Antoo Member May 1, 2019 4,507 Full on spoilers for TLOU1, TLOU2, and both seasons of the show ahead I saw this clip on the TLOU subreddit making the rounds. Neil goes into the viability of the cure, and he says this: "Could the Fireflies make a cure? Our intent was that, yes, they could. Now, is our science a little shaky that now people are questioning it? Yeah, it was a little shaky and now people are questioning that. I can't say anything. All I can say is that our intent is that they would have made a cure. That makes it a more interesting philosophical question for what Joel does." Click to expand... Click to shrink... https://www.reddit.com/r/thelastofus/comments/1krqoz0/neil_debunks_the_cure_viability_debate_once_for/ I can't post the interview due to the interviewer, however, if you are a fan of TLOU, I would suggest maybe looking for articles/posts covering it or finding the interview yourself. Neil goes into A LOT. I'll bullet point some highlights. Show highlights: - Neil says Ellie and Dina's relationship was intentionally static in the game. The same approach wouldn't work for the show because shows need movement. - The series needs constant conflict/progression because story is everything in the medium. In games, you can have nothing of high importance going on for a while and still be invested due to interactivity. - He recognizes the divisiveness of the second season from game fans. He's appreciative of their love for the material and finds it cool how people see a game as standing shoulder-to-shoulder with a HBO show. He thinks it highlights how gaming has elevated as a medium. - Abby's motivation and the porch scene were moved up due to the reality that the second game needed multiple seasons to be fully adapted. Neil and Craig felt these elements wouldn't land if they kept the game's structure due to how long TV viewers would have to wait to get to them. There was a fear that the impact of these elements would have been lost due to people not remembering the previous season clearly enough to draw connections. - Craig is very intrigued by the idea of the prophet and wants to expand on who she is in the future. Game highlights: - There was originally a sequence planned for one of the flashbacks in TLOU2 where we would play through an infected attack on Jackson as Ellie alongside Joel. - There was no intent for the WLF/Seraphite conflict to serve as an allegory for the Israeli/Palestinian conflict. He took inspiration from the latter but he also took inspiration from other conflicts. Neil feels certain people online were cherry-picking statements to fit a narrative. He views the game conflict as a secular group clashing against a religious group. - He confirms he would be open to TLOU3 like he said in the documentary but wants to ensure he has the right idea for it that lives up to the series' pedigree - Neil's top priority right now is Intergalactic above all else. He claims it has the deepest gameplay they've ever done.  Last edited: Today at 2:57 AM Red Kong XIX Member Oct 11, 2020 13,276 Never understood why people thought they couldn't. That's the whole point of the ending. Joel being selfish, not willing to sacrifice someone he cares about for the greater good.  ConflictResolver Member Jan 1, 2024 4,907 Midgar I thought it was left vague in both the game and the show until the show's latest episode.   Philippo Developer Verified Oct 28, 2017 8,836 Heh, I liked the ambiguity of not knowing if there was a 100% success guarantee out of sacrificing Ellie.   Lotus One Winged Slayer Member Oct 25, 2017 124,081 I'm still saving her.   FTF Member Oct 28, 2017 33,203 New York Philippo said: Heh, I liked the ambiguity of not knowing if there was a 100% success guarantee out of sacrificing Ellie. Click to expand... Click to shrink... Agreed. It should be left unknown.   Bansai Teyvat Traveler Member Oct 28, 2017 14,176 Maaan Neil really needs to stop, feels like he's stripping away what's left of the nuance with those latest comments on the story. Then again, his story, his right I suppose, my headcannon remains strong and stubborn though. :P btw. interesting interview 🤔  Risev "This guy are sick" Member Oct 27, 2017 3,896 Red Kong XIX said: Never understood why people thought they couldn't. That's the whole point of the ending. Joel being selfish, not willing to sacrifice someone he cares about for the greater good. Click to expand... Click to shrink... I always got the impression the game wanted you to skip through the bullshit and just assume the cure would work, but it's still mediocre writing at best: - the lead doctor was a veterinarian. - the fireflies were desperate, lacking man power, and funds. - literally almost zero testing on Ellie before Just wanting to rip her brain out of her skull - literally zero attention given to the special circumstances that could have led to ellie being immune  The Quentulated Mox Corrupted by Vengeance Member Jun 10, 2022 6,565 hell yeah, next we should ask christopher nolan if the top was gonna fall down   Mauricio_Magus Member Oct 25, 2017 15,827 Death of the author applies here, I don't really care what he has to say if it's not in the original game/text. It's clearly supposed to be ambiguous and it's staying that way for me.  Axiom Member Oct 25, 2017 308 Neil knowing the answer isn't the same as Joel knowing the answer - the only guarantee was that Ellie was going to die.   