Axiom Distributed AI: Difference between revisions

From BOINC Projects
Jump to navigation Jump to search
← Older edit
VisualWikitext
Al Piskun (talk | contribs)
1,221 edits
{{Start date and age|YEAR|MONTH|DAY}}
Al Piskun (talk | contribs)
1,221 edits
No edit summary
 
(2 intermediate revisions by the same user not shown)
Line 1: Line 1:
[[File:{{#setmainimage:Axiom-logo.png}}|alt=Axiom Distributed AI|center|frameless]]
{{Infobox software
{{Infobox software
| name                = Axiom Distributed AI
| name                = Axiom Distributed AI
| screenshot          = Axiom-logo.png
| screenshot          = Axiom-logo.png
| caption              = Logo of the Axiom Distributed AI project
| caption              = Logo of the Axiom Distributed AI project
 
| status              = Inactive
 
| category            = Experimental
| compute              = CPU and GPU
| dependencies        = None
| developer            = PyHelix / Axiom Project Team
| developer            = PyHelix / Axiom Project Team
| released            = {{Start date and age|2026|01|23}}
| released            = {{Start date and age|2026|01|23}}
| discontinued        = Yes (Completed Test / Proof of Concept)
| discontinued        = {{Start date and age|2026|03|21}}
 
| repository          = [https://github.com/PyHelix/boinc-Axiom github.com/PyHelix/boinc-Axiom]
 
| repository          = [https://github.com/PyHelix/boinc-Axiom github.com/PyHelix/boinc-Axiom]
| programming language = Python, C++
| programming language = Python, C++
| operating system    = Windows, Linux
| operating system    = Windows, Linux
 
| stats as of          = {{Start date and age|2026|03|14}}
 
| average performance  = 5398.69 GigaFLOPS
| platform            = BOINC
| active users        = 65
| genre                = Distributed computing, Artificial Intelligence
| total users          = 94
| active hosts        = 219
| total hosts          = 159
| license              = Open Source
| license              = Open Source
| website              = [https://web.archive.org/web/20260319085629/https://axiom.heliex.net/ Archived website]
| website              = [https://web.archive.org/web/20260319085629/https://axiom.heliex.net/ Archived website]
}}
}}


'''[https://axiom.heliex.net/ Axiom Distributed AI]''' was an experimental volunteer distributed computing project built on the [https://berkeley.edu BOINC] platform. Launched in early 2026, the project explored an alternative approach to artificial intelligence training by using decentralized, biologically inspired learning techniques rather than conventional centralized AI infrastructure [https://linustechtips.com].  
[[File:{{#setmainimage:Axiom-logo.png}}|alt=Axiom Distributed AI|center|frameless]]
 
'''Axiom Distributed AI''' was an experimental volunteer distributed computing project built on the [[wikipedia:Berkeley Open Infrastructure for Network Computing|BOINC]] platform.<ref>{{Cite web
|title=Axiom Distributed AI (archived)
|url=https://web.archive.org/web/20260319085629/https://axiom.heliex.net/
|website=Internet Archive
|access-date=2026-05-20
}}</ref>
 
Launched in January 2026, the project explored decentralized and biologically inspired approaches to artificial intelligence training using volunteer CPU and GPU resources contributed by participants worldwide.<ref>{{Cite web
|title=boinc-Axiom
|url=https://github.com/PyHelix/boinc-Axiom
|website=GitHub
|access-date=2026-05-20
}}</ref>
 
The platform experimented with distributed neural-network learning techniques influenced by [[wikipedia:Hebbian theory|Hebbian learning]], attempting to determine whether collaborative volunteer computing infrastructure could train predictive AI systems without relying exclusively on centralized datacenters.<ref>{{Cite web
|title=Hebbian theory
|url=https://en.wikipedia.org/wiki/Hebbian_theory
|website=Wikipedia
|access-date=2026-05-20
}}</ref>
 
