Hi, good morning


This is to share with you that our research group made publicly available the 
beta version of
*the chatbot for the Portuguese language based on open LLMs, Evaristo.ai 
<https://evaristo.ai/>
*
One of such LLMs is Gervásio 8B 
<https://huggingface.co/PORTULAN/gervasio-8b-portuguese-ptpt-decoder>, we 
developed for Portuguese and we are releasing also now,
and which is the LLM that you find active by default when arriving at this 
chatbot.

Though you may not be proficient in Portuguese, on-the-fly translators will 
likely help
you to get its content and its basic functioning. *We invite you, and your 
colleagues
and students, to visit it and try out this first test version. We welcome any 
help
you can give us in testing it, any feedback and any suggestions you can share 
with us.*



It has a unique set of features, among others: it is an open AI chatbot for
the Portuguese language; it is multi-model and multi-heteronym, as well as 
being agentic,
multi-tool and multi-modal; it does not track its users or pass on their 
content to third parties,
safeguarding user privacy and ownership of their content.

You'll find the presentation of its motivation in this press release 
<https://evaristo.ai/assets/pressRelease_EvaristoAI.pdf>(in English),
that can be complemented by the more complete description in the section about 
<https://evaristo.ai/about> (in Portuguese)



The current open LLMs available are typically between 10 and 100 times smaller 
than
the top-of-the-range closed LLMs used in commercial chatbots, so the costs 
associated
with training and operating them are much lower. The performance of open LLMs, 
however,
is much more satisfactory than this linear disproportion would suggest. They 
therefore have
an excellent ratio of performance quality versus cost, and are a viable option 
for
fully autonomous generative AI services focused on concrete use cases.

In this context, we see this chatbot as a milestone in the democratization of
generative technology for the Portuguese language, through open LLMs, by 
encouraging
more and more organizations to move forward with their own AI services,
rooted in their own computers, and focused on their concrete use cases.


Have a nice day,

António





_______________________________________________
Corpora mailing list -- [email protected]
https://list.elra.info/mailman3/postorius/lists/corpora.list.elra.info/
To unsubscribe send an email to [email protected]

Reply via email to