SOTAVerified

Description-guided molecule generation

The significance of description-based molecule generation lies in its potential to streamline the process of molecular design by enabling the production of molecules that directly meet the criteria outlined in a given description. This facilitates a more targeted approach in the creation and optimization of novel molecules, with applications in diverse fields such as drug discovery and materials science.

Papers

Showing 12 of 2 papers

TitleStatusHype
TOMG-Bench: Evaluating LLMs on Text-based Open Molecule GenerationCode1
Mol-Instructions: A Large-Scale Biomolecular Instruction Dataset for Large Language ModelsCode2
Show:102550

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1Claude-3.5wAcc35.92Unverified
2Gemini-1.5-prowAcc34.8Unverified
3GPT-4-turbowAcc34.23Unverified
4GPT-4owAcc32.29Unverified
5Claude-3wAcc30.47Unverified
6Llama-3.1-8B (OpenMolIns-large)wAcc27.22Unverified
7Galactica-125M (OpenMolIns-xlarge)wAcc25.73Unverified
8Llama3-70B-Instruct (INT4)wAcc23.93Unverified
9Galactica-125M (OpenMolIns-large)wAcc23.42Unverified
10Galactica-125M (OpenMolIns-medium)wAcc19.89Unverified