Community discussions and archived project materials indicate that Axiom Distributed AI functioned primarily as a short-lived proof-of-concept research experiment before operations concluded in March 2026.<ref>{{Cite web
|title=BOINC forum discussion on Axiom Distributed AI
|url=https://boinc.berkeley.edu/forum_thread.php?id=15694&postid=118571
|website=BOINC
|access-date=2026-05-20
}}</ref>


The project invited volunteers from around the world to donate spare CPU and NVIDIA GPU processing power to help train a shared neural network collaboratively across thousands of computers. Community analysis and forum data indicate that the project has concluded its run, operating primarily as a short-lived experimental test and proof-of-concept [https://berkeley.eduforum_thread.php?id=15694&postid=118571].
Official website: [https://web.archive.org/web/20260319085629/https://axiom.heliex.net/ Archived project website]


Official website: [https://axiom.heliex.net/ Axiom Distributed AI] 
Open-source development repository: [https://github.com/PyHelix/boinc-Axiom GitHub repository]
Open source development: [https://github.com/PyHelix/boinc-Axiom GitHub Repository]


== Overview ==
== Overview ==
Unlike most modern AI systems that rely on centralized datacenters and expensive GPU clusters, Axiom experimented with a decentralized learning architecture inspired by '''Hebbian learning''' — often summarized as:
Unlike most modern artificial intelligence systems that depend on centralized datacenters and large GPU clusters, Axiom Distributed AI experimented with decentralized machine learning architectures inspired by biological neural adaptation.
 
The project's conceptual framework drew from '''Hebbian learning''', commonly summarized as:


<blockquote>
<blockquote>
Line 38: Line 66:
</blockquote>
</blockquote>


The project distributed training tasks across volunteer computers worldwide using the BOINC infrastructure. Each participant contributed small amounts of computation toward a larger shared neural model.
Using the BOINC infrastructure, volunteers contributed spare CPU and NVIDIA GPU processing power to collaboratively train a distributed neural system across many independent computers.<ref>{{Cite web
 
|title=BOINC
Axiom is considered a finished '''proof-of-concept research project''' exploring whether distributed Hebbian learning can scale effectively in a real-world volunteer computing environment [https://boincsynergy.ca].  
|url=https://boinc.berkeley.edu/
 
|website=University of California, Berkeley
== Why Axiom Distributed AI? ==
|access-date=2026-05-20
Traditional AI systems typically depend on:
}}</ref>
* Massive centralized datacenters
* High energy consumption
* Large-scale backpropagation training
* Expensive specialized hardware


Axiom instead investigated whether learning could emerge from:
The project primarily served as an experimental proof-of-concept intended to explore whether distributed Hebbian-style learning could scale effectively within a volunteer computing environment.
* Distributed volunteer computation [https://boincsynergy.ca]
* Correlation-based learning
* Decentralized pattern recognition
* Biologically inspired neural adaptation


The long-term vision was to explore alternative methods for building AI systems that are more decentralized, energy-efficient, community-driven, and resistant to centralized control.
== Goals ==
The primary objective of Axiom Distributed AI was to investigate whether decentralized learning systems could emerge from large-scale volunteer participation rather than centralized AI infrastructure.


== Goal ==
Research goals included:
The immediate goal of the project was to demonstrate that:
:<center>'''Distributed Hebbian learning works at scale.'''</center>


If successful, the research sought to contribute toward new approaches in:
* Distributed neural-network training
* Neural network training
* Decentralized artificial intelligence
* Decentralized artificial intelligence
* Distributed machine learning
* Correlation-based learning systems
* Emergent pattern recognition
* Collaborative AI infrastructure
* Collaborative AI infrastructure
* Emergence-based pattern recognition systems
* Alternative machine learning methodologies


At present, the project's online lifecycle has completed, serving its primary purpose as an experimental research platform and proof concept [https://boincsynergy.ca].
The project also explored whether community-driven AI systems could potentially reduce reliance on expensive centralized datacenters and specialized commercial hardware.


