Qwen 2.5 Takes Down GPT4o in Coding Showdown!

Поделиться
HTML-код
  • Опубликовано: 24 янв 2025

Комментарии • 15

  • @UCs6ktlulE5BEeb3vBBOu6DQ
    @UCs6ktlulE5BEeb3vBBOu6DQ 2 месяца назад +13

    It beats this and that on benchmarks. In real life, GPT o4 will have you playing a somewhat working Tetris game in 10-15 minutes while QWEN will have you troubleshoot to death with pieces go left to right. I love QWEN. I want to use QWEN but benchmarks give false hopes.

    • @themax2go
      @themax2go 2 месяца назад

      qwen is run by chinese ai company (alibaba group actually), and chinese ai companies (at least for now) fine-tune to look good on "leaderboards"

    • @globalprosperity123
      @globalprosperity123 Месяц назад

      ​@themax2go and openai is American so transparent non-profit. 😂

  • @modoulaminceesay9211
    @modoulaminceesay9211 2 месяца назад +6

    I think 24gb will be able to handle this locally ?

  • @trendguru586
    @trendguru586 2 месяца назад +1

    There are some possibilities to extend standard context window of qwen 2.5 coder from 32k to 128k by using YaRN.
    Do you have any experience with that?

  • @AK-ox3mv
    @AK-ox3mv 2 месяца назад

    Is there anyway to run it from huggingface chat models to vscode extentions like Continue for code autocompletion?

  • @wiseguy0106
    @wiseguy0106 2 месяца назад

    Yeah, right :))))

  • @LuisBorges0
    @LuisBorges0 2 месяца назад

    Yeah... but o1-mini is unstopable

  • @RealEstate3D
    @RealEstate3D 2 месяца назад +3

    Here another clickbait title. I give up here as you cost a lot of time.

  • @madness1621
    @madness1621 2 месяца назад

    That's not true😂

  • @fathi240
    @fathi240 2 месяца назад +1

    Can you add Arabic to your videos? I don't understand English well. Thank you.❤❤❤🎉🎉

  • @themax2go
    @themax2go 2 месяца назад

    yeah you messed this one up (seems to be a trend with you) @MervinPraison... i'm going to unsub from your channel (for now)