Add documentation and updates for various storage and service tools

- Create a new document for PhotoPrism with configuration options and Docker usage.
- Update rclone README to include a command for checking bucket size.
- Add Aliyun S3 endpoint information to the S3 storage README.
- Create a new document for Samba with comprehensive configuration and usage instructions.
- Enhance the storage-awesome document with additional distributed storage options and S3-compatible tools.
- Add a new torrent client to the torrent-awesome document.
- Include Inkscape and RapidRAW in the software-awesome document.
- Add a new section for shaders in the design-awesome document.
- Update ESBuild documentation to remove outdated information about decorators.
- Add key code references in the DOM event documentation.
- Include focus-related event listeners in the DOM FAQ.
- Update editor-awesome to reflect the correct repository for Milkdown.
- Add flamegraph profiling tool to Node.js debugging documentation.
- Include new lightweight S3 clients in the JavaScript awesome document.
- Update JavaScript FAQ with detailed information on Number properties.
- Enhance JavaScript I/O documentation with notes on stream usage.
- Add FAQ section comparing orpc and trpc in the orpc documentation.
- Update TypeScript README with commands for file listing and explanation.
- Add text-overflow behavior explanation in the CSS FAQ.
- Include a FAQ section for menu icons in the icon-awesome document.

Author: wenerme

notes/ai/ai-awesome.md

@@ -20,6 +20,14 @@ tags:
     - LLM Observability, Prompt Management, LLM Evaluations, Datasets, LLM Metrics, Prompt Playground
   - https://openrouter.io
   - https://together.ai
+- Engine
+  - MLX
+  - tinygrad
+  - PyTorch
+  - llama.cpp
+  - stablediffusion.cpp
+- Accelerators
+  - [huggingface/optimum](https://github.com/huggingface/optimum)
 - Search/RAG
   - Perplexity
   - [ItzCrazyKns/Perplexica](https://github.com/ItzCrazyKns/Perplexica)
@@ -30,7 +38,7 @@ tags:
     - MIT, TS
     - MiniPerplx -> scira
     - minimalistic AI-powered search engine
-- Client/Desktop/App
+- Client/Desktop/App/Runner/Server
   - [chatbox](./service/chatbox.md)
     - GPLv3, TypeScript, React, Electron
     - 支持 Desktop, iOS, Android, Web
@@ -39,8 +47,12 @@ tags:
     - MIT
   - [CherryHQ/cherry-studio](https://github.com/CherryHQ/cherry-studio)
     - ~~Apache-2.0~~, TS
+    - 聚合客户端 / Service Aggregator
     - 功能更加全面、综合
     - 桌面端为主
+  - [lmstudio](https://lmstudio.ai/)
+    - LM Runner/Server
+    - 闭源
   - [mainframecomputer/fullmoon-ios](https://github.com/mainframecomputer/fullmoon-ios)
     - MIT, Swift
   - [languine-ai/languine](https://github.com/languine-ai/languine)
@@ -124,6 +136,8 @@ tags:
 - Agent/MCP
   - [microsoft/autogen](https://github.com/microsoft/autogen)
     - Multi Agent
+  - [FoundationAgents/OpenManus](https://github.com/FoundationAgents/OpenManus)
+    - MIT, Python
   - [magnitudedev/magnitude](https://github.com/magnitudedev/magnitude)
     - Aapche-2.0, TS
     - browser automation framework
@@ -287,11 +301,15 @@ tags:
 
 ## Reading
 
+- from scratch
+  - [jsbaan/transformer-from-scratch](https://github.com/jsbaan/transformer-from-scratch)
+  - [GeeeekExplorer/nano-vllm](https://github.com/GeeeekExplorer/nano-vllm)
 - https://github.com/microsoft/generative-ai-for-beginners
 - https://www.promptingguide.ai/zh/readings
 - https://jalammar.github.io/
 - https://karpathy.ai/zero-to-hero.html
 - https://github.com/premAI-io/state-of-open-source-ai
+- https://theaisummer.com/diffusion-models/
 
 ## Prompt
 

notes/ai/ai-glossary.md

@@ -72,6 +72,22 @@ tags:
   - 用于可视化卷积神经网络（CNN）在图像分类任务中关注的区域
   - 通过将特定类别的预测与输入图像的特征图进行关联，生成热力图，显示模型在做出决策时关注的图像区域
   - 模型可解释性
+- diffusion
+  - 一种技术范式，广义概念
+  - 扩散过程
+  - 加噪-去噪
+  - High Concentration -> Low Concentration
+  - 物理上: Passive, Concentration Gradient, Temperature, Surface Area
+  - 通常在像素空间 (Pixel Space) 直接操作
+- Stable Diffusion
+  - 一种特定的扩散模型/应用，专注于图像生成任务
+  - 在潜空间 (Latent Space) 中操作
+    - e.g. 512x512 pixels -> 64x64 latent
+  - 引入 VAE
+  - 引入 文本引导（Text-Conditioning） - CLIP
+- VAE - Variational Autoencoder - 变分自编码器
+- ControlNet
+  - [lllyasviel/ControlNet](https://github.com/lllyasviel/ControlNet)
 