FTF Member Oct 28, 2017 33,203 New York The Quentulated Mox said: hell yeah, next we should ask christopher nolan if the top was gonna fall down Click to expand... Click to shrink... lol   Threadmarks Clarification on cure New Index OP OP Antoo Member May 1, 2019 4,507 For further context, he also states that if you had a different takeaway on the ending scenario of the first game, that's fully valid as well. He's just clarifying his authorial intent while also acknowledging that a player/viewer may read the situation much differently. I think he finds the philosophical question of saving a loved one versus saving the world more interesting than the specifics of how they got to that point.   New Index harleyvwarren Member Oct 31, 2022 5,299 Illinois I always assumed there was a shot at a cure and that's what Joel denied humanity with his selfish, murderous behavior. There was no ambiguity about it for me playing the second game. It's just not subtle at all.   behOemoth Member Oct 27, 2017 6,687 ConflictResolver said: I thought it was left vague in both the game and the show until the show's latest episode. Click to expand... Click to shrink... I think his answer still keeps it vague, but emphasising that serious possibilities existed   Besiktas Member Sep 2, 2024 914 Why creators their own productruin years after a good product releases. Man just focus on making new stuff instead of clarifying theories.   Risev "This guy are sick" Member Oct 27, 2017 3,896 Antoo said: For further context, he also states that if you had a different takeaway on the ending scenario of the first game, that's fully valid as well. He's just clarifying his authorial intent while also acknowledging that a player/viewer may read the situation much differently. Click to expand... Click to shrink... With all due respect to Niel, but that'a bullshit lmao. He also said the exact same thing about Ellie forgiving Joel / knowing he killed the fireflies at the end of the first game just a month or so before the release of Part 2 which gives a clear answer and renders any other interpretation invalid   Kenzodielocke Member Oct 25, 2017 13,948 It's kind of moot if you make this so technichal Could they make it, could they deliver it, etc.  Lotus One Winged Slayer Member Oct 25, 2017 124,081 FTF said: Agreed. It should be left unknown. Click to expand... Click to shrink... The game came out over 12 years ago. The idea that a creator/author should just shut up and literally never comment on an ambiguous ending or complicated choice is so weird to me, especially when it's just his opinion at the end of the day.  Shoot Member Oct 25, 2017 5,909 Red Kong XIX said: Never understood why people thought they couldn't. That's the whole point of the ending. Joel being selfish, not willing to sacrifice someone he cares about for the greater good. Click to expand... Click to shrink... This. I was surprised to see people online saying they couldn't make a cure. It also obviously had no bearing on Joel's decision to massacre the hospital either. He just went back to doing what he used to do with Tommy for 20 years. Definitely makes Druckmann's recent comment about doing what Joel did sound sociopathic.  VAD Member Oct 28, 2017 6,099 Philippo said: Heh, I liked the ambiguity of not knowing if there was a 100% success guarantee out of sacrificing Ellie. Click to expand... Click to shrink... Yes, me too. I liked that the Fireflies were acting on the basis of hope rather than hard facts. Maybe Joel was right to save Ellie from pointless sacrifice. Maybe Ellie's savior complex was based on nothing and she was right to just live and enjoy life as it was.  Khanimus Avenger Oct 25, 2017 46,469 Greater Vancouver Woah Woah Woah... You mean Joel is a shitty self-serving asshole??!! Say it ain't so...  Zemoco Member Jan 12, 2021 2,621 Death of the author and all that, he really shouldn't confirm something like that. I suppose it's his right, but it hampers the discussion irrevocably. In either case, it does not make any sense on any level to kill the one girl with immunity milliseconds after making the deduction. Not to mention since the Fireflies are murderous, lying pricks anyway, it doesn't make any sense why Joel should believe them just because an omniscient entity (as far as the universe is concerned) confirmed it.  SirKai Member Dec 28, 2017 10,181 Washington Will never understand why people split hairs over this or claim the supposed "ambiguity" of the vaccine viability adds anything to the story. In BOTH games, every character that matters is confident in the possibility of the vaccine, and that is what is important. People so DESPERATELY want to be morally vindicated that siding with Joel is not just righteous, but also rational even pursuit of a vaccine, even though the most passing glance interpretation of the ending is OBVIOUSLY written to not satisfy that perspective. It's a trolley problem, and the trolley problem is what makes the ending, and Joel's decision, interesting. If it's not actually a trolley problem, the ending and the story lose a lot of their depth and impact.   Last edited: Today at 3:02 AM Risev "This guy are sick" Member Oct 27, 2017 3,896 Khanimus said: Woah Woah Woah... You mean Joel is a shitty self-serving asshole??!! Say it ain't so... Click to expand... Click to shrink... if only he picked up the phone when Neil was calling to tell him the cure works...   