How do I access the Forge API models?

You can access advanced models like esm3-large by obtaining an API token from the EvolutionaryScale Forge platform and using the built-in Forge client in the skill.

What are the hardware requirements for local inference?

For local use, a CUDA-enabled GPU is recommended. Installing Flash Attention is also highly suggested to optimize inference speed and memory usage.

Can I use the ESM skill for local protein modeling?

Yes, the skill supports local model usage with open-weight versions like esm3-sm-open-v1, though larger production models require the Forge API.

What is the difference between ESM3 and ESM C in this skill?

ESM3 is a generative multimodal model used for designing protein sequences, structures, and functions, while ESM C is optimized for providing efficient, high-quality embeddings for sequence representation.

Does this skill support inverse folding?

Yes, the ESM skill enables inverse folding, allowing you to design specific protein sequences that will fold into a target 3D structure provided via PDB.

ESM Protein Language Models

Name: ESM Protein Language Models
Author: x-cmd

byx-cmd

•

データサイエンスとML

Powers protein engineering workflows with state-of-the-art generative design, structure prediction, and high-quality sequence embeddings.

ESM (Evolutionary Scale Modeling) provides a comprehensive toolkit for protein language models, enabling researchers and developers to design novel proteins, predict 3D structures, and generate efficient embeddings. By integrating ESM3 for multimodal generative design across sequence, structure, and function, and ESM C for representation learning, this skill facilitates complex biological tasks such as inverse folding, functional annotation, and high-throughput protein engineering using both local models and the scalable Forge API.