Atomizr.ai

Project name: Objective(s):

Advanced

AI Model: Llama 3.2 (3B) - Fast, Good Quality Gemma 3 (4B) - Balanced Performance Llama 3.1 (8B) - Higher Quality Two-Stage (Gemma → Llama 3.1) - Best Quality, Slower Two-Stage generates with Gemma, then enhances with Llama 3.1

AI Prompt (editable): This prompt will be sent to the AI model. You can customize it to get better results.

Model Parameters

Temperature (0-2): Controls randomness: 0 = deterministic/factual, 1 = balanced, 2 = highly creative/random

Top P (0-1): Nucleus sampling: lower values = more focused, higher values = more diverse word choices

Max Tokens: Maximum length of generated response (roughly 4 characters per token)

Repeat Penalty (1-2): Penalizes repetitive text: 1.0 = no penalty, higher values reduce repetition

Context Control

Context Window Size: Size of context window in tokens (default: 2048)

Context Memory: Tokens to keep from prior context (0 = stateless/fresh each time, for repeatable testing)