Choosing the right model
The main constraints when picking a model for a given task are:
- Privacy / legal reasons
- Quality of the response
- Latency, time to provide a response
- Memory / Compute requirements
- The type of response
A leaderboard with the different models and benchmark are available here:
- https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
- https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard
Open AI
| Model | Response Quality | Memory Requirement | Purpose | Privacy |
|---|---|---|---|---|
| GPT-3.5 (API) | High | None | Multi-purpose, text generation | Poor |
| GPT-4 | High | None | Multi-purpose, text generation | Poor |
| Codex | High | None | Code completion | Poor |
Open source model
Modèles de langages Open Source pour la génération de code,\ un aperçu