FAQ

General Information About Eridu

What is Eridu? Eridu is enqAI's unfiltered and unbiased Large Language Model (LLM) designed for deep language understanding and generation across multiple languages.

What does "unfiltered" mean? An unfiltered LLM operates without restrictive guardrails. The outputs reflect the broad spectrum of the model's training dataset, without moderation or trigger warnings.

What does "unbiased" mean? Unlike many LLMs today that contain hardcoded biases to influence responses, Eridu does not incorporate any hardcoded logic to alter its outputs. Users receive responses exactly as the model generates them, based on its learning.

How extensive is Eridu's training dataset? Eridu was trained on a dataset encompassing more than a trillion tokens, drawn from a diverse range of text sources, including books, scientific articles, websites, and specialized datasets for technical, medical, legal, and creative writing. This ensures a comprehensive linguistic and factual knowledge base.

Technical Specifications

What is the parameter count of Eridu? Eridu is built with 45 billion parameters, optimizing its capacity for complex language tasks.

What computational setup was required to train Eridu? The training of Eridu required an advanced setup with several hundred GPUs, specifically optimized for deep learning tasks, over multiple weeks.

Can Eridu process text in multiple languages? Yes, Eridu is multilingual, capable of understanding and generating text in dozens of languages, with its performance optimized for English.

What is the maximum token input Eridu can handle? Eridu can process inputs up to 4096 tokens in length, facilitating detailed and context-rich text generation.

How is Eridu's performance benchmarked? Eridu's performance is evaluated using perplexity scores, task-specific accuracy metrics, and qualitative assessments from human evaluators for a range of tasks such as translation, summarization, and creative writing.

Applications and Usage

What applications is Eridu suited for? Eridu is applicable in automated content creation, conversational agents, text summarization, language translation, educational tools, and more.

Is Eridu capable of generating creative content? Yes, Eridu excels in generating creative content like stories, poetry, and informative articles, leveraging its deep understanding of language patterns.

What hardware is recommended for running Eridu efficiently? For real-time applications, state-of-the-art GPUs or AI-specific processors are recommended to handle Eridu's computational demands efficiently.

Decentralization and Crypto Token

Is enqAI decentralized? Eridu will be decentralized, enabling any entity with the computational power to run the model and connect to the enqAI network. Nodes in this network earn by running inferences and must stake enqAI tokens to participate. Tokenholders not running nodes can delegate their tokens to active nodes.

Security and Privacy

Why do I need to create an account? Creating an account ensures that Eridu is used by verified individuals and helps manage conversation memory for a personalized experience.

Are my conversations encrypted? Yes, your conversations are encrypted, and no one except you can access them.

Future Developments and How to Contribute

What are you preparing for the next releases? Future functionalities will include a tier system for increased prompt quotas, further decentralization, image generation, text to speech, document upload and analysis, extended internet access, and an expanded knowledge API.

I have datasets I am willing to share to train Eridu. What should I do? We are always looking for new datasets. You can contribute by getting in touch with us at [e-mail].

How can I contribute? If you have ideas, datasets, or want to get involved in other ways, contact us at [e-mail].