DeepSeek-Coder-v2: The BEST Opensource Coding LLM! (Beats GPT-4o and Claude 3.5 Sonnet)
ฝัง
- เผยแพร่เมื่อ 3 ต.ค. 2024
- DeepSeek AI has been consistently innovating, and their new DeepSeek Coder v2 is a game-changer. This model has outperformed some of the biggest names in the industry, like GPT-4 Turbo and Claude 3.5 Sonnet. In this video, we'll explore why DeepSeek Coder v2 is leading the pack and how it can revolutionize your coding experience.
[🔗 My Links]:
🔥 Become a Patron (Private Discord): / worldofai
☕ To help and Support me, Buy a Coffee or Donate to Support the Channel: ko-fi.com/worl... - It would mean a lot if you did! Thank you so much, guys! Love yall
🧠 Follow me on Twitter: / intheworldofai
📅 Book a 1-On-1 Consulting Call With Me: calendly.com/w...
📖 Want to Hire Me For AI Projects? Fill Out This Form: td730kenue7.ty...
🚨 Subscribe To My Second Channel: @WorldzofCrypto
Sponsor a Video or Do a Demo of Your Product, Contact me: intheworldzofai@gmail.com
[Must Watch]:
Aider + Llama 3.1: Develop a Full-stack App Without Writing ANY Code!: • Aider + Llama 3.1: Dev...
Aider + Claude 3.5: Develop a Full-stack App Without Writing ANY Code!: • Aider + Claude 3.5: De...
Claude Enginner: Powerful AI Coding Agent Framework!: • Claude Enginner: Power...
[Link's Used]:
Github Repo: github.com/dee...
Deepseek Website: www.deepseek.com/
Chat Demo: chat.deepseek....
Docs: platform.deeps...
LM Studio Tutorial: • LM Studio: Easiest Way...
Research Paper: arxiv.org/pdf/...
Model Card: huggingface.co...
Video Content
**Overview of DeepSeek Coder v2**: Learn about the latest updates, including function calling, chat completion, and the new API.
**BigCode Bench Leaderboard**: Discover how DeepSeek Coder v2 ranks at the top in practical and challenging programming tasks, competing closely with GPT-4 Turbo and Claude 3.5 Sonnet.
**Comparison with Aider Framework**: See how DeepSeek Coder v2 stands out in the pair programming evaluation, slightly behind Sonnet 3.5 but ahead of GPT-4.
**Breaking Barriers**: Understand how DeepSeek Coder v2 is an open-source Mixture-of-Experts (MoE) code language model, supporting up to 338 programming languages with a 128k context window.
**Pre-training and Performance**: Learn about the additional 6 trillion tokens used for pre-training from an intermediate checkpoint of DeepSeek V2.
If you enjoyed this video, don't forget to like, subscribe, and share! Your support helps us bring you more cutting-edge content on the latest in AI and coding.
Additional Tags and Keywords
DeepSeek Coder v2, Best Open-Source Coding LLM, GPT-4 Turbo, Claude 3.5 Sonnet, BigCode Bench, Aider Framework, AI Coding Models, Mixture-of-Experts, Open-Source AI, Programming Languages, AI Innovation
Hashtags
#DeepSeekCoderV2 #opensourceai #CodingLLM #gpt4turbo #Claude3Sonnet #aiassistants #programming #techinnovation
Want to HIRE us to implement AI into your Business or Workflow? Fill out this work form: td730kenue7.typeform.com/to/WndMD5l7
💗 Thank you so much for watching guys! I would highly appreciate it if you subscribe (turn on notifcation bell), like, and comment what else you want to see!
📆 Book a 1-On-1 Consulting Call WIth Me: calendly.com/worldzofai/ai-consulting-call-1
🔥 Become a Patron (Private Discord): patreon.com/WorldofAi
🧠 Follow me on Twitter: twitter.com/intheworldofai
Love y'all and have an amazing day fellas. Thank you so much guys! Love yall!
Thank you so much for this bro! I think this is the best video on this so far. Keep it up!
Thanks! Will do!
love the content 🖤
[Must Watch]:
Aider + Llama 3.1: Develop a Full-stack App Without Writing ANY Code!: th-cam.com/video/eyyMwSa7OSg/w-d-xo.htmlsi=H80wHMZ0-22tOVNO
Aider + Claude 3.5: Develop a Full-stack App Without Writing ANY Code!: th-cam.com/video/dzOWn8TI738/w-d-xo.htmlsi=kbLr6jYGp8n8zPQ7
Claude Enginner: Powerful AI Coding Agent Framework!: th-cam.com/video/-VW8dJbEDsk/w-d-xo.htmlsi=ewqXsj2QzSqIB80X
The problem is its knowledge cutoff in late 2021
I was gonna say this as well, it is great to have but it really defeats the purpose if it's not up to date.
If you download it locally you can give it internet access and also it to search for information beyond the cutoff
@@TheGuillotineKingthe difference between web scraping and a predefined knowledge base is massive. This is no where near Claude
Noti gang!
RagFlow: Ultimate RAG Engine - Semantic Search, Embeddings, Vector Search + Supports Graph!: th-cam.com/video/ApA-7G7FGRc/w-d-xo.html
Aider + DeepSeek-Coder-V2: Develop a Full-stack App Without Writing ANY Code /w Best Coding LLM! - th-cam.com/video/xcN3Lw0zbp4/w-d-xo.html
wow
OpenSearch AI: Opensource AI-powered Search Engine! PERPLEXITY Alternative!: th-cam.com/video/xQ53VK07bGg/w-d-xo.html
Make video about hyperwrite
I actually have before! Multiple times :)
What I want to know is how well it remembers instructions. Claude 3.5 is actually horrible. For more complex projects Claude is extremely frustrating. Projects is a joke. Claude cannot reference different chats in a project so you have to start over from scratch each time your chat limit is reached. It regularly forgets about files in the content database and the custom instructions are meaningless. I am new to coding so I ask it to not give snippets, only complete code. I have to ask every time I submit a message because it forgets. The error rate in code is so high I spend most of my messages on iterations. I even had it tell me that it did not really provide a fix but just provided the same broken code it did previously. It apologized.. I am working on a JUCE/C++ project. I will not be renewing my subscription.
Oh interesting. I never really encountered that problem when using sonnet. What you plan on using now ?
Yeah it depends on complexity of projects. For common requested projects like snake game or crud apps they work fine. LLM is still based on what they are trained on
for my Claude is very good, in some cases better even that Chatgpt and iam talking of the free version.
You have to ask her very well to get good questions, sometimes i even write examples of what i want, also give her the pseudocode help a lot.
Claude? I have had the same experience. The web version is MUCH better quality than the API version. However, the web version constantly hits the limit. It's unusable. 😢
Thats with every ai lol it has a limit
Deepseek is not that great. It makes lots of mistakes. It has to be corrected often and doesn’t remember instructions. It literally started generating the word “Context” repeatedly in the middle of a project. Switching back to Llama…