From 71e9c7a9de3d4f8bf82835661f0b5db5e593465f Mon Sep 17 00:00:00 2001 From: by321 Date: Thu, 24 Aug 2023 13:33:13 -0700 Subject: [PATCH] Update readme.md --- readme.md | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/readme.md b/readme.md index 6407d71..325e295 100644 --- a/readme.md +++ b/readme.md @@ -22,7 +22,7 @@ The most useful thing is probably the metadata command: python safetensors_util.py metadata input_file.safetensors -pm -Many safetensors files, especially LoRA files, have a __metadata__ field in the file header that records training information, such as learning rates, number of epochs, number of images used, etc. You can see how your favorite file was trained and perhaps use some of the training parameters for your own model in the future. +Many safetensors files, especially LoRA files, have a \_\_metadata\_\_ field in the file header that records training information, such as learning rates, number of epochs, number of images used, etc. You can see how your favorite file was trained and perhaps use some of the training parameters for your own model in the future. The optional **-pm** flag is meant to make output more readable. Because safetensors files only allow string-to-string dictionary in header, non-string values need to be quoted. Basically the **-pm** flag tries to turn this: @@ -36,7 +36,8 @@ into this: "img_count":60 } }, -You can create a JSON file containing a __metadata__ entry: + +You can create a JSON file containing a\_\_metadata\_\_ entry: { "__metadata__":{