SCUMMbag Prophet of Truth - Chicken Chaser Member Oct 25, 2017 7,199 Red Kong XIX said: Never understood why people thought they couldn't. That's the whole point of the ending. Joel being selfish, not willing to sacrifice someone he cares about for the greater good. Click to expand... Click to shrink... This. TLOU isn't a written masterpiece so there's some holes but the intention of those scenes were pretty clear. A lot of the ambiguity comes from things like "they did no testing" and "they decided this far too quick" which are just leaps you'd make to keep the pacing of your game.  Milk Prophet of Truth Avenger Oct 25, 2017 4,292 No shit. People trying to "um achually ☝️🤓" their way out of Joel's choice ruins the entire point of the ending in the first place. If there's no realistic way to create and disperse a vaccine then there's no choice in the first place. At the same time, I get it. Obviously you want to realistically analyze parts of a story you're experiencing. But story intent still applies, in this instance, it's literally just better to accept the Fireflies at their word and assume a cure would work.  Kalentan Member Oct 25, 2017 50,658 I feel like the cure having been likely possible is far more interesting because it means Joel's decision has more around it. Cause yeah, his decision to kill them all means a lot more than if the cure was never possible and they were just a bunch idiots cause then Joel was 100% in the right to stop them.   Glio Member Oct 27, 2017 27,779 Spain Red Kong XIX said: Never understood why people thought they couldn't. That's the whole point of the ending. Joel being selfish, not willing to sacrifice someone he cares about for the greater good. Click to expand... Click to shrink... Because the science behind it was pretty stupid, tbh. But you're right, from a dramatic point of view, it needs to be that way.  bob1001 ▲ Legend ▲ Member May 7, 2020 2,109 If they can make a cure: The ending is a moral dilemma, where Joel is willing to sacrifice humanity to save the person he loves. If they can't make a cure: Joel is saving a child from child murderers. I never understood why anyone would prefer a Mario saves Peach style ending instead of the actual interesting ending we got. When you question their ability to make a cure you are arguing the ending is worse than it is.  Kalentan Member Oct 25, 2017 50,658 bob1001 said: If they can make a cure: The ending is a moral dilemma, where Joel is willing to sacrifice humanity to save the person he loves. If they can't make a cure: Joel is saving a child from child murderers. I never understood why anyone would prefer a Mario saves Peach style ending instead of the actual interesting ending we got. When you question their ability to make a cure you are arguing the ending is worse than it is. Click to expand... Click to shrink... Exactly.  Risev "This guy are sick" Member Oct 27, 2017 3,896 Milk said: No shit. People trying to "um achually ☝️🤓" their way out of Joel's choice ruins the entire point of the ending in the first place. If there's no realistic way to create and disperse a vaccine then there's no choice in the first place. At the same time, I get it. Obviously you want to realistically analyze parts of a story you're experiencing. But story intent still applies, in this instance, it's literally just better to accept the Fireflies at their word and assume a cure would work. Click to expand... Click to shrink... I'd say this is Neil's biggest shortcoming as a writer and is worth criticizing: wanting you to make leaps and just skip through some plot holes for the service of the plot. Uncharted 4 also contains an extremely glaring plot hole that you have to gloss over to enjoy the story  Kenzodielocke Member Oct 25, 2017 13,948 The arguments about if the cure would have worked come usually from people who want to justify hie actions. The justification there actually is, love. Edit: "They didn't even ask her" point is also kind of moot because how often we heard from Ellies mouth that she would have done it.  mbpm Member Oct 25, 2017 29,491 I thought it was more interesting leaving it unknown   psynergyadept Shinra Employee Member Oct 26, 2017 19,044 It was always the case; people just obscured things to make themselves feel better about Joel's decision. The whole point of the games ending was dealing with the "Save many by the cost of one/few" trope we've seen before.  EatChildren Wonder from Down Under Member Oct 27, 2017 7,595 Boooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo The story is infinitely more interesting and thought provoking, particularly in regards to character motivations, actions, and subsequent consequences, when the climax of the first game is viewed through the lens of the unknown; that people and groups make decisions and take actions, sometimes decisively and recklessly, without knowing for sure what the totality of consequence will be, or being fundamentally unable to know if the risks taken are worth the cost. I don't even care about the science behind it (which is dumb). Knowing the cure would/wouldn't work sucks shit and is a boring lame framing of the narrative. Not knowing adds a hefty ambiguity and weight to the choices made. People making decisions, or committing to causes, without ever fully knowing for sure how subsequent events will transpire, is literally how life works is is the ultimate fuel and weight behind our personal journey through guilt and accountability.  