BERT and MosaicBERT runtime packages used by nucleotide language model ports.
Taykhoom Dalal
Taykhoom
AI & ML interests
None yet
Recent Activity
new activity 11 minutes ago
morrislab/rna-loc-fazal:Questions about binary classification labels of subcellular localization datasets updated a collection about 16 hours ago
BERT-updated updated a collection about 16 hours ago
BERT-updatedOrganizations
Minimal Hugging Face packages for eight GENA-LM genomic language model checkpoints.
Regulatory-genomics benchmark datasets from Tang et al., Genome Biology 2025.
Huggingface dataset versions of the GAMBA functional-region and ATG evaluation datasets.
HF ports of gLM2: 2 model versions - 150M and 650M parameters.
HF ports of Evo2: 7 model versions ranging from 1B to 40B parameters with different context windows.
HF ports of AIDO.RNA: 7 model versions ranging from 1M to 1.6B parameters.
HF port of CodonBERT: BERT pretrained on codon-level mRNA sequences.
HF ports of ERNIE-RNA: Base, MRL fine-tuned, and SS fine-tuned versions.
HF ports of DNABERT, DNABERT-2, and DNABERT-S: 6 model versions.
HF port of RNA-MSM: MSA Transformer trained on RNA sequence alignments.
HF ports of RiNALMo: 3 model versions ranging from 33M to 651M parameters.
Minimal Hugging Face port of the CDS-aware mRNABERT RNA language model.
Minimal Hugging Face packages for the ModernGENA base and large genomic language models.
Processed MPRA, variant-pair, and saturation-mutagenesis datasets from Siraj et al. (Nature 2026).
Huggingface dataset versions of the five DART-Eval regulatory DNA benchmark tasks.
HF ports of AIDO.DNA: 2 model versions - 300 million and 7 billion parameters.
HF ports of Evo1: 3 models spanning 8K-131K context and 300B-450B pretraining tokens.
HF ports of RNA-FM and mRNA-FM: 2 model versions for non-coding and coding RNA.
HF ports of RNAErnie: original (PaddlePaddle) and PyTorch retrain.
HF ports of SpliceBERT: 3 model versions (510nt and 1024nt, vertebrate and human).
HF ports of UTRBERT: 4 model versions ranging from 3-mer to 6-mer tokenization.
HF port of RNABERT: RNA language model for structural clustering and alignment.
Minimal Hugging Face ports of four UTR-LM checkpoints spanning MLM, MFE regression, and secondary-structure objectives.
BERT and MosaicBERT runtime packages used by nucleotide language model ports.
Minimal Hugging Face port of the CDS-aware mRNABERT RNA language model.
Minimal Hugging Face packages for eight GENA-LM genomic language model checkpoints.
Minimal Hugging Face packages for the ModernGENA base and large genomic language models.
Regulatory-genomics benchmark datasets from Tang et al., Genome Biology 2025.
Processed MPRA, variant-pair, and saturation-mutagenesis datasets from Siraj et al. (Nature 2026).
Huggingface dataset versions of the GAMBA functional-region and ATG evaluation datasets.
Huggingface dataset versions of the five DART-Eval regulatory DNA benchmark tasks.
HF ports of gLM2: 2 model versions - 150M and 650M parameters.
HF ports of AIDO.DNA: 2 model versions - 300 million and 7 billion parameters.
HF ports of Evo2: 7 model versions ranging from 1B to 40B parameters with different context windows.
HF ports of Evo1: 3 models spanning 8K-131K context and 300B-450B pretraining tokens.
HF ports of AIDO.RNA: 7 model versions ranging from 1M to 1.6B parameters.
HF ports of RNA-FM and mRNA-FM: 2 model versions for non-coding and coding RNA.
HF port of CodonBERT: BERT pretrained on codon-level mRNA sequences.
HF ports of RNAErnie: original (PaddlePaddle) and PyTorch retrain.
HF ports of ERNIE-RNA: Base, MRL fine-tuned, and SS fine-tuned versions.
HF ports of SpliceBERT: 3 model versions (510nt and 1024nt, vertebrate and human).
HF ports of DNABERT, DNABERT-2, and DNABERT-S: 6 model versions.
HF ports of UTRBERT: 4 model versions ranging from 3-mer to 6-mer tokenization.
HF port of RNA-MSM: MSA Transformer trained on RNA sequence alignments.
HF port of RNABERT: RNA language model for structural clustering and alignment.
HF ports of RiNALMo: 3 model versions ranging from 33M to 651M parameters.
Minimal Hugging Face ports of four UTR-LM checkpoints spanning MLM, MFE regression, and secondary-structure objectives.