 | en               | cn       |
 | ---------------- | -------- |

notes/ai/dev/comfyui.md

@@ -12,14 +12,15 @@ title: ComfyUI
 - Control Net
   - SD 1.5
 - Juggernaut
-- 参考
-  - https://github.com/Comfy-Org/ComfyUI-Manager
 - 生产部署
   - ComfyDeploy
   - 另存 (API 格式)
   - 调用 /prompt
 - 参考
+  - [BennyKok/comfyui-deploy](https://github.com/BennyKok/comfyui-deploy)
+  - [Comfy-Org/ComfyUI-Manager](https://github.com/Comfy-Org/ComfyUI-Manager)
   - https://docs.comfy.org/tutorials/
+  - Transform Your Videos with Professional Quality: WAN-VACE V2V Complete Workflow for ComfyUI https://civitai.com/articles/16401
 
 :::caution
 

notes/ai/dev/rag.md

@@ -0,0 +1,9 @@
+---
+title: RAG
+---
+
+# RAG
+
+- [chonkie-inc/chonkie](https://github.com/chonkie-inc/chonkie)
+  - MIT, Python
+  - RAG chunking library

notes/ai/inference.md

@@ -31,10 +31,9 @@ title: 推理
 
 ## Service
 
+- toruchrun
 - vLLM
-  - Text
 - SGLang
-  - Text, Image
 - Image
   - ComfyUI
   - SD
@@ -43,6 +42,16 @@ title: 推理
 - Embeddings
   - [michaelfeil/infinity](https://github.com/michaelfeil/infinity)
     - MIT, Python
+- [exo-explore/exo](https://github.com/exo-explore/exo)
+  - GPLv3, Python
+- [Tencent/ncnn](https://github.com/Tencent/ncnn)
+- [InternLM/lmdeploy](https://github.com/InternLM/lmdeploy)
+  - Apache-2.0, Python, C++
+  - based on [MMRazor](https://github.com/open-mmlab/mmrazor) and [MMDeploy](https://github.com/open-mmlab/mmdeploy)
+- [mit-han-lab/nunchaku](https://github.com/mit-han-lab/nunchaku)
+  - Apache-2.0, Python, C++
+  - Nunchaku is a high-performance inference engine optimized for 4-bit neural networks
+  - SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models
 
 ## 模型大小 {#model-size}
 

notes/ai/ml/label-studio.md

@@ -13,10 +13,11 @@ title: Label Studio
 - 参考
   - https://labelstud.io/
   - 前端 https://github.com/HumanSignal/label-studio/tree/develop/web/libs/editor
+- /Users/wener/Library/Application Support/label-studio
 
 ```bash
-pip install -U label-studio
-label-studio
+uv pip install -U label-studio
+LOCAL_FILES_SERVING_ENABLED=true uv run label-studio
 
 # https://hub.docker.com/r/heartexlabs/label-studio
 # https://github.com/HumanSignal/label-studio/blob/develop/docker-compose.yml
@@ -28,8 +29,8 @@ docker run --rm -it \
 # label-studio --log-level DEBUG
 
 LABEL_STUDIO_BASE_DATA_DIR=$PWD/data \
-LABEL_STUDIO_LOCAL_FILES_SERVING_ENABLED=true \
-LABEL_STUDIO_LOCAL_FILES_DOCUMENT_ROOT=$PWD/files \
+  LABEL_STUDIO_LOCAL_FILES_SERVING_ENABLED=true \
+  LABEL_STUDIO_LOCAL_FILES_DOCUMENT_ROOT=$PWD/files \
   label-studio start
 ```
 
@@ -50,7 +51,6 @@ LABEL_STUDIO_LOCAL_FILES_DOCUMENT_ROOT=$PWD/files \
 
 ## tags
 
-
 ```xml
 <View>
   <TimeSeriesLabels name="label" toName="ts">
@@ -81,22 +81,70 @@ ts.brushRange.map(n=>(+n).toFixed(2));_=r();setInterval($=>r().some((n,i)=>n!==_
 interface Obj {
   id: string;
 
-  data:any
-  value:any
+  data: any;
+  value: any;
 
   from_name: string;
   to_name: string;
   type: string;
 }
 ```
 
+- Label Studio Common Format
+
 ## ML Backend
 
 - https://github.com/HumanSignal/label-studio-ml-backend
 - https://github.com/seblful/label-studio-yolov8-backend
 