SirKai Member Dec 28, 2017 10,181 Washington Risev said: I'd say this is Neil's biggest shortcoming as a writer and is worth criticizing: wanting you to make leaps and just skip through some plot holes for the service of the plot. Uncharted 4 also contains an extremely glaring plot hole that you have to gloss over to enjoy the story Click to expand... Click to shrink... I'm not a Neil defender, but that's not a shortcoming of a writer; that's just an extremely basic aspect of storytelling to motivate the drama and create interesting circumstances. Pretty much no long-form story that depends on exceptional scenarios is going to be free of contrivance or convenience. Some stories obviously take it too far and it can make the narrative feel too arbitrarily authored and unnatural, and every individual person has their own threshold for how far they can suspend their disbelief, but the willingness to trust writers by suspending our disbelief is what makes stories strong and effective, and a writer depending on their audience to be able to do that, at least to some extent, is not a weakness.  Altairre Member Oct 25, 2017 5,211 Risev said: With all due respect to Niel, but that'a bullshit lmao. He also said the exact same thing about Ellie forgiving Joel / knowing he killed the fireflies at the end of the first game just a month or so before the release of Part 2 which gives a clear answer and renders any other interpretation invalid Click to expand... Click to shrink... It doesn't really matter what he says because within the text there clearly is ambiguity and there is basically no way to retcon that away. Considering their situation, what the audio logs say and the state of the world it's definitely a long shot but it's also THE long shot. Risev said: I'd say this is Neil's biggest shortcoming as a writer and is worth criticizing: wanting you to make leaps and just skip through some plot holes for the service of the plot. Uncharted 4 also contains an extremely glaring plot hole that you have to gloss over to enjoy the story Click to expand... Click to shrink... I'm not sure that the situation in LoU qualifies as a plot hole tbh.   Jubern Member Oct 25, 2017 1,597 Mauricio_Magus said: Death of the author applies here, I don't really care what he has to say if it's not in the original game/text. It's clearly supposed to be ambiguous and it's staying that way for me. Click to expand... Click to shrink... Exactly where I stand. Why he would want to clarify/comment on this so long after the fact leaves me dumbfounded.   FTF Member Oct 28, 2017 33,203 New York EatChildren said: Boooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo The story is infinitely more interesting and thought provoking, particularly in regards to character motivations, actions, and subsequent consequences, when the climax of the first game is viewed through the lens of the unknown; that people and groups make decisions and take actions, sometimes decisively and recklessly, without knowing for sure what the totality of consequence will be, or being fundamentally unable to know if the risks taken are worth the cost. I don't even care about the science behind it (which is dumb). Knowing the cure would/wouldn't work sucks shit and is a boring lame framing of the narrative. Not knowing adds a hefty ambiguity and weight to the choices made. People making decisions, or committing to causes, without ever fully knowing for sure how subsequent events will transpire, is literally how life works is is the ultimate fuel and weight behind our personal journey through guilt and accountability. Click to expand... Click to shrink... Yeah, this is what I meant and said sooo much better lol.   Cantaim Member Oct 25, 2017 35,072 The Stussining I always thought it would have worked as well. The entire game is building up to Joel facing the Trolley problem but with Ellie on the track. I don't think it really has any teeth if you just say killing Ellie doesn't do anything.   Crossing Eden Member Oct 26, 2017 58,520 Kenzodielocke said: It's kind of moot if you make this so technichal Click to expand... Click to shrink... You aren't supposed to because it's not real life Cantaim said: I always thought it would have worked as well. The entire game is building up to Joel facing the Trolley problem but with Ellie on the track. I don't think it really has any teeth if you just say killing Ellie doesn't do anything. Click to expand... Click to shrink... It's funny because literally nothing about the story ever implies that the cure wouldn't work. For every single thing that gets addressed in a "grounded" way that particular tidbit has never been more than people using it as an excuse to justify/lighten the severity of Joel's actions. "Eh does it really matter that he shot up the hospital at the end of the day? Not like the cure would've worked anyways. I, the player/Joel did nothing wrong."  