LLM : Performance Benchmarks

 Kishan,

 

Checked the counters ( Visitors So far / Questions So far ) on the dashboard > https://hemenparekh-dashboard.vercel.app

 

Looks good

 

That leaves buttons >  Download / Share ( on Consensus Block ) .. and … SUBSCRIBE ( for LLM : Performance Benchmark )

 

Of these, I suppose the “ Download / share “ buttons are very standard and may not take you much time

 

As far as that “ LLM : Performance Benchmark “ button is concerned, I believe you will need to “ experiment “ ( to see the actual results ) for , may be 8 / 10 days , and satisfy yourself that there is actually a “ noticeable improvement “ in the SCORES of these LLMs , as more and more questions get asked / answered

 

Then there is the question of telling the visitors, what this SCORE is all about ?  What is its significance ? How is it computed ?

 

This would require , placing on the second page a link which reads : LLM :  Performance Benchmarks

 

Clicking this will pop up following message ( to be suitably modified by you ) :

 

Dear Visitors :

 

Among many unique features of this “ Collaborative / Cooperative “ platform of 4 LLMs, one that gravitates towards an AGI is :

 

Ø  At the end of each round of ANSWERS, each participating LLM uses the input from its predecessor, as “ Training Matter “ and IMPROVES its own performance along following BENCHMARK criteria :

 

#   Completeness   #   Practicality    #   Truthfulness   #   Ethicality

 

To view , select :

 

Ø  LLM  ………… [  ChatGPT – Gemini – Claude - Grok   ] …….drop-list

Ø  Year………….. [  2025 – 2026 – 2027  ] …… drop-list ( further selection of MONTHS )

 

Those of you who want this tabulation delivered , every Sunday , in your mailbox , please SUBSCRIBE (  Your Email……. / SUBMIT )

 

 

 

hcp