Access the Uncensored Model Weights
DeepSeek AI and its open-source language models have taken over the news cycle this week, raising concerns about data privacy, security, and Chinese-government-enforced censorship within their training.
Try R1 via AI Platforms
Perplexity, an AI search platform, has open-sourced the model weights for R1 1776, an uncensored version of R1 that has been post-trained to provide unbiased, accurate, and factual information. Perplexity notes that its decensoring training process did not negatively impact other parts of the model’s performance, including math and reasoning.
Results from Perplexity’s benchmark training comparing their uncensored version of R1 with the original model.
Access R1 on HuggingFace or Perplexity’s Sonar API
The standard (and thus presumably still censored) version of DeepSeek R1 is available on platforms including Azure and Github, AWS, and Nvidia.
DeepSeek R1 on Perplexity
As of last month, Perplexity now hosts DeepSeek R1. The free plan gives users three Pro-level queries per day, which you could use with R1, but you’ll need the $20 per month Pro plan to access the model more than that.
Try R1 on Perplexity
In another post, the company confirmed that it hosts DeepSeek "in US/EU data centers — your data never leaves Western servers," assuring users that their data would be safe if using the open-source models on Perplexity.
Get R1 on You.com
You.com, an AI assistant, offers V3 and R1, similarly only through its Pro tier, which is $15 per month (discounted from the usual $20) without any free queries.
R1 and V3 on You.com
Bryan McCann, You.com cofounder and CTO, explained in an email to ZDNET that users can access R1 and V3 via the platform in three ways, all of which use "an unmodified, open-source version of the DeepSeek models hosted entirely within the United States to ensure user privacy."
Conclusion
In conclusion, Perplexity and You.com have found a way to access the uncensored model weights of R1, providing users with a more accurate and unbiased version of the model. By using the uncensored model, users can access a more comprehensive and accurate source of information, free from censorship.
FAQs
Q: What is DeepSeek AI?
A: DeepSeek AI is a Chinese startup that has developed an open-source language model that has raised concerns about data privacy, security, and Chinese-government-enforced censorship within their training.
Q: What is Perplexity?
A: Perplexity is an AI search platform that has open-sourced the model weights for R1 1776, an uncensored version of R1.
Q: How do I access R1 on Perplexity?
A: You can access R1 on Perplexity’s Sonar API or HuggingFace.
Q: What are the differences between the censored and uncensored versions of R1?
A: The uncensored version of R1 has been post-trained to provide unbiased, accurate, and factual information, while the censored version has been trained to follow Chinese-government-enforced censorship.
