Ready when you are

Ask anything, or give it a task.

Fine-tune a model

Runs locally through praisonai-train.

Base model A Hugging Face id.

Run on Needs an NVIDIA GPU.

Method Trains on completions.

SSH host An alias from your ~/.ssh/config.

Remote python Interpreter on that machine.

Remote directory Where the run works. Created if absent.

Dataset A Hugging Face dataset id.

Split

Epochs

Max steps 0 runs the full epochs.

Advanced

LoRA rank

LoRA alpha

Learning rate

Max sequence length

Batch size per device

Gradient accumulation

Load in 4-bit

Output directory

Channels

Telegram, Slack, and Discord. Tokens use env: refs from ~/.praisonai/.env

Your channels

Channel log

Select a channel and click “View log”.

Knowledge (RAG)

Index a folder of documents, then enable “Use knowledge” in chat.

Folder paths (one per line)

Not configured

Image generation

Create images with DALL·E (OpenAI API key in Settings or ~/.praisonai/.env).

For MiniMax, set MINIMAX_API_KEY in the engine environment.

Prompt

Model

Size

Video generation

Optional: set REPLICATE_API_TOKEN in ~/.praisonai/.env (Replicate minimax/video-01).

Model

Prompt

Multi-bot gateway

Run Telegram, Slack, and Discord together on port 8765. Stop individual bots first.

stopped

Creative Studio

Projects, prompts, image & video generation, timeline export.

Project

Set API keys in Settings.

Prompt

Image settings (testing & keyframes)

Image model

Quality

Size

Video model

First-frame image

Duration (sec)

Variations

Jobs

Assets & timeline

0 clips