This is the the inference container which can be used with Weaviate
text2vec-transformers module. You can download it directly from Dockerhub
using one of the pre-built images.
It is built to support only static (model2vec) models.
Model2Vec models significantly surpass other static embeddings models like GLoVe and BPEmb in performance, and the pre-built Docker images containing these models are notably compact in size.
| Model Name | Language | Image Name |
|---|---|---|
potion-multilingual-128M |
Multilingual | semitechnologies/model2vec-inference:minishlab-potion-multilingual-128M |
potion-retrieval-32M |
English | semitechnologies/model2vec-inference:minishlab-potion-retrieval-32M |
potion-base-4M |
English | semitechnologies/model2vec-inference:minishlab-potion-base-4M |
potion-base-8M |
English | semitechnologies/model2vec-inference:minishlab-potion-base-8M |
potion-base-32M |
English | semitechnologies/model2vec-inference:minishlab-potion-base-32M |