Find open-source science resources

A domain-optimized reasoning model built on DeepSeek-R1-Distill-Qwen-32B, refined through a multi-stage pipeline of GPTQ quantization-aware training and QLoRA fine-tuning. Achieves 84% on MedQA — within 4 points of GPT-4o — in a ~20GB package that fits on a single L40/L40s GPU.

# ibm/biomed.sm.mv-te-84m-MoleculeNet-ligand_scaffold-CLINTOX-101 biomed.sm.mv-te-84m is a multimodal biomedical foundation model for small molecules created using MMELON (Multi-view Molecular Embedding with Late Fusion), a flexible approach to aggregate multiple views (sequence, image, graph) of…

For a convenient overview and download list, visit our model page for this model.

Fine-tuned version of google/gemma-4-E4B-it across three professional domains — Medical, Legal, and Finance — using QLoRA (4-bit NF4) with Optuna-tuned hyperparameters, trained on Kaggle T4 GPU.

# ChemGPT 19M ChemGPT is based on the GPT-Neo model and was introduced in the paper Neural Scaling of Deep Chemical Models.

# ibm/biomed.sm.mv-te-84m-MoleculeNet-ligand_scaffold-QM7-101 biomed.sm.mv-te-84m is a multimodal biomedical foundation model for small molecules created using MMELON (Multi-view Molecular Embedding with Late Fusion), a flexible approach to aggregate multiple views (sequence, image, graph) of…

datasets: - UMLS

# ibm/biomed.sm.mv-te-84m-MoleculeNet-ligand_scaffold-SIDER-101 biomed.sm.mv-te-84m is a multimodal biomedical foundation model for small molecules created using MMELON (Multi-view Molecular Embedding with Late Fusion), a flexible approach to aggregate multiple views (sequence, image, graph) of…

In search engines, rerankers are crucial for improving the accuracy of your retrieval system.

Drugs must satisfy stringent criteria for both efficacy and safety. This model predicts the likelihood of failure in clinical toxicity trials for small-molecule drugs, represented using SMILES (Simplified Molecular Input Line Entry System) strings.

![Language: Multilingual]()

An Evolutionary-scale Model (ESM) for protein function prediction from amino acid sequences using the Gene Ontology (GO). Based on the ESM2 Transformer architecture, pre-trained on UniRef50, and fine-tuned on the AmiGO dataset, this model predicts the GO subgraph for a particular protein sequence -…

# ibm/biomed.sm.mv-te-84m-MoleculeNet-ligand_scaffold-ESOL-101 biomed.sm.mv-te-84m is a multimodal biomedical foundation model for small molecules created using MMELON (Multi-view Molecular Embedding with Late Fusion), a flexible approach to aggregate multiple views (sequence, image, graph) of…

Dans-PersonalityEngine-V1.3.0-24b Dans-PersonalityEngine-V1.3.0-24b ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⠀⠄⠀⡂⠀⠁⡄⢀⠁⢀⣈⡄⠌⠐⠠⠤⠄⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⡄⠆⠀⢠⠀⠛⣸⣄⣶⣾⡷⡾⠘⠃⢀⠀⣴⠀⡄⠰⢆⣠⠘⠰⠀⡀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠃⠀⡋⢀⣤⡿⠟⠋⠁⠀⡠⠤⢇⠋⠀⠈⠃⢀⠀⠈⡡⠤⠀⠀⠁⢄⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠁⡂⠀⠀⣀⣔⣧⠟⠋⠀⢀⡄⠀⠪⣀⡂⢁⠛⢆⠀⠀⠀⢎⢀⠄⢡⠢⠛⠠⡀⠀⠄⠀⠀ ⠀⠀⡀⠡⢑⠌⠈⣧⣮⢾⢏⠁⠀⠀⡀⠠⠦⠈⠀⠞⠑⠁⠀⠀⢧⡄⠈⡜⠷⠒⢸⡇⠐⠇⠿⠈⣖⠂⠀…

# GPN trained on Arabidopsis thaliana and 7 other Brassicales See https://github.com/songlab-cal/gpn for more details.

vitreg1s14lsdino-v2-dist-bio is a compact Bio-DINO image encoder distilled from the larger Bio-DINO SoViT-150M/14 model. It keeps the same natural-photography biodiversity scope as the teacher model, but uses a much smaller ViT-S/14-style student with 21.7M backbone parameters and 384-dimensional…

# ibm/biomed.sm.mv-te-84m-MoleculeNet-ligand_scaffold-MUV-101 biomed.sm.mv-te-84m is a multimodal biomedical foundation model for small molecules created using MMELON (Multi-view Molecular Embedding with Late Fusion), a flexible approach to aggregate multiple views (sequence, image, graph) of…

An Evolutionary-scale Model (ESM) for protein function prediction from amino acid sequences using the Gene Ontology (GO). Based on the ESM2 Transformer architecture, pre-trained on UniRef50, and fine-tuned on the AmiGO dataset, this model predicts the GO subgraph for a particular protein sequence -…

## Description: Geneformer is a foundational transformer model pretrained on a large-scale corpus of single-cell transcriptomes to enable context-specific predictions in settings with limited data in network biology.

