The R1 reasoning model introduced by Deepseek was criticized for censorship of China -related issues. Perplexity has modified this open source model and releases a new version that no longer censors content.
According to Neoowin, the PERPLEXITY startup has introduced its new model R1 1776, which, unlike the official version of Deepsic, deletes China -related censorship and gives users detailed and neutral information. This model is full of open source and is available through “HuggingFace” and “API Sonar Perplexity”.
PERPLEXITY Uncensored Dip -Sick Model
Perplexity has taken a different approach to improve R1. Using human experts, the Perplexity team identified and designated 300 sensitive issues that the Chinese government censored and designed the criterion for identifying multiple censorship. This model has been taught using 40,000 multifaceted questions to improve its reasoning model and not censor its answers.
As a result, the R1 1776 has been able to effectively use impartial processing and its accuracy has increased compared to previous versions.
Accessibility and how to get the R1 1776 model
The R1 1776 is now available through the HuggingFace and API Sonar Perplexity and can be easily downloaded and used. This model is especially for those looking for impartial and accurate information in various fields.
Perplexity is now intended to consistently improve its models by focusing on more accurate and removing any censorship.
As can be seen in the image above, there are differences between the original version of the R1 and R1 1776 in benchmarks, but these are not much.
The R1 1776 is designed from Perplexity to eliminate Chinese censorship and provide more accurate and impartial information. Using advanced machine learning methods and human analysis, this model has been able to provide users with impartial responses. This model is especially for those looking for valid and uncensored information in various fields.
Latest Passing over countries : Spain | Dominica | United Arab Emirates
RCO NEWS