== Methods ==
== Methods ==
Axiom trained neural systems to recognize patterns in data using the spare processing power of volunteer computers. Instead of conventional large language model (LLM) training, Axiom focused on learning:
Axiom Distributed AI trained experimental neural systems using the spare processing power of volunteer computers connected through BOINC.
* Bit-level compression patterns [https://boincsynergy.ca]
 
* Sequence prediction (e.g., predicting the next bit in a stream)
Instead of focusing on large language model (LLM) development, the platform concentrated on lower-level predictive learning techniques including:
* Correlations within data streams
 
* Bit-level compression analysis
* Sequence prediction
* Correlation discovery within data streams
* Generalized pattern recognition
* Generalized pattern recognition
* Distributed learning aggregation


=== Key Technical Concepts ===
=== Technical Concepts ===
# '''Not an LLM'''
==== Not a Large Language Model ====
#* Axiom was '''not''' training a Large Language Model (LLM). The system instead attempted to learn predictive patterns at the bit-stream level.
The project was not designed as a conventional large language model. Instead, the system focused on predictive pattern learning at the bit-stream and correlation level.
# '''Distributed Learning'''
 
#* Training workloads were distributed across volunteers worldwide through BOINC. Each computer contributed partial learning updates to the shared model.
==== Distributed Learning ====
# '''Correlation-Based Credit System'''
Training workloads were distributed globally through volunteer BOINC clients. Individual systems processed partial learning tasks that collectively contributed to the broader neural model.
#* Random or low-quality data naturally contributed less to the model [https://boincsynergy.ca]. Useful learning signals correlated more strongly with other contributors' gradients. This created a natural weighting system favoring meaningful patterns.
 
# '''Noise Reduction'''
==== Correlation-Based Weighting ====
#* Random noise generally fails to correlate with broader learning patterns. As a result, noisy data contributed less to overall model development.
Project documentation suggested that useful data naturally produced stronger correlations with other contributors' results, while random or noisy inputs contributed less significantly to the evolving model.<ref>{{Cite web
# '''Generalization Through Diversity'''
|title=boinc-Axiom
#* Diverse datasets were intended to improve the model's ability to generalize across different data types. Text, code, images, documents, and structured data could all contribute useful patterns.
|url=https://github.com/PyHelix/boinc-Axiom
|website=GitHub
|access-date=2026-05-20
}}</ref>
 
==== Diversity and Generalization ====
The platform encouraged diverse input sources in an attempt to improve generalization capabilities across multiple forms of data, including text, source code, documents, and structured datasets.


=== AI Principal Investigator Experiment ===
=== AI Principal Investigator Experiment ===
According to documentation from early 2026 deployments, the project uniquely experimented with an automated workflow [https://linustechtips.com]. A large language model served as a "Principal Investigator," autonomously designing experiments, deploying tasks to the volunteer client hardware, reviewing return results, and automatically awarding BOINC credit based on the scientific quality of data returned.
Early project descriptions indicated that Axiom experimented with partially automated workflows in which a large language model acted as a form of automated "Principal Investigator".<ref>{{Cite web
|title=Linus Tech Tips BOINC discussion
|url=https://linustechtips.com/
|website=Linus Tech Tips
|access-date=2026-05-20
}}</ref>
 
According to community discussions, the system attempted to automate portions of experiment design, workload deployment, validation review, and BOINC credit allocation based on the quality of returned computational results.


== Monitoring Learning Progress ==
== Monitoring Learning Progress ==
The Axiom website provided live statistics and progress monitoring during its operational window. One key metric tracked was:
The Axiom website provided live monitoring statistics throughout the project's operational period.
 
One of the primary metrics tracked was:
 
:<center>'''Bit loss'''</center>
:<center>'''Bit loss'''</center>


When bit loss trended downward over time, it suggested the neural model was learning meaningful structures and predictive relationships rather than memorizing random noise [https://boincsynergy.ca].  
A declining bit-loss trend was interpreted as evidence that the neural system was identifying meaningful predictive structures rather than memorizing random noise.


