermt-embed
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: kermt-embed
Source: https://github.com/NVIDIA-BioNeMo/bionemo-agent-toolkit/tree/main/plugins/bionemo-agent-toolkit/skills/kermt-embed
Command: npx skills add https://github.com/NVIDIA-BioNeMo/bionemo-agent-toolkit --skill kermt-embed

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires docker, nvidia-container-toolkit, rdkit, and includes scripts (resource) components.

What problem does it solve?

This skill automates the complex process of generating per-molecule embeddings from KERMT checkpoints, removing the need for manual feature engineering or complex pipeline orchestration.

Core Features & Use Cases

  • Automated Embedding Generation: Extracts atom and bond-level embeddings from various KERMT encoder checkpoints.
  • Integrated Validation: Performs pre-flight checks on system requirements, checkpoints, and input data to ensure successful execution.
  • Use Case: A researcher needs to generate high-dimensional representations for a large library of SMILES strings to perform downstream clustering or property prediction; this skill handles the entire pipeline from data cleaning to outputting the final .npy files.

Quick Start

Use the kermt-embed skill to extract molecular embeddings from the provided molecules.csv file using the default pretrained model.

Frequently Asked Questions about kermt-embed

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract molecular embeddings from SMILES strings for downstream analysis?▼

You can extract molecular embeddings by processing large-scale SMILES strings through KERMT encoder checkpoints, which generates high-dimensional .npy files suitable for clustering or property prediction.

What hardware and software do I need to generate KERMT molecular embeddings?▼

Generating KERMT molecular embeddings requires a CUDA-capable NVIDIA GPU, Docker, the nvidia-container-toolkit, and rdkit to execute the containerized extraction workflow.

Can I process a large library of molecules automatically without manual feature engineering?▼

Yes, automated embedding generation handles the entire pipeline from data cleaning to outputting final .npy files, removing the need for manual feature engineering or complex pipeline orchestration.

Does the molecular embedding extraction pipeline validate checkpoints and input data?▼

Yes, integrated validation performs pre-flight checks on system requirements, KERMT checkpoints, and input data to ensure successful execution of the molecular embedding extraction.

What level of molecular embeddings can I extract from KERMT checkpoints?▼

KERMT checkpoint extraction supports generating atom-level and bond-level molecular embeddings, alongside multi-readout outputs for comprehensive chemical analysis.

Are there limitations when using Docker for molecular embedding extraction?▼

The containerized workflow depends on Docker and the nvidia-container-toolkit to access GPU resources, meaning execution is limited to systems with supported NVIDIA GPU hardware configurations.

Related Skills

Explore More Education & Research Skills→
NVIDIA-BioNeMoNVIDIA-BioNeMo
413

kermt-finetune

Finetune pretrained KERMT encoder models on labeled CSV datasets.

Official
Advanced
NVIDIA-BioNeMoNVIDIA-BioNeMo
413

kermt-infer

Execute KERMT model inference workflows on SMILES chemical datasets.

Official
Advanced
NVIDIA-BioNeMoNVIDIA-BioNeMo
413

kermt-add-cmim-pretrain

Convert grover_base checkpoints into hybrid cMIM-enabled models for continued pretraining.

Official
Advanced
jinzhezenggroupjinzhezenggroup
124

unimol

Run Uni-Mol embedding, training, and prediction workflows via a CLI wrapper.

Official
Advanced