Instructions to use Catniti/catrex-1.0 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Catniti/catrex-1.0 with Transformers:

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

pipe = pipeline("text-generation", model="Catniti/catrex-1.0")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Catniti/catrex-1.0")
model = AutoModelForCausalLM.from_pretrained("Catniti/catrex-1.0", 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

llama.cpp

How to use Catniti/catrex-1.0 with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Catniti/catrex-1.0:F16
# Run inference directly in the terminal:
llama cli -hf Catniti/catrex-1.0:F16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Catniti/catrex-1.0:F16
# Run inference directly in the terminal:
llama cli -hf Catniti/catrex-1.0:F16

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf Catniti/catrex-1.0:F16
# Run inference directly in the terminal:
./llama-cli -hf Catniti/catrex-1.0:F16

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf Catniti/catrex-1.0:F16
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Catniti/catrex-1.0:F16

Use Docker

docker model run hf.co/Catniti/catrex-1.0:F16

LM Studio
Jan

vLLM

How to use Catniti/catrex-1.0 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Catniti/catrex-1.0"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Catniti/catrex-1.0",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Catniti/catrex-1.0:F16

SGLang

How to use Catniti/catrex-1.0 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 "Catniti/catrex-1.0" \
    --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": "Catniti/catrex-1.0",
		"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 "Catniti/catrex-1.0" \
        --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": "Catniti/catrex-1.0",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use Catniti/catrex-1.0 with Ollama:
```
ollama run hf.co/Catniti/catrex-1.0:F16
```

Unsloth Studio

How to use Catniti/catrex-1.0 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 Catniti/catrex-1.0 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 Catniti/catrex-1.0 to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Catniti/catrex-1.0 to start chatting

Docker Model Runner
How to use Catniti/catrex-1.0 with Docker Model Runner:
```
docker model run hf.co/Catniti/catrex-1.0:F16
```

Lemonade

How to use Catniti/catrex-1.0 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Catniti/catrex-1.0:F16

Run and chat with the model

lemonade run user.catrex-1.0-F16

List all available models

lemonade list

Atomic Chat

🐱 Catrex 1.0

Двуязычная (RU/EN) чат-модель на 65M параметров, обученная с нуля: случайная инициализация, ни одного предобученного веса.

Результаты


val loss	5.6104
perplexity	273.3
Параметров	64.82M
Контекст	512 токенов
Шагов обучения	120
Обучение	1× T4

Запуск в LM Studio

Скачай catrex-1.0-f16.gguf
Положи в ~/.lmstudio/models/Catniti/catrex-1.0/
Загрузи в LM Studio — chat template уже внутри файла

Запуск через llama.cpp

llama-cli -hf Catniti/catrex-1.0:f16 -p "Привет!"

Python (transformers)

from transformers import AutoModelForCausalLM, AutoTokenizer

tok = AutoTokenizer.from_pretrained("Catniti/catrex-1.0")
model = AutoModelForCausalLM.from_pretrained("Catniti/catrex-1.0")

msgs = [{"role": "user", "content": "Привет!"}]
ids = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt")
out = model.generate(ids, max_new_tokens=100, temperature=0.8, do_sample=True)
print(tok.decode(out[0][ids.shape[1]:], skip_special_tokens=True))

Формат промпта

ChatML:

<|im_start|>user
Привет!<|im_end|>
<|im_start|>assistant

Честно об ограничениях

Модель на 65M параметров, обученная несколько часов на одной T4. Это примерно в 1000 раз меньше вычислений, чем у GPT-3.5.

Чего ожидать:

связная речь, простые диалоги, короткие истории
английский заметно сильнее русского: токенизатор SmolLM2 тратит на русский вдвое больше токенов

Чего не ожидать: фактической точности, рассуждений, длинного контекста, кода. Модель может уверенно выдумывать.

Архитектура Llama выбрана намеренно — только так GGUF открывается в LM Studio, Ollama и llama.cpp. Веса при этом полностью свои.

Downloads last month: 268

Safetensors

Model size

64.8M params

Tensor type

F32

Catniti
/

catrex-1.0