How much GPU memory does USER2-small-ONNX need?

About 1.2 GB: 0.14 GB of weights and the runtime's own memory.

What is the cheapest GPU to run USER2-small-ONNX?

Right now one GTX 1650 at $0.06 an hour; RTX PRO 5000 at $0.45 an hour is the best value for its speed. Prices are live offers from GPU clouds, checked every 15 minutes.

Can USER2-small-ONNX run on an RTX 4090?

Yes: it needs about 1.2 GB and an RTX 4090 has 24 GB.

Models / USER2-small

onnx-community/USER2-small-ONNX

Embeddings · served with ONNX Runtime · ↓ 759 downloads a month · View on Hugging Face

onnx-community/USER2-small-ONNX needs about 1.2 GB of GPU memory. It runs on one GTX 1650 from $0.06/h right now; best value: RTX PRO 5000 at $0.45/h.

Weights: 140 MB
GPU memory needed: about 1.2 GB
Memory · disk: 8 GB · 30 GB
Licence: apache-2.0

Versions

Smaller versions (8-bit, 4-bit) need less GPU memory and answer a little less precisely.

full precisiononnx/model.onnx140 MB · 1.2 GB GPU 4-bit (bnb)onnx/model_bnb4.onnx90 MB · 1.1 GB GPU float16onnx/model_fp16.onnx70 MB · 1.1 GB GPU 8-bitonnx/model_int8.onnx30 MB · 1.1 GB GPU 4-bitonnx/model_q4.onnx90 MB · 1.1 GB GPU 4-bit, float16onnx/model_q4f16.onnx50 MB · 1.1 GB GPU 8-bitonnx/model_quantized.onnx30 MB · 1.1 GB GPU 8-bitonnx/model_uint8.onnx30 MB · 1.1 GB GPU

Hardware and deploy

Deploying ONNX models from the catalog is coming soon. Today, download the files from Hugging Face and upload them in Inference.

Questions

How much GPU memory does USER2-small-ONNX need?: About 1.2 GB: 0.14 GB of weights and the runtime's own memory.
What is the cheapest GPU to run USER2-small-ONNX?: Right now one GTX 1650 at $0.06 an hour; RTX PRO 5000 at $0.45 an hour is the best value for its speed. Prices are live offers from GPU clouds, checked every 15 minutes.
Can USER2-small-ONNX run on an RTX 4090?: Yes: it needs about 1.2 GB and an RTX 4090 has 24 GB.

Loading…