Open Devin

 幾星期前， AI 界的焦點曾落於

🌟 “Devin AI” ✨ 之上 ，它被喻為首個最有能力，貼近人類 Software Engineer 多元能力的 AI 工具， 大概就係講出開發的主題和核心需求，然後 AI agent 就會分工進行不同崗位（包括設計功能、環境設置、編程、測試及修正、撰寫文檔）的工作，自行合作去完成編程任務。
誰知過咗幾日後，GitHub 上就出現了 ⚡️Open Devin 🤖（免費開源版本）， 用類似的 UX框架，配合不同自定義的LLM ，也可以進行編程任務了。
如果有興趣試玩用的朋友留意， 由於 Open Devin 仍然好新，所以， 代碼仍然不停被社群更新，安裝的方法也在變化中，令好多網上的教學都 out date 了。 如果你用 Linux 或者 Mac 會容易一些， 如果用 Windows 下安裝，暫時仍十分複雜 😅 （我都係撞下撞下咁裝到🤷‍♂️，主要涉及幾方面的問題要解決：
1/ UTF-8 問題 （讀寫logs檔會出錯)
2/ frontend 與 backend 的 Port 不懂自動相連，要按個別的網絡設置環境去調節
3/ 有些預設的 dependency 無法安裝，因不支援 Windows 環境 （e.g: vuloop)
4/ 需要 docker , WSL , Conda 設置
5/ 用’make’ 方式來 build , 和 run ， 所以在 windows 下要用 Chocolatey 去安裝。）
裝完之後， 設置設工作目錄和所用的 openAI API Key ， 透過 browser 就可以用到。我簡單地試咗， 叫佢用 python 寫個「人 vs 電腦的Tic Tac Toe 遊戲」 ， 佢大概用了HK$2-3 蚊的 token 成本 gpt-4-turbo-preview 就寫了出來， 執行起來， 冇 bug 🙆🏻‍♂️ 這的確有點驚喜， 至於複雜的開發任務，我等 open devin 再更新到更穩定的版本時會再試。


Comments

Post a Comment

Popular posts from this blog

state government roles website

South Australia (SA) https://iworkfor.sa.gov.au New South Wales (NSW): The I Work for NSW portal advertises job vacancies across various NSW government agencies. https://www.randstad.com.au/jobs/new-south-wales/ Victoria (VIC): The Careers.Vic website provides listings for positions within the Victorian Public Service. https://careers.vic.gov.au/ Queensland (QLD): The Smart Jobs and Careers platform showcases job opportunities in the Queensland Government. https://smartjobs.qld.gov.au/jobtools/jncustomsearch.jobsearch?in_organid=14904 Western Australia (WA): The WA Government Jobs Board features vacancies across Western Australian government departments. https://search.jobs.wa.gov.au/page.php?pageID=215 Tasmania (TAS): The Jobs Tasmania site lists employment opportunities within the Tasmanian State Service. https://www.jobs.tas.gov.au/ Australian Capital Territory (ACT): The Jobs ACT portal provides information on positions available in the ACT Government. https://www.jobs.act.gov.au/ ...
Read more

To enhance embedding in your Retrieval-Augmented Generation (RAG) application

 To enhance embedding in your Retrieval-Augmented Generation (RAG) application based on the "Python RAG Tutorial.txt" notes, follow these steps in simple English: 1. **Use High-Quality Embeddings**:    - Choose high-quality embeddings to ensure accurate matching between your queries and the relevant data chunks. Consider using services like OpenAI or AWS Bedrock, as they provide reliable embeddings. 2. **Consistent Embedding Function**:    - Use the same embedding function for both creating the database and querying it. This ensures consistency and better performance. 3. **Manage Large Documents**:    - Split large documents into smaller chunks. Use tools like Langchain's recursive text splitter. Smaller chunks improve indexing and retrieval accuracy. 4. **Update the Vector Database**:    - Add a unique ID to each data chunk. Use the file path, page number, and chunk number to create these IDs. This helps in updating the database without dupl...
Read more

SQL Tutorials 10 hours

Image
Certainly! Here's a tutorial in the form of a flash card table for MySQL SQL commands: SQL Command & Explanation SQL Example SELECT - Retrieves data from one or more tables. SELECT first_name, last_name FROM employees; FROM - Specifies the table to retrieve data from. SELECT first_name FROM employees; WHERE - Filters records based on a condition. SELECT * FROM employees WHERE age > 30; AND , OR - Combine conditions in a WHERE clause. SELECT * FROM employees WHERE age > 30 AND department = 'HR'; ORDER BY - Sorts the result set. SELECT * FROM employees ORDER BY last_name ASC; INSERT INTO - Adds new records into a table. INSERT INTO employees (first_name, last_name) VALUES ('John', 'Doe'); UPDATE - Modifies existing records in a table. UPDATE employees SET age = 31 WHERE first_name = 'John' AND last_name = 'Doe'; DELETE - Removes records from a table. DELETE FROM employees WHERE last_name = 'Doe'; CREATE DATABASE - Cre...
Read more