BrtGPT-1-Conversation

Introduction

Our most powerful model, trained on conversational data, allows you to chat.

It was trained on 60,000 chat data containing 3-8 messages.

Our model's context length is 1024 tokens (approximately 750 tokens, approximately 7.5 paragraphs). Our model was trained on A100 GPUs.

Model is generates long and random output.

Model is only for test!!!

USE

Please start conversation like this: "Heyyyy what’s up?! 😊" if model output has: "ŁĺĤ" or anything like this, its means "emoji". CODE:

from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
import torch
from threading import Thread

# === MODEL and TOKENIZER ===
model_id = "Bertug1911/BrtGPT-1"

tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True)
model.eval().to("cuda" if torch.cuda.is_available() else "cpu")

# === CLEAN ===
def clean(text):
    return text.replace(" ", "").replace("Ġ", " ").replace("Ċ", "\n")

# === PAST ===
chat_history = []

# === LOOP: QA ===
while True:
    try:
        # Soru al
        question = input("\n🧠 Soru: ")

        if question.strip().lower() in ["q", "quit", "exit"]:
            print("Çıkılıyor...")
            break

        # Mesaj geçmişine kullanıcı mesajını ekle
        chat_history.append({"role": "user", "content": question})

        # Mesajı chat formatında hazırla (önceki geçmiş + yeni mesaj)
        inputs = tokenizer.apply_chat_template(
            chat_history,
            add_generation_prompt=True,
            return_tensors="pt"
        ).to(model.device)

        # Streamer tanımla
        streamer = TextIteratorStreamer(
            tokenizer,
            skip_prompt=True,
            skip_special_tokens=True
        )

        # Üretim thread'i
        def generate():
            model.generate(
                input_ids=inputs,
                streamer=streamer,
                max_new_tokens=256,
                do_sample=True,
                top_k=10,
                temperature=0.1,
            )

        thread = Thread(target=generate)
        thread.start()

        # Cevap tokenlarını temizleyerek yazdır
        print("🤖 Cevap:", end=" ", flush=True)
        full_answer = ""
        for token in streamer:
            cleaned = clean(token)
            full_answer += cleaned
            print(cleaned, end="", flush=True)

        # Cevabı da geçmişe assistant olarak ekle
        chat_history.append({"role": "assistant", "content": full_answer})

    except KeyboardInterrupt:
        print("\nManuel olarak durduruldu.")
        break

TRAINING DETAILS

Model is trained on 4x A100 for 30 minutes.

Model trained on 60.000~ 3-8 messages conversations.

Platform: RunPod

RISKS

There is no any risk but model can generates harmfull content.

USE AT YOUR OWN RISK!

CONTACT

[email protected] or [email protected]