Skip to content
Search models or paste a model link. Press Enter to search.

ViT Base/32 (224, QuickGELU, CLIP, OpenAI)

timm/vit_base_patch32_clip_quickgelu_224.openai
Left and right arrow keys switch the vendor.
Embedl-optimized
timm vit_base_patch32_clip_quickgelu_224.openai architecture graph

vit_base_patch32_clip_quickgelu_224.openai is an image encoder from timm: it turns an image into a 512-dimensional embedding, for search, retrieval and as features for other models.

Methodology

  • Latency: trtexec on the NVIDIA L4, batch 1, 201 iterations after warm-up, during the build: GPU compute time per inference, mean and p99, and throughput.
  • Baseline TensorRT FP16: the same Embedl-optimized model before quantization, converted to FP16 with ModelOpt AutoCast (float32 inputs and outputs), compiled with TensorRT 11.3.0.99 and timed with the same trtexec settings on an NVIDIA L4.
  • Quantization: Embedl Deploy, activations INT8, weights INT8 per-channel; calibrated on Imagenette (1,000 images). Exported as ONNX opset 20, compiled with TensorRT 11.3.0.99, built on NVIDIA L4 (sm_89).
  • Build: config vit_base_patch32_clip_quickgelu_224/openai-timm_int8_embedl_tensorrt_l4, on an NVIDIA L4.

Files

Image Size

Model

docker.io/embedl/model-artifacts:vit_base_patch32_clip_quickgelu_224.openai-timm_int8_embedl_tensorrt_l4.f7db27bebb6967e8

0.79 GiB

Runtime

docker.io/embedl/models:runtimes.tensorrt.amd64.2026102005300fa3

0.52 GiB