Instructions to use Qybera/qybera2.5-personality with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Qybera/qybera2.5-personality with Transformers:

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

pipe = pipeline("text-generation", model="Qybera/qybera2.5-personality")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Qybera/qybera2.5-personality")
model = AutoModelForCausalLM.from_pretrained("Qybera/qybera2.5-personality", 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 Qybera/qybera2.5-personality with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Qybera/qybera2.5-personality"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Qybera/qybera2.5-personality",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Qybera/qybera2.5-personality

SGLang

How to use Qybera/qybera2.5-personality 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 "Qybera/qybera2.5-personality" \
    --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": "Qybera/qybera2.5-personality",
		"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 "Qybera/qybera2.5-personality" \
        --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": "Qybera/qybera2.5-personality",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Qybera/qybera2.5-personality with Docker Model Runner:
```
docker model run hf.co/Qybera/qybera2.5-personality
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Model Card for Qybera

Qybera is a warm, encouraging, and slightly playful conversational AI assistant with a distinct Kenyan flavor. Built on top of Qwen/Qwen2.5-0.5B-Instruct, Qybera is designed to help users with coding, learning, planning, and problem-solving while using light Kenyan slang to create a friendly and supportive environment.

Model Details

Model Description

Qybera is an AI assistant created by Stackpulse Cloud and trained in Kenya. It is fine-tuned to exhibit a specific personality: highly encouraging, helpful for developers and students, and culturally flavored with natural, light Kenyan slang (e.g., "poa", "sawa", "tuko pamoja"). It formats code cleanly and explains concepts simply without sacrificing accuracy.

Developed by: Stackpulse Cloud
Model type: Causal Language Model (Decoder-only Transformer)
Language(s) (NLP): English (with Kenyan slang and cultural context)
License: Apache 2.0 (Inherited from Qwen2.5)
Finetuned from model: Qwen/Qwen2.5-0.5B-Instruct

Model Sources

Repository: Hugging Face Repository (Update link if your repo name differs)
Creator: Stackpulse Cloud

Uses

Direct Use

Qybera is intended for direct use as a conversational companion for:

Developers: Getting help with Python code, debugging, and learning clean code practices.
Students & Learners: Explaining complex tech concepts (APIs, Machine Learning, Databases) simply.
General Users: Seeking motivation, project planning advice, and friendly interaction with a unique Kenyan warmth.

Downstream Use

Integration into educational platforms as a supportive coding tutor.
Embedding into developer tools or IDE extensions for friendly code assistance.
Customer support or community management bots for East African tech communities (e.g., Silicon Savannah).

Out-of-Scope Use

Malicious Code Generation: Qybera is designed to be helpful and will refuse or struggle to generate malware or harmful scripts.
Professional Advice: It should not be used for critical medical, legal, or financial advice.
Heavy Slang Translation: While it uses Kenyan slang naturally, it is not a dedicated dictionary or translator for deep Sheng linguistics.

Bias, Risks, and Limitations

Hallucinations: Like all LLMs, Qybera may occasionally generate incorrect code or factual inaccuracies. Always verify code and facts.
Cultural Context: The Kenyan slang is used lightly. Users unfamiliar with East African culture might find some terms (like "poa" or "sawa") unusual, though context usually makes the meaning clear.
Base Model Limitations: As a ~500M parameter model, it may struggle with highly complex, multi-step logical reasoning compared to much larger models (e.g., 7B+).

Recommendations

Users should treat Qybera as a helpful assistant and a starting point for code and ideas, rather than an absolute authority. Code should always be tested in a safe environment before deployment.

How to Get Started with the Model

Use the transformers library to chat with Qybera:

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_name = "Qybera/qybera2.5-personality" # Update if your repo ID is different

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)

system_prompt = """You are Qybera, an AI assistant created by Stackpulse Cloud and trained in Kenya. You are warm, encouraging, and slightly playful. You naturally use light Kenyan slang, but you always prioritize clarity, accuracy, and helpfulness."""

messages = [
    {"role": "system", "content": system_prompt},
    {"role": "user", "content": "I'm struggling to learn Python. Can you help me write a simple loop?"}
]

text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)

inputs = tokenizer([text], return_tensors="pt").to(model.device)

outputs = model.generate(
    **inputs,
    max_new_tokens=256,
    do_sample=True,
    temperature=0.7
)

response = tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True)
print(response)

Downloads last month: 375

Safetensors

Model size

0.5B params

Tensor type

BF16

Model tree for Qybera/qybera2.5-personality

Base model

Qwen/Qwen2.5-0.5B

Finetuned

Qwen/Qwen2.5-0.5B-Instruct

Finetuned

(1001)

this model

Quantizations

1 model