Instructions to use cijov/cijov-lang-1B-1V with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use cijov/cijov-lang-1B-1V with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="cijov/cijov-lang-1B-1V")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("cijov/cijov-lang-1B-1V", dtype="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use cijov/cijov-lang-1B-1V with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "cijov/cijov-lang-1B-1V" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "cijov/cijov-lang-1B-1V", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/cijov/cijov-lang-1B-1V
- SGLang
How to use cijov/cijov-lang-1B-1V 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 "cijov/cijov-lang-1B-1V" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "cijov/cijov-lang-1B-1V", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'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 "cijov/cijov-lang-1B-1V" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "cijov/cijov-lang-1B-1V", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use cijov/cijov-lang-1B-1V with Docker Model Runner:
docker model run hf.co/cijov/cijov-lang-1B-1V
Cijov-lang-1B Release Package
This folder is a self-contained release layout for inference.
Model Identity
- Developed by: Alin Cijov
- Model name: Cijov Language - cijov-lang
- Text generation languages: English (EN), French (FR), Romanian (RO), Spanish (SP/ES)
- Tokenizer: cijov/cijov-lang-tokenizer
Structure
release/
model/ # drop model checkpoint files here
generate.py # text generation script
LICENSE # Cijov Apache 2.0 license
NOTICE # attribution notice
1) Model Files
The model weights are already included in release/model/ (checkpoint
pytorch_model.bin, plus tokenizer and config). No extra download is required.
Layout of release/model/:
config.json
tokenizer.json (from cijov/cijov-lang-tokenizer)
tokenizer_config.json
special_tokens_map.json
chat_template.jinja
pytorch_model.bin
2) Install Runtime
From this release/ folder:
python -m venv .venv
source .venv/bin/activate
pip install -U pip
pip install torch transformers safetensors sentencepiece accelerate
3) Run Generation
python generate.py \
--model-dir model \
--prompt "Write a short bilingual FR/RO story about a train station." \
--max-new-tokens 180 \
--temperature 0.8 \
--top-p 0.9
Example Generations
The examples below were produced by this release checkpoint with
--max-new-tokens 90 --temperature 0.7 --top-p 0.9. The model keeps a single
language per prompt (no cross-language drift) across all four supported languages.
English
Prompt: The history of the Roman Empire began
to be the subject of some controversy over the medieval period. The Greek Empire, under a great extent, was a period of radical and religious and political upheaval and civilizations. In the 18th century, the empire was considered one of the greatest military power in the region. The empire was more than just a political and economic system, and a system of military influence led to the emergence of new lands in the region.
French
Prompt: L'intelligence artificielle est
une branche de l'intelligence et de l'économie qui vise à créer des systèmes d'IA capables de résoudre des problèmes et de s'adapter aux changements. Elle se concentre sur la manière dont les algorithmes sont traités, et comment les humains peuvent résoudre des problèmes plus rapidement et plus précisément. L'intelligence artificielle est largement utilisée dans les domaines de l'informatique et de l'énergie.
Spanish
Prompt: La ciudad de Madrid es
sede del gobierno de España. Sus actividades principales son: turismo, deportes, comida y entretenimiento.
Romanian
Prompt: Pădurea era tăcută în acea
zi, și încă mai era în continuare cea mai strălucitoare insulă din afara orașului. În prezent se află locul al doilea, însă acesta era cel care se va întâmpla în primul rând. În anii ’70, acest lucru a afectat locul, dar a fost întotdeauna un loc mai puternic.
Note: This is a 1.2B base model. It is fluent and language-consistent but not a reliable source of facts. For cleaner long-form output, use
--repetition-penalty 1.15.
4) License Requirement (Important)
This release uses Cijov Apache License 2.0.
If you redistribute this model package or derivatives, include:
LICENSENOTICE
Keep the license and notice with the distributed files/binaries.