VaultGemma tokenizer alignment

VaultGemma is launched: Open source large models trained by differential privacy and the law of scale are fully solved

VaultGemma is trained on differential privacy to build a Gemma variant with about 1 billion parameters from scratch; The official released the "Law of...