PharMolix Large Reasoning Models
Performance on Small Molecule Docking
ChatDD- R1
Large reasoning model series; other proprietary models: OCSU…
PharMolix-FM
All-atom foundation model
BioMedGPT
Multimodal reasoning model: 17B
OpenBioMed
960
Starred
Integrates more than ten in-house advanced algorithms; an easy-to-use workflow and agent-building solution that simulates scientific experiments; helps researchers construct multi-tool collaborative workflows to tackle complex scientific problems, simulate trial-and-error, and derive potential discoveries via large language models.
Knowledge-Graph Technology
SnowballKG Agent
- Iterative extraction: adopts a few-shot learning scheme starting with a small seed dataset and a human-in-the-loop feedback mechanism to codify expert knowledge into the KG.
- Self-learning & self-adaptation: in each iteration the system performs active-learning tuning based on extraction results, continuously improving performance on vertical subtasks.
Multilingual Cross-modal Retrieval Technology
- Multimodal fusion: combine data from different modalities for more comprehensive understanding and retrieval—for example, aligning text descriptions with image content to improve accuracy and relevance.
- Semantic understanding & matching: leverage deep learning to perform semantic understanding and feature extraction across modalities, enabling matching at the semantic level.
Academic Research
Arvix
BioMedGPT: Open Multimodal Generative Pre-trained Transformer for BioMedicine
KDD2024
Learning Multi-view Molecular Representations with Structured and Unstructured Knowledge
ICML2024
ESM All-Atom: Multi-Scale Protein Language Model for Unified Molecular Modeling
ICML2024
Mol-AE: Auto-Encoder Based Molecular Representation Learning With 3D Cloze Test Objective
ICML2024
LangCell: Language-Cell Pre-training for Cell Identity Understanding
NeurIPS2024
MutaPLM: Protein Language Modeling for Mutation Explanation and Engineering
SPJ
ECG-LM: Understanding Electrocardiogram with a Large Language Model