Vesnica Vesnica

BuildKit 作用

docker run -dti \
    --name buildkitd \
    --security-opt seccomp=unconfined \
    --security-opt apparmor=unconfined \
    --device /dev/fuse \
    -p 8080:8080 \
 -v $(pwd)/buildkitd.toml:/etc/buildkitd.toml \

https://forums.developer.nvidia.com/t/problem-running-optix-7-6-in-wsl/239355/2

MIT Technology Review
Nature
https://ciechanow.ski/

Official tutorial won't work if you behind a corp proxy

Add the following code to fix:

import openai

openai.proxy = "http://IP:PORT"

Additionally, you can set api key like:

openai.api_key_path = "path/to/the/key"

https://docstore.mik.ua/orelly/unix/upt/ch15_02.htm
touch tmp32332hello32232_zzz
ls tmp*([0-9])*([a-z])*([0-9])_+(z)
visudo -s -c -f rulefile 检查规则文件语法是否无误

启动一个alpine容器：docker run -dti --name alpine alpine sh
登入该容器：docker exec -ti alpine sh
安装必要的包（可能需要先设置代理）：apk add wget alpine-sdk linux-headers libowfat-dev zlib-dev
从https://git.alpinelinux.org/aports/tree/community/opentracker/APKBUILD中获取opentracker源码包位置
下载opentracker源码包：wget https://dev.alpinelinux.org/opentracker/opentracker-0_git20210823.tar.gz
解压并进入目录：tar zxvf https://dev.alpinelinux.org/opentracker/opentracker-0_git20210823.tar.gz; cd opentracker
编译：make PREFIX=/usr/include LDFLAGS+=-static LDFLAGS+=-l:libowfat.a
得到静态编译的opentracker二进制程序，将其拷贝出容器

Use xvfb to start jupyter-lab: xvfb-run -s "-screen 0 1400x900x24" jupyter-lab
Must install old version gym: pip3 install gym==0.19
Must install old version pyglet: pip3 install pyglet==1.5.27

	# Force model to always use specified device
	# Place in `ComfyUI\custom_nodes` to use
	# City96 [Apache2]
	#
	import types
	import torch
	import comfy.model_management

	class OverrideDevice:
	@classmethod

	import pathlib
	import json
	import itertools

	from sentence_transformers import SentenceTransformer
	from qdrant_client import QdrantClient, models

	m = SentenceTransformer("shibing624/text2vec-base-chinese-nli")

	client_mem = QdrantClient(":memory:")