RAGFlow

# 克隆 ragflow 仓库
git clone https://github.com/infiniflow/ragflow.git

vi docker/.env

# Optimizations for MacOS
# Uncomment the following line if your OS is MacOS:
MACOS=1

# 进入项目根目录
cd ragflow/
poetry env use python3
source .venv/bin/activate

# 这一步是在 https://github.com/infiniflow/ragflow/issues/5367 这里提到的
pip3 install huggingface_hub nltk --index-url https://mirrors.aliyun.com/pypi/simple/

# import urllib.request
import requests

for url in urls:
    filename = url.split("/")[-1]
    print(f"Downloading {url}...")
    if not os.path.exists(filename):
        # urllib.request.urlretrieve(url, filename)
        response = requests.get(url, verify=True)
        with open(filename, 'wb') as f:
            f.write(response.content)

Downloading nltk wordnet...
[nltk_data] Error loading wordnet: <urlopen error [SSL:
[nltk_data]     CERTIFICATE_VERIFY_FAILED] certificate verify failed:
[nltk_data]     unable to get local issuer certificate (_ssl.c:1000)>

import ssl

try:
    _create_unverified_https_context = ssl._create_unverified_context
except AttributeError:
    pass
else:
    ssl._create_default_https_context = _create_unverified_https_context

# 漫长的下载
python3 download_deps.py

# 构建镜像
docker build -f Dockerfile.deps -t infiniflow/ragflow_deps .
docker build --build-arg NEED_MIRROR=1 --build-arg LIGHTEN=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .

# 为镜像打上最新标签
docker tag infiniflow/ragflow:nightly-slim infiniflow/ragflow:latest
docker tag infiniflow/ragflow:nightly-slim infiniflow/ragflow:v0.17.0
# 重新构建镜像
docker build --no-cache --build-arg NEED_MIRROR=1 --build-arg LIGHTEN=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .

# 拉取 ubuntu:22.04；先使用代理拉取，避免容器内还使用代理，造成安装缓慢。
docker pull docker.io/library/ubuntu:22.04

cd ragflow/docker
docker compose -f docker-compose-macos.yml pull
docker compose -f docker-compose-macos.yml up -d

# MLX-lm 用来在苹果 silicon 芯片上提供高效的 LLM 推理
pip install "xinference[mlx]" --index-url https://mirrors.aliyun.com/pypi/simple/
XINFERENCE_MODEL_SRC=modelscope XINFERENCE_HOME=./xinference xinference-local --host 0.0.0.0 --port 9997

# 这样可以启动下载好的模型
xinference launch --model-name bge-large-zh-v1.5 --model-type embedding
xinference launch --model-name bge-reranker-v2-m3 --model-type rerank
xinference launch --model-name ChatTTS --model-type audio
xinference launch --model-name SenseVoiceSmall --model-type audio

--- 查看模型设置
SELECT tenant_id, llm_factory, model_type, llm_name, max_tokens
FROM tenant_llm
WHERE tenant_id = 'df75265cf7da11ef9a840242ac120006';

ollama run deepseek-r1:70b

ollama run shaw/dmeta-embedding-zh:latest

ollama run llava:34b

pip3 install funasr

pip3 install sentence-transformers

pip3 install ChatTTS

ERROR: Knowledge bases use different embedding models.