NV-Embed-v2
NV-Embed-v2 presents several new designs, including having the LLM attend to latent vectors for better pooled embedding output, and demonstrating a two-staged instruction tuning method to enhance the accuracy of both retrieval and non-retrieval tasks.
text → embedding · made by NVIDIA
Sold by
Nobody in the catalogue publishes a price for this yet.
About
NV-Embed-v2 — a vectors model from NVIDIA.
It takes text and returns vectors. It was published in August 2024. The catalogue files it under embedding. Its weights are published, but the licence forbids selling what it produces.
Every current figure
- Maker
- NVIDIA
- Register
- model
- Takes
- text
- Returns
- embedding
- Published
- August 2024
- Parameters
- 7.9 billion · read from its own weights
- Licence
- cc-by-nc-4.0
Known as 1 name
nvidia/NV-Embed-v2