Deeplearning.ai text-embedding model error. Change to text-embedding-005
Deeplearning.ai text-embedding model error for „textembedding-gecko@001” requires You to simply change it to another one. Instead of the original pretrained models we should use the newer ones.
Those models are succeeding the gecko series in Vertex AI.
Google Vertex AI Options
- text-embedding-005: English/code-focused, 768 dims, up to 2048 tokens.
- gemini-embedding-001: Highest quality, 3072 dims, multilingual support.
- text-multilingual-embedding-002: For non-English tasks, 768 dims.
OpenAI Options
Those are just some of the available models.
Usage Comparison
| Model | Provider | Dims | Max Tokens | Best For |
|---|---|---|---|---|
| text-embedding-005 | 768 | 2048 | English/code | |
| gemini-embedding-001 | 3072 | Varies | High quality | |
| text-embedding-3-small | OpenAI | 1536 | 8191 | General/RAG |
Summary
Once upon a time, and still we can find many many things thanks to start overflow and people willing to share knowledge… let us hope the next time you google something like :
„Deeplearning.ai text-embedding model not working”
will be able to provide us an answer.
