Charterbeta

Sandbox

Configuration

Model Selection

Choose a base model for your training job. You can select from popular presets or specify a custom Hugging Face model.

Model Selection

Choose a base model for training or enter a custom Hugging Face model ID

Popular Models

SmolLM2-1.7B
1.7B params

Lightweight 1.7B parameter model - great for quick experiments and learning

LoRA r=81 GPU1000 tokens/lease
Mistral-7B-Instruct
7B params

Popular 7B parameter model with strong performance across tasks

LoRA r=162 GPUs2000 tokens/lease
Llama3-8B-Instruct
8B params

Latest Llama model with excellent instruction following capabilities

LoRA r=162 GPUs2000 tokens/lease

Custom Model

Format: organization/model-name (e.g., microsoft/DialoGPT-medium)

Run Summary

Status
Not validated
Configuration
Model:Not selected
Dataset:Not selected
LoRA Rank:8
Target Modules:2
Tokens/Lease:1,000
GPUs:1
Provider:salad
Cost Estimate
Total:$0.5
$0.50/GPU • salad
Need Help?
• Check validation messages above
• Start with model and dataset selection
• Use training presets for quick setup