Volunteers were able to monitor:
Volunteers could monitor:
* Model learning progress
 
* Training statistics
* Learning progress
* Network participation
* Network participation statistics
* Workunit activity
* Workunit processing activity
* Validation metrics
* Validation metrics
* Host performance data


== Contributing Data ==
== Contributing Data ==
Volunteers were given the option to contribute local files locally to help train the system via unique directory ingestion points on client devices:
Participants could optionally contribute local datasets to assist model training by placing files into designated ingestion directories on their local systems.


=== Windows ===
=== Windows ===
Place files in:
<pre>
<pre>
%USERPROFILE%\Axiom\contribute\
%USERPROFILE%\Axiom\contribute\
</pre>
</pre>


=== Linux Mint 22.3 ===
=== Linux ===
Place files in:
<pre>
<pre>
\boinc\Axiom\contribute\
/boinc/Axiom/contribute/
</pre>
</pre>


=== Supported Contribution Types ===
=== Supported Contribution Types ===
Users were able to contribute many different file types, including:
Supported file categories reportedly included:
 
* Documents
* Documents
* Images
* Images
Line 128: Line 167:
* Logs
* Logs
* Structured datasets
* Structured datasets
* Experimental real-time data streams
* Experimental data streams
 
== Reception and conclusion ==
Axiom Distributed AI attracted attention within portions of the BOINC and technology communities due to its unconventional decentralized AI concepts.
 
Discussion threads on the official BOINC forums questioned both the long-term scientific utility and practical scalability of the system.<ref>{{Cite web
|title=BOINC forum discussion on Axiom Distributed AI
|url=https://boinc.berkeley.edu/forum_thread.php?id=15694&postid=118571
|website=BOINC
|access-date=2026-05-20
}}</ref>
 
The project was also discussed within community BOINC tracking threads on technology forums such as Linus Tech Tips.<ref>{{Cite web
|title=Linus Tech Tips BOINC discussion
|url=https://linustechtips.com/
|website=Linus Tech Tips
|access-date=2026-05-20
}}</ref>
 
No scientific publications associated with Axiom Distributed AI are currently listed within the BOINC Publications Database.<ref>{{Cite web
|title=BOINC Publications
|url=https://boinc.berkeley.edu/pubs.php
|website=BOINC
|access-date=2026-05-20
}}</ref>
 
By March 2026, active work generation and public infrastructure activity had ceased, leaving the project archived as a short-lived experimental volunteer AI initiative.


== Reception and Project Conclusion ==
== Open-source development ==
Axiom Distributed AI entered the BOINC landscape as an experimental project, drawing notable interest from technical communities like the [https://linustechtips.com Linus Tech Tips Monthly BOINC Logs]. However, it also faced scrutiny on the official [https://berkeley.eduforum_thread.php?id=15694&postid=118571 Berkeley BOINC Forums] regarding the ultimate scientific utility of its distributed outputs, with multiple community members noting that the platform behaved more as a brief conceptual test than a long-term infrastructure project.  
Axiom Distributed AI was developed as open-source software. Source code and development repositories remain available on GitHub.


There are no recorded permanent entries for scientific papers authored by Axiom within the official [https://berkeley.edupubs.php Berkeley BOINC Publications Database]. The software repositories and active work unit generations have since wound down, cementing its status as an early-2026 experimental test.
* [https://github.com/PyHelix/boinc-Axiom boinc-Axiom GitHub repository]


== Open Source Development ==
== Project team ==
Axiom Distributed AI was developed as open-source software. Source code and development logs remain accessible on GitHub:
The project was associated with:
* [https://github.com/PyHelix/boinc-Axiom boinc-Axiom GitHub Repository]


== Project Team / Sponsors ==
* PyHelix
* PyHelix
* Axiom Project
* Axiom Project Team
 
