Karpathy on DeepSeek-OCR paper: Are pixels better inputs to LLMs than text?

Hacker NewsTuesday, October 21, 2025 at 5:43:16 PM
NeutralTechnology
Andrej Karpathy recently discussed the implications of the DeepSeek-OCR paper, which explores whether using pixels as inputs for large language models (LLMs) could be more effective than traditional text inputs. This conversation is significant as it could reshape how we think about data input in AI, potentially leading to advancements in machine learning and natural language processing.
— Curated by the World Pulse Now AI Editorial System

Was this article worth reading? Share it

Recommended Readings
General Motors will integrate AI into its cars, plus new hands-free assist
PositiveTechnology
General Motors is taking a bold step by integrating artificial intelligence into its vehicles, along with introducing a new hands-free assist feature. This move reflects GM's confidence in the potential of AI to enhance driving experiences and safety. As technology continues to evolve, the incorporation of AI could revolutionize how we interact with our cars, making them smarter and more intuitive. This development is significant not only for GM but for the entire automotive industry, as it sets a precedent for future innovations.
LLMs can get "brain rot"
NeutralTechnology
Recent discussions have emerged around the phenomenon of 'brain rot' in large language models (LLMs), highlighting potential issues in their performance and reliability. This matters because as LLMs become more integrated into various applications, understanding their limitations is crucial for developers and users alike.
The Karpathy Interview, 6 Months After AI 2027
NeutralTechnology
In a recent interview, Andrej Karpathy reflects on the developments in artificial intelligence since the predictions made for 2027. He discusses the advancements in technology and their implications for the future, providing insights that are both thought-provoking and relevant for anyone interested in the field. This conversation is significant as it highlights the rapid pace of change in AI and encourages a dialogue about its potential impact on society.
Neural audio codecs: how to get audio into LLMs
NeutralTechnology
The article discusses the emerging field of neural audio codecs and their potential applications in large language models (LLMs). As audio processing technology evolves, understanding how to effectively integrate audio into LLMs could enhance their capabilities, making them more versatile in handling various forms of data. This is significant as it opens up new avenues for innovation in AI and machine learning.
Getting DeepSeek-OCR working on an Nvidia Spark via brute force with Claude Code
PositiveTechnology
A recent article discusses the successful implementation of DeepSeek-OCR on Nvidia Spark using a brute force approach with Claude Code. This achievement is significant as it showcases the potential of combining advanced OCR technology with powerful data processing frameworks, which can enhance efficiency in data handling and analysis. The community's interest in this development highlights the ongoing innovation in tech, making it a noteworthy topic for those following advancements in machine learning and data processing.
Latest from Technology
All UK TV viewers getting a new free channel – and the return of a much-loved comedy favourite
PositiveTechnology
Great news for UK TV viewers! Warner Bros. Discovery is launching a new free channel, making one of its popular paid TV exclusives accessible to everyone. This move not only broadens entertainment options for the audience but also marks the return of a beloved comedy favorite, bringing joy and laughter back into homes. It's a fantastic opportunity for viewers to enjoy quality content without any cost, highlighting the importance of accessible entertainment in today's media landscape.
VST3 audio plugin format is now MIT
PositiveTechnology
The VST3 audio plugin format has officially transitioned to the MIT license, a move that is expected to enhance accessibility and innovation in audio software development. This change allows developers to freely use and modify the format, fostering a more collaborative environment in the audio community. It matters because it opens up new possibilities for creators and encourages the growth of diverse audio tools.
After Apple TV's latest price hike, I'm even more convinced that an Apple One subscription is the superior choice
PositiveTechnology
With Apple TV's recent price increase, many users are realizing that the Apple One subscription offers a more cost-effective solution for accessing multiple services. This shift highlights the value of bundling services, making it a smart choice for Apple users looking to save money while enjoying a range of features.
Nokia Profit Surges Past Estimates As AI, Cloud Demand Grows
PositiveTechnology
Nokia Oyj has reported a significant surge in its adjusted profit for the latest quarter, exceeding analyst expectations. This impressive growth is largely attributed to the rising demand for artificial intelligence and cloud services, highlighting the company's strong position in these rapidly evolving markets. As businesses increasingly turn to AI and cloud solutions, Nokia's performance reflects a broader trend in the tech industry, making it a key player to watch.
STMicro Revenue Forecast Misses as Chip Recovery Stalls
NegativeTechnology
STMicroelectronics has announced a fourth-quarter revenue forecast that falls short of analysts' expectations, indicating that the anticipated recovery in the chip industry may be losing momentum. This news is significant as it reflects broader challenges within the semiconductor market, which many had hoped would rebound after recent downturns.
Tokyo craft meets Zurich upcycling as Hender Scheme reimagines two FREITAG icons
PositiveTechnology
In an exciting collaboration, Hender Scheme has teamed up with FREITAG to create a limited-edition collection that beautifully merges Swiss tarp design with Japanese leather craftsmanship. This partnership not only showcases the unique artistry of both brands but also highlights the growing trend of sustainable fashion through upcycling. It's a significant step in promoting innovative design while respecting the environment, making it a noteworthy development in the fashion industry.