Base model: google/gemma-4-26b-it Architecture: MoE — 26B total / ≈4B active parameters (1 shared expert + 8 routed from a pool of 128 per MoE layer, 30 MoE layers) Method: Activation-directed expert surgery — 128 → 64 experts per layer (50% reduction) Quantization: Q4KM (≈9.7 GB on disk) Tags:…

## Description: Geneformer is a foundational transformer model pretrained on a large-scale corpus of single-cell transcriptomes to enable context-specific predictions in settings with limited data in network biology. This model version was continually pretrained on ~14 million cancer transcriptomes…

# or·a·cle /ˈôrəkəl/ — a source of wise counsel; one who provides authoritative knowledge. From Latin ōrāculum, meaning divine announcement. In computer science, an oracle is a black box that always returns the correct answer — you don't ask it how it knows, you ask and it answers.

ONNX export of the Cellpose cpsam (Cellpose-SAM) model for cell segmentation in microscopy images.

darkknight25/deepseek-16b-medical-GPT is a fine-tuned version of deepseek-ai/deepseek-l6b-moe-chat, optimized for medical question answering, reasoning, and clinical summarization using QLoRA and open-access healthcare datasets.

> [!NOTE] > This model has been optimized using NVIDIA's TransformerEngine > library. Slight numerical differences may be observed between the original model and the optimized > model. For instructions on how to install TransformerEngine, please refer to the > official documentation.

For a convenient overview and download list, visit our model page for this model.

## Description: Geneformer is a foundational transformer model pretrained on a large-scale corpus of single-cell transcriptomes to enable context-specific predictions in settings with limited data in network biology.

COMO (Closed-loop Optical Molecule recOgnition) is a deep learning framework for Optical Chemical Structure Recognition (OCSR). It recognizes chemical structure diagrams from images and predicts SMILES strings with atom-level 2D coordinates and bond matrices.

ChemFIE-BED is a sentence-transformers based on gbyuvd/chemselfies-base-bertmlm fine-tuned on around (for now) 2 million pairs of valid molecules' SELFIES (Krenn et al. 2020) taken from COCONUTDB (Sorokina et al. 2021) and ChemBL34 (Zdrazil et al. 2023).

This model is a lightweight model pre-trained on SELFIES (Self-Referencing Embedded Strings) representations of molecules. It is trained on 2.7M unique and valid molecules taken from COCONUTDB and ChemBL34, with 7.3M total generated masked examples.

A compact protein language model distilled from ProtGPT2 using complementary-regularizer distillation---a method that combines uncertainty-aware position weighting with calibration-aware label smoothing to achieve 87% better perplexity than standard knowledge distillation at 20x compression.

Github | Cite

InstaNovo-P is a specialized transformer-based model for de novo peptide sequencing from phosphoproteomics mass spectrometry data. This model is specifically trained and optimized for identifying phosphorylated peptides and their modification sites.

An Evolutionary-scale Model (ESM) for protein function prediction from amino acid sequences using the Gene Ontology (GO). Based on the ESM2 Transformer architecture, pre-trained on UniRef50, and fine-tuned on the AmiGO dataset, this model predicts the GO subgraph for a particular protein sequence -…

Genos-m is a foundation model for human-associated microbial genomes. It is trained to model microbial DNA sequences at single-nucleotide resolution and supports ultra-long genomic contexts up to one million tokens.

# ModernGENA base ModernGENA is a DNA foundation model based on ModernBERT (a modernized BERT-style encoder architecture) adapted for genomic sequence modeling. ModernGENA base is the 377M-parameter version introduced in the paper Back to BERT in 2026: ModernGENA as a Strong, Efficient Baseline for…

!fffffff.png

Abstract:

This repository contains the full Bio-DINO DINOv2 training weights for a SoViT-150M/14 Vision Transformer trained on natural photographs of living organisms. It is the companion release to the Birder backbone checkpoints at .

ScANVI is a variational inference model for single-cell RNA-seq data that can learn an underlying latent space, integrate technical batches and impute dropouts. In addition, to scVI, ScANVI is a semi-supervised model that can leverage labeled data to learn a cell-type classifier in the latent space…