What AI says about GENERator

GENERator is a suite of long-context generative genomic foundation models from GenerTeam that use a transformer decoder to generate DNA sequences, trained on large RefSeq-derived genomic data. The repository highlights model releases such as GENERator-eukaryote-1.2b-base and 3b-base, with updates and availability on platforms like HuggingFace. A key usage note is that input sequence lengths must be multiples of 6 due to a 6-mer tokenizer; if not, an out-of-vocabulary token may be appended, reducing generation quality.