razhangwei

#!/bin/bash

# === Cache Cleanup Script ===
# Cleans caches for Homebrew, UV, Hugging Face, and NPM
# Usage: bash cleanup.sh [options]
#   Options:
#     -a, --all       Clean all caches
#     -b, --brew      Clean Homebrew cache
#     -u, --uv        Clean UV cache
#     -h, --huggingface Clean Hugging Face cache

razhangwei

How can I find the model implementation for a particular model?

• Usually it's avaiable from transformers/models/xxxx, e.g., for qwen2_5_vl

razhangwei

from typing import List, Tuple, Optional, Union, Callable
import numpy as np

# Define a Tensor type for clarity in this pseudocode
Tensor = np.ndarray  # In real implementation, this would be a framework-specific tensor type

# Qwen2.5-VL Vision Encoder Pseudocode

class Qwen25VisionEncoder:
    def __init__(self, 

razhangwei

Personal Website Tech Stack Cheatsheet

Core Technologies

Technology	Version	Purpose
Next.js	14.2.0	React framework with file-based routing, static site generation
React	18.2.0	UI library for component-based development
TypeScript	5.3.3	Static type-checking for JavaScript
TailwindCSS	3.4.1	Utility-first CSS framework

razhangwei

The Walrus Operator in Python: A Comprehensive Guide

The walrus operator (:=) was introduced in Python 3.8 as part of PEP 572. It's officially called the "assignment expression" operator, but it earned its nickname because := resembles a walrus with tusks when viewed sideways. This operator allows you to assign values to variables as part of an expression, rather than as a separate statement.

Basic Syntax and Purpose

The syntax is:

variable := expression

razhangwei

Key Features

• 10-20x faster than pip
• Written in Rust for high performance
• Single static binary (no Python dependency)
• Compatible with pip command interface
• Can replace pip, pip-tools, and virtualenv
• Efficient dependency resolution

UV Python Package Manager Cheatsheet

razhangwei

CUDA Cheatsheet for DeepGEMM

CUDA Fundamentals

Term	Description
SM (Streaming Multiprocessor)	Computational unit in NVIDIA GPUs; Hopper has up to 132 SMs
Warp	Group of 32 threads that execute in lockstep
Thread Block	Group of threads that execute on the same SM and can synchronize
Grid	Collection of thread blocks that execute the same kernel

razhangwei

1. Create Dockerfile via UI and build.

FROM codercom/code-server:latest

USER root

# Install system dependencies
RUN apt-get update && apt-get install -y \
    python3-pip \
    python3-venv \

razhangwei

This is an excellent LangGraph cheat sheet! It provides a comprehensive and practical overview of LangGraph's key features and usage. Here's a slightly refined version with added clarity and formatting for better readability:

---

LangGraph Cheat Sheet

1. Import Necessary Modules

from typing import TypedDict, Dict, Any, Callable
from langchain_core.runnables import Runnable

razhangwei

TypeScript Cheatsheet

#typescript #cheatsheet

A strongly-typed superset of JavaScript. TypeScript adds static typing, modern JS features, and great tooling for large-scale applications.

---

1. Basic Types

let name: string = "John";