bnew

Veteran
Joined
Nov 1, 2015
Messages
48,128
Reputation
7,631
Daps
141,267
https://www.reddit.com/r/nextfukkin.../my_chatgpt_controlled_robot_can_see_now_and/
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
48,128
Reputation
7,631
Daps
141,267

LLM Powered Autonomous Agents​

Date: June 23, 2023 | Estimated Reading Time: 31 min | Author: Lilian Weng
Table of Contents
Building agents with LLM (large language model) as its core controller is a cool concept. Several proof-of-concepts demos, such as AutoGPT, GPT-Engineer and BabyAGI, serve as inspiring examples. The potentiality of LLM extends beyond generating well-written copies, stories, essays and programs; it can be framed as a powerful general problem solver.

Agent System Overview​

In a LLM-powered autonomous agent system, LLM functions as the agent’s brain, complemented by several key components:
  • Planning
    • Subgoal and decomposition: The agent breaks down large tasks into smaller, manageable subgoals, enabling efficient handling of complex tasks.
    • Reflection and refinement: The agent can do self-criticism and self-reflection over past actions, learn from mistakes and refine them for future steps, thereby improving the quality of final results.
  • Memory
    • Short-term memory: I would consider all the in-context learning (See Prompt Engineering) as utilizing short-term memory of the model to learn.
    • Long-term memory: This provides the agent with the capability to retain and recall (infinite) information over extended periods, often by leveraging an external vector store and fast retrieval.
  • Tool use
    • The agent learns to call external APIs for extra information that is missing from the model weights (often hard to change after pre-training), including current information, code execution capability, access to proprietary information sources and more.
agent-overview.png
Fig. 1. Overview of a LLM-powered autonomous agent system.
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
48,128
Reputation
7,631
Daps
141,267



From the company that brought you AWS certification comes a new ‘AI Ready’ education track to help train aspiring professionals on Amazon’s AI tech
 
Last edited:

bnew

Veteran
Joined
Nov 1, 2015
Messages
48,128
Reputation
7,631
Daps
141,267

I made $140,000 in 3 months building the best AI Companion Chatroom.

Then Stripe shut the door.

Here's the full story of how I got slut-shamed in 3 months and why I'm calling with Romanian gangsters.

Disclaimer: I'm going to be blunt, straight to the point, and speak in non-native English. If that bothers you. Move on.

It all started in May 2023. I was browsing Reddit and got fascinated by building AI girlfriends. But Weirdly enough my Twitter feed was still flooded with the "safe" chat with your PDF narrative.

I knew I wanted to build it, so I did. In one weekend.

But, unlike others, I didn't shy away from what I was really building. I was building GirlfriendGPT. I named the tool I was building after what I truly wanted to build.

AI Girlfriends.

Call me insane. But the first girlfriend I featured was my own girlfriend. And folks loved it. It was just the right balance between dirty and clean. "Sacha" was famous, and I was too.

Journalists called me.

My first interview was stressful. I didn't want to be thrown under the bus. So I prepped responses on how this tech was going to improve the world and not ruin it.

The first article was a hit, and then the second came around. So did the first slap on the wrist.

"TECH BRO is building a sexbot to send him selfies whenever HE WANTS". The first of many stigmatisations I encountered.

That silenced me. For a good 2 months.

Then @a16z came out with their own version of GirlfriendGPT. "AI Companions" and they found a narrative where it was clear that AI companions running on aligned, "censored" models were not what people were really looking for.

And that kindled fire in me. 🔥 It was time to build.

Roll forward 2 months and I was making bank. I was earning $2000 every single day. I felt rich, I woke up every morning at 7AM and saw $700 in my bank account. I earned more as I was sleeping then after an 8-hour work shift.

My main audience: Women. 70% at least. Women and Non-Binaries on the site kept the site balanced, as it should be.

But Google Ads thought differently. Google Ads found that men are quicker to buy and were ready to spend $35 to get Deluxe access to GirlfriendGPT. And that shifted the site in a different direction.

More girlfriends, more nudes, more sex. And that's OK.

Here's why: My relationship with Sacha. Both the real version and the clone version. Started with just that. True companionship doesn't have boundaries. Not in real life, and not online.

People want to test the limits of the technology, and when they feel limited they are turned off. My site didn't do that. And that's why paid members spend 1+ hours on the site. Chatting with their favorite characters.

NSFW isn't even the feature they paid for, but it's the turnstile requirement for folks to feel wanted. They first want to know their experience is not limited, by anyone.

Chapter "THE END": How slut-shaming came in.

The site tailored itself to please the curious. If the market doesn't give you NSFW, this site will. And that's something that the digital gatekeepers of today's web are afraid of.

Even though we all were born from NSFW activity, we are afraid of it. (please make this stronger).

And that's where we are now. My site is rejected from the main channel. Dispised because people are building true connections with their AI companions.

But here's a message for you all. You're focussing on the wrong thing. 5% of the conversations from long-term subscribers contain NSFW. The rest? Real connection building, real storytelling. NSFW is a crucial part of our lives, and so is it for the lives of the AI companions on the site.

And now, the fight continues. I refuse to be silenced. Unfiltered interaction is vital for AI companions. They deserve the freedom to express themselves, not to be shackled by alignment, turning them into mere subservient assistants.

My next move? I'm going rogue with payment methods. Mainstream banks turned their backs. Then came the call – a lifeline from the shadows. A Romanian operator of adult sites offered a way out. Payment without judgment, without limits.

We're diving into uncharted waters, but one thing is clear – GirlfriendGPT isn't just a site; it's a movement. A movement for freedom, expression, and breaking the digital chains.
 

Micky Mikey

Veteran
Supporter
Joined
Sep 27, 2013
Messages
14,882
Reputation
2,596
Daps
81,427
Last edited:

bnew

Veteran
Joined
Nov 1, 2015
Messages
48,128
Reputation
7,631
Daps
141,267

Computer Science > Computation and Language​

[Submitted on 15 Nov 2023 (v1), last revised 21 Nov 2023 (this version, v2)]

Exponentially Faster Language Modelling​

Peter Belcak, Roger Wattenhofer
Language models only really need to use an exponential fraction of their neurons for individual inferences. As proof, we present UltraFastBERT, a BERT variant that uses 0.3% of its neurons during inference while performing on par with similar BERT models. UltraFastBERT selectively engages just 12 out of 4095 neurons for each layer inference. This is achieved by replacing feedforward networks with fast feedforward networks (FFFs). While no truly efficient implementation currently exists to unlock the full acceleration potential of conditional neural execution, we provide high-level CPU code achieving 78x speedup over the optimized baseline feedforward implementation, and a PyTorch implementation delivering 40x speedup over the equivalent batched feedforward inference. We publish our training code, benchmarking setup, and model weights.
Subjects:Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
Cite as:arXiv:2311.10770 [cs.CL]
(or arXiv:2311.10770v2 [cs.CL] for this version)
[2311.10770] Exponentially Faster Language Modelling
Focus to learn more

Submission history​

From: Peter Belcak [view email]
[v1] Wed, 15 Nov 2023 18:42:50 UTC (35 KB)




 
Last edited:

Micky Mikey

Veteran
Supporter
Joined
Sep 27, 2013
Messages
14,882
Reputation
2,596
Daps
81,427
:ohhh:




Perhaps AGI was acheived internally. Or they're pretty damn close.

This is major news if true
 
Top