Instructions to use aiprojecom/kumru-2b-flirt-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use aiprojecom/kumru-2b-flirt-lora with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("vngrs-ai/Kumru-2B")
model = PeftModel.from_pretrained(base_model, "aiprojecom/kumru-2b-flirt-lora")

Transformers

How to use aiprojecom/kumru-2b-flirt-lora with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="aiprojecom/kumru-2b-flirt-lora")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("aiprojecom/kumru-2b-flirt-lora", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use aiprojecom/kumru-2b-flirt-lora with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "aiprojecom/kumru-2b-flirt-lora"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "aiprojecom/kumru-2b-flirt-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/aiprojecom/kumru-2b-flirt-lora

SGLang

How to use aiprojecom/kumru-2b-flirt-lora with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "aiprojecom/kumru-2b-flirt-lora" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "aiprojecom/kumru-2b-flirt-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "aiprojecom/kumru-2b-flirt-lora" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "aiprojecom/kumru-2b-flirt-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Studio new

How to use aiprojecom/kumru-2b-flirt-lora with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for aiprojecom/kumru-2b-flirt-lora to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for aiprojecom/kumru-2b-flirt-lora to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for aiprojecom/kumru-2b-flirt-lora to start chatting

Load model with FastModel

pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="aiprojecom/kumru-2b-flirt-lora",
    max_seq_length=2048,
)

Docker Model Runner
How to use aiprojecom/kumru-2b-flirt-lora with Docker Model Runner:
```
docker model run hf.co/aiprojecom/kumru-2b-flirt-lora
```

Flört veri seti ile eğitilmiş Kumru 2B Instruct Lora

Model Detayları

Temel model: vngrs-ai/Kumru-2B
Eğitim yöntemi: Unsloth + LoRA (r=16)
Eğitim süresi: 2 epoch (~35 dk, T4 GPU)
Son eval loss: ~5.39
Veri seti: Flört uygulamalarından toplanmış Türkçe konuşmalar (~38.537 satır)
Adaptörü Oluşturan: @TlCARET

Bu model ne yapar / ne yapmaz?

Base modele göre cevap farklılıkları:

Argo ve emoji kullanımı arttı
Klasik sohbet başlangıçlarında cevap tutarlılığı iyileşti
Flörtöz sorularda alakalı cevaplar üretiyor
Flörtöz sohbet devamlılığı iyileşti

Eksiklikler:

Model parametresi göz önünde bulundurulunca verilen cevaplarda akıl yürütmesi noksan.
Epoch sayısı artsa ve QA'da tekrar sayısı artsa da loss benzer kalıyor bu veri setinde.
Base model çok fazla sözlük etkisinde kaldığı için alakasız yerlerde Bknz: diyebiliyor.

Not(lar):

Tamamen eğlence ve deney amaçlıdır. Profesyonel kullanıma uygun değildir.
Eğitim verilerine system prompt dahil edilmemiştir.

Kullanım

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
import torch

base_model = "vngrs-ai/Kumru-2B"
lora_path = "aiproje/kumru-2b-flirt-lora"   # kendi repo adını yaz

tokenizer = AutoTokenizer.from_pretrained(base_model)
model = AutoModelForCausalLM.from_pretrained(
    base_model,
    load_in_4bit=True,
    device_map="auto"
)
model = PeftModel.from_pretrained(model, lora_path)

# Örnek sohbet
prompt = "Nasılsın şekerim? 😏"
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
outputs = model.generate(
    **inputs,
    max_new_tokens=120,
    temperature=0.85,       # biraz daha yaratıcı
    top_p=0.92,
    repetition_penalty=1.2, # daha fazla tekrar 
    max_new_tokens=300,
    repetition_penalty=1.25,
    do_sample=True
)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Downloads last month: 2

Model tree for aiprojecom/kumru-2b-flirt-lora

Base model

vngrs-ai/Kumru-2B

Adapter

(7)

this model