Yarr citizens of the high seas! The Stop Killing Games movement is still ongoing and we've recently had a second wind. It's within reach!

We're all lovers of media in here, and games currently have no safeguard that guarantees that they won't be locked down long after being released and abandoned. If crackers help us, they can still be played long into the future, but many times there isn't such a possibility, specially in multiplayer games.

This initiative seeks to change that by mainly:

• Disallowing planned obsolesce in paid video games. (Ex: By disallowing phone-home based DRM after the game reaches end of life. Like in Ubisoft's The Crew)
• Ensuring that paid multiplayer games can still be reasonably played long into the future. (Ex: By releasing relevant server hosting software)

If you didn't sign yet, there is only one month left. Tell your friends too.

Do you live in the EU?

• You can sign it here: https://eci.ec.europa.eu/045/public/#/screen/home

Do you live in the UK?

• You can also sign a different one here: https://petition.parliament.uk/petitions/702074/

Do you live elsewhere or would like to know more?

• https://www.stopkillinggames.com/

Disclaimer: Reminder post, sort of relevant since piracy movements have much to benefit from this initiative.

Have a fine day!

cross-posted from: https://feddit.uk/post/32177558

Please support the initiative of Stop Killing Games!

EU Petition

UK Petition

A real woman. A real story. A real triumph. Julia Roberts won an Academy Award® for Best Actress in her role as Erin Brockovich, a feisty young mother who fought for justice any way she knew how. Desperate for a job to support herself and her three children, she convinces attorney Ed Masry (Albert Finney) to hire her, and promptly stumbles upon a monumental law case against a giant corporation. Now, Erin's determined to take on this powerful adversary even though no law firm has dared to do it before. And while Ed doesn't want anything to do with the case, Erin won't take 'no' for an answer. So the two begin an incredible and sometimes hilarious fight that will bring a small town to its feet and a huge company to its knees.

Special Features and Technical Specs:

• DOLBY VISION/HDR PRESENTATION OF THE FILM
• Deleted Scenes
• Erin Brockovich: A Look at a Real-Life Experience
• Spotlight on Location: The Making of Erin Brockovich
• Theatrical Trailer
• 100 Years of Universal: Academy Award Winners
• 100 Years of Universal: The Lot
• Optional English SDH subtitles for the main feature

Pre-order: GRUV.com

Comments

Artist: Mac Star | pixiv | twitter | danbooru

I have always used Mullvad Leta as my search engine, but after changing smartphone and download Fennec, I ended up leaving DuckDuckgo that comes standard for now. However, after the last Fennec update (140.0.0), I was surprised to have Google as a standard search engine. What do you think of this? Someone with this same experience?

Andy Stitzer (Steve Carell) has gone 40 years without doing it. Now his pals are making it their mission to help him score…fast! Can he survive their hilariously bad advice? Will he land in the arms of the way-too experienced or the way-too-drunk? Or can he find true love where he least expects—from a gorgeous grandmother (Catherine Keener)? With both the Unrated and Theatrical version, The 40-year-Old Virgin on Blu-ray Hi-Def makes the pleasure last forever! Own this classic comedy for the first time on 4K Blu-ray.

Special Features and Technical Specs:

• DOLBY VISION/HDR PRESENTATION OF THE FILM
• DOLBY ATMOS AUDIO TRACK
• 20th Anniversary Discussion - Director Judd Apatow joins Steve Carell, Catherine Keener, Jane Lynch, Kat Dennings, and Gerry Bednob for a lively discussion at the Academy Museum of Motion Pictures.

Pre-order: GRUV.com

It's a snowy owl.

This paper introduces DiffuCoder, a 7B-scale open-source masked diffusion large language model (dLLM) specifically designed for code generation.

The research provides insights into how dLLMs generate content, distinguishing their decoding behavior from that of autoregressive (AR) models. Unlike AR models, dLLMs can intrinsically adjust their generation causality and increasing sampling temperature diversifies not just token choices but also their generation order, creating a rich search space for reinforcement learning (RL).

This flexibility allows dLLMs to be more non-autoregressive and generate tokens in a less sequential, more “human-like” code writing manner.

To leverage this diversity and improve performance, the paper proposes coupled-GRPO RL algorithm. This method utilizes a coupled-sampling scheme that constructs complementary mask noise during training to reduce the variance of token log-likelihood estimates while maintaining training efficiency.

Experimentally, coupled-GRPO significantly boosts DiffuCoder’s performance on code generation benchmarks, notably improving EvalPlus scores by 4.4% with training on only 21K samples. The research also shows that coupled-GRPO trained models experience a smaller performance drop when decoding steps are halved (resulting in a 2x speedup), indicating increased parallelism and reduced reliance on AR bias during decoding.

available at https://huggingface.co/apple/DiffuCoder-7B-cpGRPO