+## Template
+
+- data 是静态的、被观察的对象
+- annotations 是动态的、被创建的结果
+- value 绑定 data
+- 控制标签创建 annotations
+
+```tsx
+<Text name='text' value='$prediction' />
+```
+
+- name 是内部名字
+- value
+
+---
+
+- 参考
+  - https://labelstud.io/tags/
+
+## Storage
+
+```
+/project-folder/
+├── images/
+│   ├── image1.jpg
+│   ├── image2.png
+│   └── subfolder/
+│       └── image3.gif
+└── my_tasks.json
+```
+
+- Treat every bucket object as a source file
+- 本地存储
+  - `LABEL_STUDIO_LOCAL_FILES_SERVING_ENABLED=true`
+  - LABEL_STUDIO_LOCAL_FILES_DOCUMENT_ROOT
+    - 限定访问目录
+  - `/data/local-files?d=ds/images/1.jpg`
+    - 实际相当于访问 `http://localhost:8080/data/local-files?d=ds/images/1.jpg`
+- 上传存储
+  - `/data/upload/$PROJECT_ID/`
+- https://labelstud.io/guide/storage.html
+
 # FAQ
 
+- OCR 指标
+  - 字符错误率 (CER)
+  - 词错误率 (WER)
+
 ## video frameCount
 
 - framerate 默认 24

notes/ai/ml/labeling.md

@@ -21,13 +21,19 @@ tags:
 - 参考
   - https://github.com/KKKSQJ/DeepLearning/tree/master/others/label_convert
   - https://openvinotoolkit.github.io/datumaro/latest/docs/data-formats/formats/index.html
+  - [A Guide to Bounding Box Formats](https://dragoneye.ai/blog/a-guide-to-bounding-box-formats)
 - [datumaro](./datumaro.md)
-- box
-  - xywh
-  - xywhn
-  - xyxy
-  - xyxyn
-  - cycywh
+
+| format  | spec                              | adopted by     |
+| ------- | --------------------------------- | -------------- |
+| xyxy    | left, top, right, bottom          | Pascal VOC     |
+| xyxyn   |                                   | Albumentations |
+| xywh    | left, top, width, height          | COCO           |
+| xywhn   |                                   |                |
+| cxcywh  | center x, center y, width, height | CreateML       |
+| cxcywhn |                                   | YOLO           |
+
+- n -> Normalized
 
 ## YOLO
 

notes/ai/ml/ml-dataset.md

@@ -20,6 +20,7 @@ title: Dataset
   - 增值税发票数据集-适配PaddleOCR https://aistudio.baidu.com/datasetdetail/165561
 - [huggingface/datasets](https://huggingface.co/docs/datasets/index)
 - [ConardLi/easy-dataset](https://github.com/ConardLi/easy-dataset)
+  - AGPLv3, JS
   - creating fine-tuning datasets for LLM
   - 不支持 图片数据集
 - [open-sciencelab/GraphGen](https://github.com/open-sciencelab/GraphGen)
@@ -31,6 +32,18 @@ cd easy-dataset
 pnpm i
 pnpm run build
 pnpm run start
+
+
+#
+docker cp easy-dataset:/app/prisma/db.sqlite easy-dataset/prisma/
+
+docker run --rm -it \
+  -p 1717:1717 \
+  -v $PWD/easy-dataset/data:/app/local-db \
+  -v $PWD/easy-dataset/prisma:/app/prisma \
+  --name easy-dataset ghcr.io/conardli/easy-dataset
+
+# docker exec -it easy-dataset npx prisma db push
 ```
 
 - 数据集格式

notes/ai/ml/onnx/onnx-runtime.md

@@ -1,6 +1,6 @@
 ---
 tags:
-- Runtime
+  - Runtime
 ---
 
 # ONNX Runtime
@@ -17,6 +17,8 @@ tags:
     - [oramasearch/onnx-go](https://github.com/oramasearch/onnx-go)
     - [AdvancedClimateSystems/gonnx](https://github.com/AdvancedClimateSystems/gonnx)
   - https://onnxruntime.ai/
+  - https://github.com/Aimol-l/OrtInference
+    - yolov10,yolov10+SAM ,yolov10+bytetrack , SAM2, paddleOCR
 - Inferencing
   - [Generative AI](https://github.com/microsoft/onnxruntime-genai) 实验阶段
 

notes/ai/ml/paddle/README.md

@@ -113,6 +113,26 @@ docker run --rm -it \
   --name paddlex ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlepaddle/paddle:3.0.0b2 /bin/bash
 ```
 
+## hpip
+
+```py
+from paddlex import create_pipeline
+
+pipeline = create_pipeline(
+    pipeline="OCR",
+    use_hpip=True,
+    hpi_config={"backend": "onnxruntime"}
+)
+```
+
+- backend
+  - paddle_infer
+  - openvino
+  - onnx_runtime
+  - tensorrt
+- High-Performance Inference
+- https://paddlepaddle.github.io/PaddleX/3.0/pipeline_deploy/high_performance_inference.html
+
 ## Paddle2ONNX
 
 - https://github.com/PaddlePaddle/PaddleOCR/blob/main/deploy/paddle2onnx/readme_ch.md