| Feature | Gemini 1.0 | Gemini 1.5 |
|---|---|---|
| Number of parameters | 1.5T | 540B |
| Training data | 1.56T words | 1.56T words |
| Model architecture | Transformer-based decoder-only | Transformer-based decoder-only with MoE |
| Context window size | 32k tokens | 1M tokens |
| Inference speed | 10x faster than GPT-3 | 2x faster than Gemini 1.0 |
| Performance on language benchmarks | SOTA on GLUE, SuperGLUE, and CoQA | SOTA on GLUE, SuperGLUE, CoQA, and Winograd Schema Challenge |
| Availability | Invite-only beta | Public beta |
-
-
Save tanaikech/157bb4778149429e865d37312ad8a5da to your computer and use it in GitHub Desktop.
Difference between Gemini 1.0 and Gemini 1.5
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment