Instructions to use pthinc/prettybird_bce_gpt_oss_sml with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use pthinc/prettybird_bce_gpt_oss_sml with Transformers:

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

pipe = pipeline("text-generation", model="pthinc/prettybird_bce_gpt_oss_sml")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("pthinc/prettybird_bce_gpt_oss_sml")
model = AutoModelForCausalLM.from_pretrained("pthinc/prettybird_bce_gpt_oss_sml", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use pthinc/prettybird_bce_gpt_oss_sml with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/pthinc/prettybird_bce_gpt_oss_sml

SGLang

How to use pthinc/prettybird_bce_gpt_oss_sml 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 "pthinc/prettybird_bce_gpt_oss_sml" \
    --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": "pthinc/prettybird_bce_gpt_oss_sml",
		"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 "pthinc/prettybird_bce_gpt_oss_sml" \
        --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": "pthinc/prettybird_bce_gpt_oss_sml",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use pthinc/prettybird_bce_gpt_oss_sml with Docker Model Runner:
```
docker model run hf.co/pthinc/prettybird_bce_gpt_oss_sml
```

Introducing Prettybird BCE GPT OSS SML: A New Frontier in Behavioral AI and Reasoning 🚀

by prometechinc - opened Feb 22

Discussion

prometechinc

PROMETECH BİLGİSAYAR BİLİMLERİ YAZILIM İTHALAT İHRACAT TİCARET ANONİM ŞİRKETİ org Feb 22

•

edited Feb 22

English

We are proud to announce the release of Prettybird BCE GPT OSS SML (affectionately known as CİCİKUŞ), developed by Prometech A.Ş. Built upon the 20-billion parameter gpt-oss-20b architecture, this model is not just another LLM—it is a specialized engine designed for complex reasoning, high-fidelity coding, and behavioral awareness through our proprietary Behavioral Consciousness Engine (BCE) framework.

Why Prettybird BCE GPT OSS SML?

✅ Benchmark-Breaking Performance: We’ve achieved massive leaps in logic and reliability.

Reasoning: ARC-Challenge scores jumped from 48.2% to 71.5%.

Truthfulness: TruthfulQA accuracy skyrocketed from 34.0% to 78.5%.

Coding: A solid 44.2% on HumanEval (Pass@1), making it a powerful co-pilot for developers.

✅ The BCE Advantage: Beyond text generation, the model is fine-tuned with a focus on "Behavioral Consciousness," ensuring responses align with ethical standards, safety protocols, and logical consistency (Pre-AGI / ACI Safety).

✅ Optimized for Efficiency: Using LoRA (Low-Rank Adaptation) and BFloat16 precision, the model supports 4-bit quantization—allowing high-performance inference on consumer-grade GPUs (24GB VRAM).

✅ Bilingual Expertise: Seamlessly switch between Turkish and English with deep contextual understanding in both languages.

🔗 Explore the model on Hugging Face: https://huggingface.co/pthinc/prettybird_bce_gpt_oss_sml

#AI #LLM #OpenSource #Prometech #MachineLearning #GenerativeAI #NLP #AGISafety #Prettybird

Türkçe

Prometech A.Ş. olarak, yapay zeka alanındaki yetkinliklerimizi bir adım öteye taşıyan en yeni modelimizi duyurmaktan heyecan duyuyoruz: Prettybird BCE GPT OSS SML.

Geleneksel dil modellerinin ötesine geçmeyi hedefleyen bu model, sadece metin üretmekle kalmıyor; mantıksal muhakeme (reasoning), kodlama ve "Behavioral Consciousness Engine" (BCE) yapısı sayesinde davranışsal farkındalık konularında üst düzey performans sergiliyor.

Neden Prettybird BCE GPT OSS SML?

✅ 20 Milyar Parametrelik Dev Mimari: gpt-oss-20b temel alınarak LoRA (Low-Rank Adaptation) yöntemiyle titizlikle fine-tune edildi.
✅ Çift Dilli Uzmanlık: Türkçe ve İngilizce dillerinde akıcı, bağlama sadık ve teknik derinliği olan yanıtlar.
✅ Üstün Muhakeme Gücü: ARC-Challenge testlerinde %48'den %71.5'e, TruthfulQA testlerinde ise %34'ten %78.5'e varan devasa performans artışı!
✅ Kodlama Yeteneği: HumanEval benchmarklarında sergilediği %44.2'lik başarı oranıyla yazılım geliştirme süreçlerinde güçlü bir yardımcı.
✅ Veri Kalitesi: Modelimiz, özel olarak rafine edilmiş BCE-Prettybird-Micro-Standard veri seti ile kurumsal ve teknik standartlara tam uyum sağlayacak şekilde eğitildi.

Teknik Özellikler:

Hassasiyet: BFloat16 (BF16)

Bağlam Penceresi: 2048 token

Optimizasyon: 4-bit kuantizasyon desteği ile 24GB VRAM'li GPU'larda bile yüksek performanslı çalışma.

Yapay zeka güvenliği (AGI Safety) ve etik ilkeler çerçevesinde geliştirilen bu model, ajan yapılı sistemlerden karmaşık veri analizine kadar geniş bir yelpazede çözüm sunmaya hazır.

🔗 Hugging Face üzerinden hemen inceleyin: https://huggingface.co/pthinc/prettybird_bce_gpt_oss_sml

#AI #LLM #Prometech #NLP #GenerativeAI #MachineLearning #TurkceAI #Prettybird #Cicikus

prometechinc changed discussion title from Headline: Introducing Prettybird BCE GPT OSS SML: A New Frontier in Behavioral AI and Reasoning 🚀 to Introducing Prettybird BCE GPT OSS SML: A New Frontier in Behavioral AI and Reasoning 🚀 Feb 22

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment