TeleChat-3B

Training compute

1.4×10²² FLOP

Parameters

3B

Published

Apr 1, 2024

TeleChat-3B is an AI model developed by China Telecom (China), first published in April 2024. It works in the language domain, on tasks such as language modeling/generation, chat, question answering and 3 more.

Training it took an estimated 1.4×10²² FLOP of compute (estimation method: operation counting). The model has 3,000,000,000 parameters. Training ran on 640 NVIDIA A100 SXM4 40 GB.

Access: Unreleased. Its weights are not openly released. The reference paper has 13 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: China Telecom
Country of organization: China
Domain: Language
Task: Language modeling/generation, Chat, Question answering, Text summarization, Code generation, Translation
Training compute: 1.4×10²² FLOP
Compute estimation method: Operation counting
Parameters: 3,000,000,000
Training hardware: NVIDIA A100 SXM4 40 GB
Chips used: 640
Training power draw: 506.2 kW
Model accessibility: Unreleased
Open weights: No
Citations: 13
Epoch confidence: Confident