From a4c4877064f6b90c7e699901e7f933ce861b5ae7 Mon Sep 17 00:00:00 2001 From: Sean Sube Date: Sat, 23 Dec 2023 12:46:02 -0600 Subject: [PATCH] fix(docs): add note about MIopen errors (#444) --- docs/user-guide.md | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/docs/user-guide.md b/docs/user-guide.md index 382cfa46..22816230 100644 --- a/docs/user-guide.md +++ b/docs/user-guide.md @@ -127,6 +127,7 @@ Please see [the server admin guide](server-admin.md) for details on how to confi - [Shape mismatch attempting to re-use buffer](#shape-mismatch-attempting-to-re-use-buffer) - [Cannot read properties of undefined (reading 'default')](#cannot-read-properties-of-undefined-reading-default) - [Missing key(s) in state\_dict](#missing-keys-in-state_dict) + - [Missing MIopen.so.1](#missing-miopenso1) - [Output Image Sizes](#output-image-sizes) ## Outline @@ -1740,6 +1741,16 @@ RuntimeError: Error(s) in loading state_dict for AutoencoderKL: Unexpected key(s) in state_dict: "encoder.mid_block.attentions.0.key.bias", "encoder.mid_block.attentions.0.key.weight", "encoder.mid_block.attentions.0.proj_attn.bias", "encoder.mid_block.attentions.0.proj_attn.weight", "encoder.mid_block.attentions.0.query.bias", "encoder.mid_block.attentions.0.query.weight", "encoder.mid_block.attentions.0.value.bias", "encoder.mid_block.attentions.0.value.weight", "decoder.mid_block.attentions.0.key.bias", "decoder.mid_block.attentions.0.key.weight", "decoder.mid_block.attentions.0.proj_attn.bias", "decoder.mid_block.attentions.0.proj_attn.weight", "decoder.mid_block.attentions.0.query.bias", "decoder.mid_block.attentions.0.query.weight", "decoder.mid_block.attentions.0.value.bias", "decoder.mid_block.attentions.0.value.weight". ``` +#### Missing MIopen.so.1 + +This can happen when you upgrade Torch to a version that was built against a more recent version of ROCm than the one +you are using. + +This can sometimes be fixed by upgrading ROCm, but make sure to check the supported architectures before you upgrade, +since they often remove support for older GPUs. + +If you cannot upgrade ROCm, downgrade Torch to the correct version for the libraries available on your machine. + ## Output Image Sizes You can use this table to figure out the final size for each image, based on the combination of parameters that you are