Google DeepMind·LaMDA / PaLM

PaLM

A 540B dense language model trained with Pathways, demonstrating large-scale multitask and few-shot capabilities.

LegacyLanguageWeights not released

MODEL NUMBER255

Specifications

Only published specifications are shown.

Total parameters

540B

Architecture

Dense

Input modalities

Text

Output modalities

Text

Open weights

Status

Legacy

A 540B dense language model trained with Pathways, demonstrating large-scale multitask and few-shot capabilities.

2022-04-04Announcement

PaLM was announced.

Select a card to view the related model.