== See also ==
* [[wikipedia:Berkeley Open Infrastructure for Network Computing|BOINC]]
* [[wikipedia:Volunteer computing|Volunteer computing]]
* [[wikipedia:Distributed computing|Distributed computing]]
* [[wikipedia:Artificial intelligence|Artificial intelligence]]
* [[wikipedia:Machine learning|Machine learning]]
* [[wikipedia:Hebbian theory|Hebbian learning]]
 
== External links ==
* [https://web.archive.org/web/20260319085629/https://axiom.heliex.net/ Archived Axiom Distributed AI website]
* [https://github.com/PyHelix/boinc-Axiom GitHub repository]
* [https://boinc.berkeley.edu/ BOINC official website]


== External Links ==
== References ==
* [https://pyhelix.com Official Axiom Distributed AI Website]
{{Reflist}}
* [https://github.com Axiom GitHub Repository]
* [https://boincsynergy.ca Axiom Profile on BOINC Synergy Wiki]

Latest revision as of 14:21, 20 May 2026




Axiom Distributed AI
Logo of the Axiom Distributed AI project
Project
StatusInactive
CategoryExperimental
ComputeCPU and GPU
RequiresNone
Development
DeveloperPyHelix / Axiom Project Team
Initial releaseJanuary 23, 2026  (0 years ago)
DiscontinuedMarch 21, 2026  (0 years ago)
Repositorygithub.com/PyHelix/boinc-Axiom
Software
Written inPython, C++
Operating systemWindows, Linux
BOINC statistics
Stats as ofMarch 14, 2026  (0 years ago)
Performance5398.69 GigaFLOPS
Active users65
Total users94
Active hosts219
Total hosts159
Metadata
WebsiteArchived website
LicenseOpen Source

[[File:{{#setmainimage:Axiom-logo.png}}|alt=Axiom Distributed AI|center|frameless]]

Axiom Distributed AI was an experimental volunteer distributed computing project built on the BOINC platform.[1]

Launched in January 2026, the project explored decentralized and biologically inspired approaches to artificial intelligence training using volunteer CPU and GPU resources contributed by participants worldwide.[2]

The platform experimented with distributed neural-network learning techniques influenced by Hebbian learning, attempting to determine whether collaborative volunteer computing infrastructure could train predictive AI systems without relying exclusively on centralized datacenters.[3]

Community discussions and archived project materials indicate that Axiom Distributed AI functioned primarily as a short-lived proof-of-concept research experiment before operations concluded in March 2026.[4]

Official website: Archived project website

Open-source development repository: GitHub repository

Overview

Unlike most modern artificial intelligence systems that depend on centralized datacenters and large GPU clusters, Axiom Distributed AI experimented with decentralized machine learning architectures inspired by biological neural adaptation.

The project's conceptual framework drew from Hebbian learning, commonly summarized as:

"Neurons that fire together wire together."

Using the BOINC infrastructure, volunteers contributed spare CPU and NVIDIA GPU processing power to collaboratively train a distributed neural system across many independent computers.[5]

The project primarily served as an experimental proof-of-concept intended to explore whether distributed Hebbian-style learning could scale effectively within a volunteer computing environment.

Goals

The primary objective of Axiom Distributed AI was to investigate whether decentralized learning systems could emerge from large-scale volunteer participation rather than centralized AI infrastructure.

Research goals included:

  • Distributed neural-network training
  • Decentralized artificial intelligence
  • Correlation-based learning systems
  • Emergent pattern recognition
  • Collaborative AI infrastructure
  • Alternative machine learning methodologies

The project also explored whether community-driven AI systems could potentially reduce reliance on expensive centralized datacenters and specialized commercial hardware.

Methods

Axiom Distributed AI trained experimental neural systems using the spare processing power of volunteer computers connected through BOINC.

