myylogic

myylogic / cevahir-ai

Public

Full-stack open-source AI engine for building language models — tokenizer training, transformer architecture, cognitive reasoning and chat pipeline.

ai- ai-engine bpe cognitive-ai- conversational-ai
12
1
89% credibility
Found Mar 14, 2026 at 12 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Cevahir AI is a complete open-source toolkit for everyday people to create, train, and chat with custom language models and assistants.

How It Works

1
🔍 Discover Cevahir AI

You hear about this free toolkit that lets anyone build a smart AI helper for chatting in Turkish or any language.

2
📥 Get the ready pieces

Download the trained brain and learning examples shared openly so you can start right away without building from scratch.

3
📚 Teach it your info

Feed in your own stories, questions, and answers so it learns exactly what matters to you.

4
💭 Chat with your AI

Ask questions and get thoughtful replies that feel personal and smart, like talking to a helpful friend.

5
🌐 Share it online

Put your custom helper on the web so friends or family can chat with it too.

🎉 Your own AI is alive

Now you have a private smart assistant trained just for you, ready anytime with answers that truly understand.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 12 to 12 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is cevahir-ai?

Cevahir-ai is a Python full-stack open-source AI engine for building custom language models end-to-end, handling tokenizer training, transformer inference, cognitive reasoning like Tree of Thoughts and RAG, plus a chat pipeline with Flask API. It empowers devs to train Turkish-optimized (but language-agnostic) models on limited hardware, delivering CLI scripts for training/inference and REST endpoints for sessions, users, and chat. Download pre-trained weights and 680k-sample datasets to skip from-scratch work.

Why is it gaining traction?

Unlike fragmented frameworks needing glued-together repos, this packs everything—tokenizer BPE with Unicode fixes, model management, cognitive strategies, and production API with JWT/auth—in one full stack github project. Devs notice the optimized Turkish handling (İ/ı, morphology) extending to any language, plus cognitive perks like critics and tools boosting response quality without extra services. The manifesto hooks resource-strapped builders challenging giants.

Who should use this?

AI hobbyists training domain LLMs (e.g., legal Turkish docs) on consumer GPUs; full stack developers github users needing a chat backend with user sessions/memory; Turkish NLP teams (cevahir ailesi-inspired?) prototyping without cloud costs.

Verdict

Grab for custom LLM experiments—solid user API and training flow shine—but 11 stars and 0.9% credibility score signal early immaturity: README docs ok, no visible tests. Fork if full stack open source AI appeals; production needs polish.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.