Llama Panel 🦙🦙🦙

🦙🦙🦙 Llama Panel 🦙🦙🦙

A multi-model agentic AI application that provides internet backed, consensus based answers from a configurable panel of LLMs.

Open Source Greatness in Numbers

Open Source LLMs have come a long way and can now challenge closed source models! Excitingly several Instruction Tuned open source models are now out and proving quite capable. Moreover, thanks to great improvements in efficiency and quantization, capable models are now small enough to run many on one system.

Building on the idea that groups outperform individuals on various cognitive tasks, I set out to create at simple application to see if these latest models perform better as a group!

Python Leading Llama Leading Llamas 🐍🦙🔎🦙🦙🦙

For simplicity, I built the app in Python on top of Ollama. A large capable model, ideally instruction tuned, leads the investigation, querying the panel of models and internet until a reasonable consensus is reached!

Sample Run!



gregarious@hal9000:~/llama_panel$ ./llama_panel.py "Are we past the AI event horizon today?"
Initializing expert panel...
Expert: mistral-small3.2 (temp: 0.0)
Panel: gemma3:4b (temp: 0.50), granite3.3:2b (temp: 0.50), qwen3:4b (temp: 0.50)

--- Expert Reasoning Step 1/20 ---
> Expert wants to use tool: 'llama_panel'
> Reason for tool selection: I need to understand the panel's perspective on the current state of AI development.

> Consulting the panel with the question: 'Are we past the AI event horizon today?'
  - Querying panelist 'gemma3:4b' (temp: 0.50)...
  - Querying panelist 'granite3.3:2b' (temp: 0.50)...
  - Querying panelist 'qwen3:4b' (temp: 0.50)...
> Panel consultation complete.

--- Expert Reasoning Step 2/20 ---
> Expert wants to use tool: 'search_web'
> Reason for tool selection: I need to find current information on whether we have passed the AI event horizon in 2025.

> Performing Google search for: 'Are we past the AI event horizon today 2025'
> Found 10 URLs.

--- Expert Reasoning Step 3/20 ---
> Expert wants to use tool: 'get_webpage'
> Reason for tool selection: This is Sam Altman's personal blog post where he discusses the AI event horizon and the gentle singularity.

> Fetching content from URL: https://blog.samaltman.com/the-gentle-singularity
> Successfully fetched and parsed content from https://blog.samaltman.com/the-gentle-singularity

--- Expert Reasoning Step 4/20 ---
> Expert wants to use tool: 'get_webpage'
> Reason for tool selection: This article provides a different perspective on whether we have crossed the AI event horizon.

> Fetching content from URL: https://www.cityam.com/have-we-crossed-the-ai-event-horizon/
> Successfully fetched and parsed content from https://www.cityam.com/have-we-crossed-the-ai-event-horizon/

--- Expert Reasoning Step 5/20 ---
> Expert wants to use tool: 'llama_panel'
> Reason for tool selection: I need to gather diverse opinions from the panel to form a consensus.

> Consulting the panel with the question: 'Based on the information provided, do you agree that we have passed the AI event horizon? Please provide reasoning.'
  - Querying panelist 'gemma3:4b' (temp: 0.50)...
  - Querying panelist 'granite3.3:2b' (temp: 0.50)...
  - Querying panelist 'qwen3:4b' (temp: 0.50)...
> Panel consultation complete.

--- Expert Reasoning Step 6/20 ---
> Expert wants to use tool: 'llama_panel'
> Reason for tool selection: I need to gather diverse opinions from the panel to form a consensus.

> Consulting the panel with the question: 'Based on the information provided, do you agree with the conclusion that while we may not have crossed the 'event horizon' in a technical sense, the rapid integration of AI into society suggests that we are on the cusp of a transformative era—one that will require both innovation and responsibility?'
  - Querying panelist 'gemma3:4b' (temp: 0.50)...
  - Querying panelist 'granite3.3:2b' (temp: 0.50)...
  - Querying panelist 'qwen3:4b' (temp: 0.50)...
> Panel consultation complete.

--- Expert Reasoning Step 7/20 ---
> Expert wants to use tool: 'final_answer'
Based on the diverse opinions gathered from the panel, we can conclude the following: 
While we may not have crossed the 'event horizon' in a technical sense—achieving
human-level AI (AGI)—AI's integration into society is rapidly transforming our world, 
blurring the lines between human and machine intelligence. The 'event horizon' 
metaphorically represents the point where AI becomes so pervasive that its impact is 
irreversible and ingrained in daily life, shaping our expectations and norms. The debate 
reflects a tension between the potential of near-term advancements and the challenges of 
achieving AGI. Balancing optimism with caution is essential as we navigate this 
transformative era of AI integration, focusing on ethical development, safety measures, 
and fostering collaboration between humans and AI to maximize benefits while mitigating 
risks.

Application, including usage and details on GitHub: GregariousEngineering/llama_panel


Up next:

Group-to-Individual thinking!

Comments

Post a Comment

Popular posts from this blog

Hackles Feedback Hypothesis

Image
Hackles Feedback Hypothesis “The  facial feedback hypothesis , rooted in the conjectures of  Charles Darwin  and  William James , is that one's facial expression directly affects their  emotional  experience. Specifically, physiological activation of the facial regions associated with certain emotions holds a direct effect on the elicitation of such emotional states, and the lack of or inhibition of facial activation will result in the suppression (or absence altogether) of corresponding emotional states.” https://en.m.wikipedia.org/wiki/Facial_feedback_hypothesis I ask, does smoothing or raising a dog’s hackles affect their emotional experience??
Read more

Turn Off Those Super Annoying Autoplay Ads in FireTV

Image
Bad Decisions: In moves seen aimed at ensure I finally switch devices, in addition to recently adding ads to Prime, Amazon decided to start autoplaying ads as soon as the device turns on. Buried Settings: Fortunately, this can be disable under Settings > Preferences > Featured Content. While in Preferences, I stopped in Privacy Settings to turn Off the slew of tracking On by default. Repeated Pain Sadly these settings are per device, not account, so you’ll have to make them on every device.
Read more

Change these “Off-Facebook” Facebook Privacy Settings

Image
Change these “Off-Facebook” Facebook (and Instagram) Privacy Settings New Settings! OK, perhaps not new. It’s been many years since I checked any social media regularly and I should really just delete my accounts, but like many, I leave them collecting dust… But then, over the weekend I received a slew of recovery tokens for my Facebook account, which prompted me to head over to Settings to check activity and enable MFA. While there I noticed the new settings, buried deep as you’d expect. I am genuinely ecstatic about these settings. Their defaults leave nearly everyone open to widespread invasion of privacy, but they still have the settings available. I suspect we have the ever increasing barrage of new consumer privacy oriented laws from cities and states (e.g. California Consumer Privacy Act) to country unions (GDPR). Apply these in addition to standard privacy settings, as described well in Consumer Reports’ How to Use Facebook Privacy Settings . They also apply to connected Instag...
Read more