Instead of focusing on large language model (LLM) development, the platform concentrated on lower-level predictive learning techniques including:

  • Bit-level compression analysis
  • Sequence prediction
  • Correlation discovery within data streams
  • Generalized pattern recognition
  • Distributed learning aggregation

Technical Concepts

Not a Large Language Model

The project was not designed as a conventional large language model. Instead, the system focused on predictive pattern learning at the bit-stream and correlation level.

Distributed Learning

Training workloads were distributed globally through volunteer BOINC clients. Individual systems processed partial learning tasks that collectively contributed to the broader neural model.

Correlation-Based Weighting

Project documentation suggested that useful data naturally produced stronger correlations with other contributors' results, while random or noisy inputs contributed less significantly to the evolving model.[6]

Diversity and Generalization

The platform encouraged diverse input sources in an attempt to improve generalization capabilities across multiple forms of data, including text, source code, documents, and structured datasets.

AI Principal Investigator Experiment

Early project descriptions indicated that Axiom experimented with partially automated workflows in which a large language model acted as a form of automated "Principal Investigator".[7]

According to community discussions, the system attempted to automate portions of experiment design, workload deployment, validation review, and BOINC credit allocation based on the quality of returned computational results.

Monitoring Learning Progress

The Axiom website provided live monitoring statistics throughout the project's operational period.

One of the primary metrics tracked was:

Bit loss

A declining bit-loss trend was interpreted as evidence that the neural system was identifying meaningful predictive structures rather than memorizing random noise.

Volunteers could monitor:

  • Learning progress
  • Network participation statistics
  • Workunit processing activity
  • Validation metrics
  • Host performance data

Contributing Data

Participants could optionally contribute local datasets to assist model training by placing files into designated ingestion directories on their local systems.

Windows

%USERPROFILE%\Axiom\contribute\

Linux

/boinc/Axiom/contribute/

Supported Contribution Types

Supported file categories reportedly included:

  • Documents
  • Images
  • Source code
  • PDFs
  • Logs
  • Structured datasets
  • Experimental data streams

Reception and conclusion

Axiom Distributed AI attracted attention within portions of the BOINC and technology communities due to its unconventional decentralized AI concepts.

Discussion threads on the official BOINC forums questioned both the long-term scientific utility and practical scalability of the system.[8]

The project was also discussed within community BOINC tracking threads on technology forums such as Linus Tech Tips.[9]

No scientific publications associated with Axiom Distributed AI are currently listed within the BOINC Publications Database.[10]

By March 2026, active work generation and public infrastructure activity had ceased, leaving the project archived as a short-lived experimental volunteer AI initiative.

Open-source development

Axiom Distributed AI was developed as open-source software. Source code and development repositories remain available on GitHub.

Project team

The project was associated with:

  • PyHelix
  • Axiom Project Team

See also

External links

References

  1. ↑ Axiom Distributed AI (archived). Internet Archive. Retrieved 2026-05-20}.
  2. ↑ boinc-Axiom. GitHub. Retrieved 2026-05-20}.
  3. ↑ Hebbian theory. Wikipedia. Retrieved 2026-05-20}.
  4. ↑ BOINC forum discussion on Axiom Distributed AI. BOINC. Retrieved 2026-05-20}.
  5. ↑ BOINC. University of California, Berkeley. Retrieved 2026-05-20}.
  6. ↑ boinc-Axiom. GitHub. Retrieved 2026-05-20}.
  7. ↑ Linus Tech Tips BOINC discussion. Linus Tech Tips. Retrieved 2026-05-20}.
  8. ↑ BOINC forum discussion on Axiom Distributed AI. BOINC. Retrieved 2026-05-20}.
  9. ↑ Linus Tech Tips BOINC discussion. Linus Tech Tips. Retrieved 2026-05-20}.
  10. ↑ BOINC Publications. BOINC. Retrieved 2026-05-20}.
Retrieved from "https://boincsynergy.ca/wiki/index.php?title=Axiom_Distributed_AI&oldid=1216"