PaLM 2 (Google) and Wu Dao 2.0 (Beijing Academy of Artificial Intelligence / BAAI) are both frontier AI models. PaLM 2 was published in May 2023 and Wu Dao 2.0 in May 2021.
PaLM 2 was trained on 7.3 x 10^24 FLOP, about 4.8x the compute of Wu Dao 2.0 at 1.5 x 10^24 FLOP. Training compute is the closest available proxy for how much was invested in a model, though it says nothing on its own about how well that compute was spent.
These two models share no benchmark on which both have been scored, so no direct performance comparison is possible here. The specification table below is a comparison of inputs, not of results.
| Field | PaLM 2 | Wu Dao 2.0 |
|---|---|---|
| Organization | Beijing Academy of Artificial Intelligence / BAAI | |
| Published | May 10, 2023 | May 31, 2021 |
| Training compute | 7.3 x 10^24 FLOP | 1.5 x 10^24 FLOP |
| Parameters | 340B | 1.8T |
| Dataset size | 3.6T | 4.9T |
| Training hardware | Google TPU v4 | — |
| Training cost (2023 USD) | $5M | $3M |
| Accessibility | API access | API access |
| Open weights | No | No |
| Country | United States | China |
