Hugging Face's logo

nvidia
/
llama-nemotron-embed-vl-1b-v2

Feature Extraction
Transformers
Safetensors
multilingual
llama_nemotron_vl
retrieval
visual document retrieval
vlm embedding
page image embedding
text embedding
semantic search
question-answering retrieval
rag
custom_code
Model card Files
xet
Community
4

Use default attention implementation with option to override

#2
by nvidia-oliver-holworthy - opened
base: refs/heads/main
←
from: refs/pr/2
Discussion Files changed
+6
-11
nvidia-oliver-holworthy
 NVIDIA org
•
edited Jan 7

Enables specifying attn_implementation when loading model including spda

 Add support for sdpa and enable changing attn_implementation5b5ca69c
mrdbourke

Thank you!

nvidia-oliver-holworthy changed pull request status to open
nvidia-oliver-holworthy
 NVIDIA org

Merged this functionality to configure attn_implementation in #4

nvidia-oliver-holworthy changed pull request status to closed

Sign up or log in to comment