Konuyu Oyla:
  • Derecelendirme: 0/5 - 0 oy
  • 1
  • 2
  • 3
  • 4
  • 5
Yapay zekaya bakış açınız nedir?
#21
torvalds

HMPMNIPbUAA8g-W?format=jpg&name=medium
Cevapla
#22
İzlemenizi öneririm.

https://www.youtube.com/watch?v=2VVV4fAcXrQ

Bu arada bende gemini claude gpt ücretsiz karışık kullanıyorum.

Ayrıca evde local llm ekosistemimi kurdum. Gayet başarılı kodlama ve bunun haricinde genelde angarya işleri kolaylaştırıyor..
PostgreSQL - Linux - Delphi, Poliüretan
WWW
Cevapla
#23
Bir şekilde en az GLM-5.2 seviyesinde model çalıştırılacak fiziksel bir sistem elde etmeliyiz. Ama tek başımıza ama bir grup olarak.
Cevapla
#24
Profesyonel programcı değilim ama işim gereği kurum içinde çalışan uygulamalar yazıyorum. genelde linux üzerinde python yada php ile çalışıyorum. son 1 yıldır çoğu uygulamayı vibe coding ile geliştiriyorum. YZ olarak antigravity ve cursor üyeliğim mevcut. agent modunda geliştiriyorum. Açıkçası %100 projeyi kendi başına yapamıyor ama adım adım parça parça ilerleyip sorunları çözüyor. Birde frontend tarafıdna gerçekten başarılı sonuçlar veriyor. Delphi IDE içerisinde benzer bir agent modu gelir mi bilmiyorum ama bence sektör bu tarafa kaymaya başladı.
Cevapla
#25
Apple 768 GB RAM li Mac Studio dan sonra 1,5 TB RAM li Mac Studio sunacak gibi.
Bu açık frontier LLM lerin yerel olarak çalıştırılabilmesi demek. Tabi hız ne olur göreceğiz.

[img]data:image/png;base64,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[/img]

Takipteyiz.
Cevapla
#26
(05-07-2026, Saat: 15:16)engerex Adlı Kullanıcıdan Alıntı: Bir şekilde en az GLM-5.2 seviyesinde model çalıştırılacak fiziksel bir sistem elde etmeliyiz. Ama tek başımıza ama bir grup olarak.

AI araçları yapılandırma adına bilginiz tecrübeniz nedir bilmiyorum. Fakat kişilerin altına giremeyeceği kadar mali yük anlamına geliyor. Açıkçası böyle işlerde modelin ağırlığı belirleyici oluyor. Yani parametre sayısı aslında daha çok dosya boyutu. Buna karşın kalite de tek başına parametre olmasa da paralelinde kaliteyi belirliyor.

Uzun sözün özü şu anda GLM 5.2 şurada https://huggingface.co/zai-org/GLM-5.2 belirtildiğine göre 753B parametre demek buda yaklaşık tahmini bir o kadar dosya boyutu anlamı geliyor. en az 500GB deseniz. Bunu çalıştıracak VRAM hesabını siz yapın. Üzerine birde RAM CPU DISK derken ev parasını dökmen gerekebilir.

Ben yakın zamanda 5090+64+2TB disk kombinasyonlu soğutması psu kasa fan vs derken 300k civarına rakamlar çıkıyor.

32VRam oluyor.

https://openzeka.com/urun/nvidia-h200-nv...-0040-000/?
burada h200 kart var. 141GB VRam bundan 5-6 tane kullanman gerekir.

Ayrıca bir süre sonra bu cihazların yıpranma paylarını da hesap edin.
PostgreSQL - Linux - Delphi, Poliüretan
WWW
Cevapla
#27
(06-07-2026, Saat: 14:01)3ddark Adlı Kullanıcıdan Alıntı:
(05-07-2026, Saat: 15:16)engerex Adlı Kullanıcıdan Alıntı: Bir şekilde en az GLM-5.2 seviyesinde model çalıştırılacak fiziksel bir sistem elde etmeliyiz. Ama tek başımıza ama bir grup olarak.

AI araçları yapılandırma adına bilginiz tecrübeniz nedir bilmiyorum. Fakat kişilerin altına giremeyeceği kadar mali yük anlamına geliyor. Açıkçası böyle işlerde modelin ağırlığı belirleyici oluyor. Yani parametre sayısı aslında daha çok dosya boyutu. Buna karşın kalite de tek başına parametre olmasa da paralelinde kaliteyi belirliyor.

Uzun sözün özü şu anda GLM 5.2 şurada https://huggingface.co/zai-org/GLM-5.2 belirtildiğine göre 753B parametre demek buda yaklaşık tahmini bir o kadar dosya boyutu anlamı geliyor. en az 500GB deseniz. Bunu çalıştıracak VRAM hesabını siz yapın. Üzerine birde RAM CPU DISK derken ev parasını dökmen gerekebilir.

Ben yakın zamanda 5090+64+2TB disk kombinasyonlu soğutması psu kasa fan vs derken 300k civarına rakamlar çıkıyor.

32VRam oluyor.

https://openzeka.com/urun/nvidia-h200-nv...-0040-000/?
burada h200 kart var. 141GB VRam bundan 5-6 tane kullanman gerekir.

Ayrıca bir süre sonra bu cihazların yıpranma paylarını da hesap edin.


Evet. Haklısınız, bu işin maliyeti çok yüksek. Yerel çalıştırmaktansa abonelikler çok çok daha uygun maliyete geliyor.
 Benim senaryom maliyet için değil güçlü modellerin erişilememesine karşı önlem. Amerika modellerin halka açılmadan önce kendi deneyip onaya istinaden halka açılmasını istiyor. Fable 5 modelinde bunu gördük yaşadık. Modeli o kadar çok kısıtlamışlar ki selam versen güvenlik nedeniyle Opus modeline geçiyor. Zaten haftalık krediyi 1 günde rahat dolduruyorsunuz. ChatGPT 5.6 da da aynı izin mevzusu uygulanıyor.
 Apple 768GB RAMli Mac Studio hazırlıyor haberleri düştü. evet maliyetli ama dediğim gibi maliyet ilk sırada değil.
 Ayrıca sadece VRAM e bakmamak gerekli. 5070ti+64+2TB li sistemde normalde 25B-35B VRAM e zor sığarken RAM sayesinde 70B lik model 35-40 tokens/sec ile çalışabiliyor.
Cevapla


Konu ile Alakalı Benzer Konular
Konular Yazar Yorumlar Okunma Son Yorum
  Yapay zeka icat oldu foruma uğrayanlar azaldı tuna 5 356 05-07-2026, Saat: 01:51
Son Yorum: 10.Köy
  Delphi ile Yapay Zeka Webinarı AhmetNuri 1 750 27-11-2024, Saat: 09:42
Son Yorum: yhackup
  Algoritma Nedir? Nasıl Çalışır? (Güzel bir örnek) ozcanm 3 2.187 23-10-2022, Saat: 12:27
Son Yorum: Hayati
  Yapay zeka sohbet botları ? Mr.Developer 22 16.743 07-09-2020, Saat: 11:06
Son Yorum: bydelphi
  Bilin bakalım bu nedir :) yhackup 6 6.542 22-04-2017, Saat: 10:37
Son Yorum: yhackup



Konuyu Okuyanlar: 1 Ziyaretçi