From a11e00055ea8f8fea88b46480267014446fda1fa Mon Sep 17 00:00:00 2001 From: Daniel Korat Date: Thu, 22 Feb 2024 14:38:52 +0200 Subject: [PATCH] Update setfit-onnx-optimum.ipynb --- notebooks/setfit-onnx-optimum.ipynb | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/notebooks/setfit-onnx-optimum.ipynb b/notebooks/setfit-onnx-optimum.ipynb index 29f2d790..3914d8a1 100644 --- a/notebooks/setfit-onnx-optimum.ipynb +++ b/notebooks/setfit-onnx-optimum.ipynb @@ -935,7 +935,7 @@ "id": "gvdggvIbvowO" }, "source": [ - "By applying ONNX, we were able to improve the latency from 13.43ms per sample to 2.19ms per sample, for a speedup of 5.41x!\n", + "By applying ONNX, we were able to improve the latency from 8.31ms per sample to 1.54ms per sample, for a speedup of 5.41x!\n", "\n", "For further improvements, we recommend increasing the inference batch size, as this may also heavily improve the throughput. For example, setting the batch size to 128 reduces the latency further down to 0.3ms, and down to 0.2ms at a batch size of 2048." ]