search

dusty-nv / NanoDB

Zero-copy multimodal vector DB with CUDA and CLIP/SigLIP
MIT License
24 stars 3 forks source link

regarding dobuts about NanoDb #2

Open thalapandi opened 3 weeks ago

thalapandi commented 3 weeks ago

we are using nano Db for 'ViT-L/14@336px' but we wanted to use our tensorrt model for embedding extraction is how can we do that?

dusty-nv commented 2 weeks ago

Hi pandian, the code for NanoDB is pretty compact, I would just edit it to add your custom model. Or you can add to clip_trt, that is where my VIT implementations go (using torch2trt)

From: pandian p @.> Sent: Saturday, August 24, 2024 12:27:59 AM To: dusty-nv/NanoDB @.> Cc: Subscribed @.***> Subject: [dusty-nv/NanoDB] regarding dobuts about NanoDb (Issue #2)

we are using nano Db for @.***' but we wanted to use our tensorrt model for embedding extraction is how can we do that?

— Reply to this email directly, view it on GitHubhttps://github.com/dusty-nv/NanoDB/issues/2, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADVEGKYYUMTBLDAWLZNDYCLZTADU7AVCNFSM6AAAAABNBHUUQWVHI2DSMVQWIX3LMV43ASLTON2WKOZSGQ4DIMJZGMYDIMI. You are receiving this because you are subscribed to this thread.Message ID: @.***>

thalapandi commented 2 weeks ago

Thanks for your reply sir

thalapandi commented 2 weeks ago

Sir one more help for me Can you suggest me which one is the best model face embedding extraction for face recognition by applying faiss lite?