Text Generation
Transformers
Safetensors
GGUF
French
qwen3
twitch
chat
emotion
irony
french
slang
lora
conversational
text-generation-inference
Instructions to use aelmizeb/chatvibe-fr-0.6b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use aelmizeb/chatvibe-fr-0.6b with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="aelmizeb/chatvibe-fr-0.6b") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("aelmizeb/chatvibe-fr-0.6b") model = AutoModelForCausalLM.from_pretrained("aelmizeb/chatvibe-fr-0.6b", 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=256) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use aelmizeb/chatvibe-fr-0.6b with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "aelmizeb/chatvibe-fr-0.6b" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "aelmizeb/chatvibe-fr-0.6b", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/aelmizeb/chatvibe-fr-0.6b
- SGLang
How to use aelmizeb/chatvibe-fr-0.6b 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 "aelmizeb/chatvibe-fr-0.6b" \ --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": "aelmizeb/chatvibe-fr-0.6b", "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 "aelmizeb/chatvibe-fr-0.6b" \ --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": "aelmizeb/chatvibe-fr-0.6b", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use aelmizeb/chatvibe-fr-0.6b with Docker Model Runner:
docker model run hf.co/aelmizeb/chatvibe-fr-0.6b
chatvibe-fr-0.6b
Mini-modรจle qui dรฉcode la vibe des messages de chat gaming francophone : argot (ยซ c'est chaud ยป, ยซ jpp ยป, ยซ wesh ยป), emotes Twitch (KEKW, Sadge, monkaS, Kappaโฆ) et ironie (ยซ bravo le gรฉnie ๐ ยป).
Il renvoie une fiche JSON par message, pensรฉe pour รชtre affichรฉe (terminal, overlay OBS) :
{"emotion":"moquerie","intensite":3,"ironie":true,"icone":"๐คก"}
รmotions : joie, hype, rire, surprise, amour, tristesse, frustration, colere, peur, moquerie, question, neutre
Utilisation
pip install "chatvibe-fr[model] @ git+https://github.com/aelmizeb/chatvibe-fr"
chatvibe --model aelmizeb/chatvibe-fr-0.6b analyse "KEKW il est tombรฉ dans le vide"
chatvibe --model aelmizeb/chatvibe-fr-0.6b twitch <chaine> --overlay
Rรฉsultats (data/test.jsonl, messages รฉcrits ou annotรฉs ร la main uniquement)
| Dรฉcodeur | JSON valide | รmotion | Ironie F1 | รcart intensitรฉ |
|---|---|---|---|---|
| Rรจgles (baseline) | 100 % | 70,0 % | 0,75 | 1,02 |
| Qwen3-0.6B (sans fine-tuning) | 7,5 % | 12,5 % | 0,00 | 1,93 |
| chatvibe-fr-0.6b (v0.1, donnรฉes synthรฉtiques) | 100 % | 85,0 % | 0,75 | 0,35 |
Entraรฎnement
- Base : Qwen3-0.6B, LoRA r=16 sur toutes les projections linรฉaires, 3 รฉpoques, lr 2e-4, bf16
- Donnรฉes : 2 841 exemples synthรฉtiques (gabarits + argot + bruit) ; aucun message rรฉel annotรฉ pour cette v0.1 (prรฉvu pour la v1)
- รvaluation : 40 messages รฉcrits ร la main, jamais vus ร l'entraรฎnement
- Matรฉriel : NVIDIA RTX 3060 12 Go, durรฉe 7 min 42 s
Limites
- Franรงais gaming uniquement ; l'argot รฉvolue vite.
- L'ironie reste difficile, surtout sans contexte du stream.
- Ne pas utiliser pour modรฉrer ou sanctionner des personnes : c'est un outil d'affichage.
Donnรฉes
Messages collectรฉs sur des chats publics, sans pseudonymes, liens et mentions filtrรฉs.
- Downloads last month
- -