« Hermes » : différence entre les versions
De Le Wiki de Lug
Autres actions
| Ligne 21 : | Ligne 21 : | ||
# apt install curl git | # apt install curl git | ||
# curl -fsSL <nowiki>https://</nowiki>raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash | # curl -fsSL <nowiki>https://</nowiki>raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash | ||
<font color = gold>How would you like to set up Hermes?</font> | |||
> <font color = lightgreen>(○) Quick setup — provider, model & messaging (recommended)</font> | |||
=== Ollama local === | |||
<font color = gold>Select provider:</font> | |||
><font color = lightgreen>(○) Custom endpoint (enter URL manually)</font> | |||
Custom OpenAI-compatible endpoint configuration: | |||
API base URL [e.g. <nowiki>https://</nowiki>api.example.com/v1]: http://<font color = blue>IP_OLLAMA</font>:<font color = green>11434</font>/v1 | |||
API key [optional]: | |||
Detected model: <font color = green>mon_model</font> | |||
Use this model? [Y/n]: <font color = blue>Y</font> | |||
Context length in tokens [leave blank for auto-detect]: | |||
Display name [192.168.2.216:11434]: | |||
Remplir les options désirées puis fin de l'installation ; | |||
Launch hermes chat now? [Y/n]: <font color = blue>y</font> | |||
Version du 23 avril 2026 à 19:25
Prérequis
(Hermes seul)
- Une machine ou VM dédiée (LXC compatible)
Pour un Hermes de base (sans LLM local) :
- RAM :
- 1 Go → minimum (test uniquement)
- 2 Go → recommandé
- 4 Go -> confortable
- CPU / vCPU :
- 1 vCPU → suffisant
- 2 vCPU → recommandé
- Espace disque :
- 5 Go → minimum
- 10 Go → confortable
- GPU :
- Inutile
Installation Hermes (service dédié)
Installation (Debian)
# apt update && apt upgrade # apt install curl git # curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
How would you like to set up Hermes? > (○) Quick setup — provider, model & messaging (recommended)
Ollama local
Select provider: >(○) Custom endpoint (enter URL manually)
Custom OpenAI-compatible endpoint configuration: API base URL [e.g. https://api.example.com/v1]: http://IP_OLLAMA:11434/v1 API key [optional]:
Detected model: mon_model Use this model? [Y/n]: Y
Context length in tokens [leave blank for auto-detect]:
Display name [192.168.2.216:11434]:
Remplir les options désirées puis fin de l'installation ;
Launch hermes chat now? [Y/n]: y