From a4df279fd2a5ac929375f1a87dd1c5e67d5ea09f Mon Sep 17 00:00:00 2001 From: Brett Kuprel Date: Wed, 29 Jun 2022 13:56:29 -0400 Subject: [PATCH] simplified attention and keys_values state resulted in decrease in inference time to 7.3 seconds (from ~10 seconds) --- README.md | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index d5d1afc..0f84204 100644 --- a/README.md +++ b/README.md @@ -2,7 +2,9 @@ [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/kuprel/min-dalle/blob/main/min_dalle.ipynb) -This is a minimal implementation of [DALL·E Mini](https://github.com/borisdayma/dalle-mini). It has been stripped to the bare essentials necessary for doing inference, and converted to PyTorch. The only third party dependencies are numpy, torch, and flax (and optionally wandb to download the models). DALL·E Mega inference with PyTorch takes about 10 seconds in Colab. +This is a minimal implementation of [DALL·E Mini](https://github.com/borisdayma/dalle-mini). It has been stripped to the bare essentials necessary for doing inference, and converted to PyTorch. The only third party dependencies are numpy, torch, and flax (and optionally wandb to download the models). + +DALL·E Mega inference with PyTorch takes 7.3 seconds in Colab to generate an avocado armchair ### Setup