FunSearch

Training compute

3.9×10²³ FLOP

Parameters

15B

Published

Dec 14, 2023

FunSearch is an AI model developed by Google DeepMind (United States), first published in December 2023. It works in the language and search domain, on tasks such as code generation.

Training it took an estimated 3.9×10²³ FLOP of compute (estimation method: hardware). The model has 15,000,000,000 parameters.

Access: Open weights (unrestricted). Its weights are openly available. It is built on top of PaLM 2. The reference paper has 593 citations. Epoch AI rates the confidence of this record as speculative.

Full record

Organization: Google DeepMind
Country of organization: United States
Domain: Language, Search
Task: Code generation
Training compute: 3.9×10²³ FLOP
Compute estimation method: Hardware
Parameters: 15,000,000,000
Training time: 48 h
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Base model: PaLM 2
Citations: 593
Epoch confidence: Speculative