TacoSupreme Member Jul 26, 2019 2,092 SirKai said: I'm not a Neil defender, but that's not a shortcoming of a writer; that's just an extremely basic aspect of storytelling to motivate the drama and create interesting circumstances. Pretty much no long-form story that depends on exceptional scenarios is going to be free of contrivance or convenience. Some stories obviously take it too far and it can make the narrative feel too arbitrarily authored and unnatural, and every individual person has their own threshold for how far they can suspend their disbelief, but the willingness to trust writers by suspending our disbelief is what makes stories strong and effective, and a writer depending on their audience to be able to do that, at least to some extent, is not a weakness. Click to expand... Click to shrink... Yeah, I'm going to have to disagree with this. It's not about suspending disbelief, it's about the game deliberately making the whole situation with the Fireflies seem sketchy. I genuinely spit out my drink and started laughing when it was revealed that they were going to instantly take the precious immune person and dissect her almost immediately after getting their hands on her. This goes beyond contrivance or convenience and into the realm of deliberately misleading the player into thinking there's ambiguity. All it would have taken is something denoting the passage of time prior to wanting to scoop out her brain and it would have been fine. It's not about suspending disbelief, it's about making unneeded decisions that mislead the player.  GMM Member Oct 27, 2017 5,797 If they really would have made a cure or not kinda doesn't matter in the context of the story, it's about how Joel put himself over the needs of the world. Everyone out there trying to survive would agree that Ellie's sacrifice would be worth restoring some semblance of peace to the world even if it wasn't a safe bet, Ellie herself would have wanted to save the world but Joel made that choice for her. It's all about Joel being the selfish person he is, he chose himself over everyone else.  Terbinator Member Oct 29, 2017 13,379 Honestly don't think the cure being viable or not matters at all. Joel makes the decision to save Ellie to save his second daughter. It's really not that deep and you also have no agency over this in the game. Whether that's the moral thing to do on the promise of a cure is an open question.  MrKlaw Member Oct 25, 2017 36,871 Reality doesn't matter (I disagree - they are barely properly staffed, they've never done this before or seen it before so its a hail mary at best etc etc all the discussion) but from a story perspective it makes sense that at least Joel believes its possible to have a cure - it makes the narrative and his reaction stronger, and the 'my life could have meant something' from Ellie's side stronger to create that necessary tension. But I don't like it.  Sinah Member Jun 2, 2022 1,254 I mean yeah so? Honestly personally i don't think it even really matters at that point world was already in a absolute shit state with literal cannibal and murderers everywhere and the infected can not be cured so you still have millions of monsters running around everywhere ripping ppl apart. There was nothing worth saving even if they did manage to make a cure and actually distribute it which is definitely the bigger problem here considering the state the Fireflys where in and the logistics involved.  Last edited: Today at 3:24 AM Vyse One Winged Slayer Member Oct 25, 2017 1,641 Joel might have pressed a 100% cure button that kills his daughter but even a 1% chance it was a hail mary by sketchy people guaranteed the slaughter.   Agni Kai Member Nov 2, 2017 10,001 None of you (and I do mean none of you) would let your child die to save other people. This new piece of information changes nothing. Joel could've never know.  Crossing Eden Member Oct 26, 2017 58,520 Agni Kai said: None of you (and I do mean none of you) would let your child die to save other people. This new piece of information changes nothing. Joel could've never know. Click to expand... Click to shrink... He never once doubted that it would work though.  

The Cannes film festival isn’t typically associated with video games, but this year it’s playing host to an unusual collaboration. Lili is a co-production between the New York-based game studio iNK Stories (creator of 1979 Revolution: Black Friday, about a photojournalist in Iran) and the Royal Shakespeare Company, and it’s been turning heads with its eye-catching translocation of Macbeth to modern-day Iran.“It’s been such an incredible coup to have it as the first video game experience at Cannes,” says iNK Stories co-founder Vassiliki Khonsari. “People have gone in saying, I’m not familiar playing games, so I may just try it out for five minutes. […] But then once they’re in, there is this growing sense of empowerment that people from the film world are feeling.”The Cannes festival’s Immersive Competition began in 2024, although the lineup doesn’t usually feature traditional video games. “VR films and projection mapping is the thrust of it,” says iNK Stories’ other co-founder, Vassiliki’s husband Navid Khonsari. But Lili weaves live-action footage with video game mechanics in a similar way to a game such as Telling Lies or Immortality. Its lead, Zar Amir Ebrahimi, won best actress at Cannes three years ago.Lili focuses on the story of Lady Macbeth, here cast as the ambitious wife of an upwardly mobile officer in the Basij (a paramilitary volunteer militia within the Islamic Revolutionary Guard in Iran). As in the play, she plots a murder to secure her husband’s rise. “I think that the narrative of Lady Macbeth is that she’s manipulative, and that’s exactly what got us interested,” says Navid.“The social limitations based on her gender forced her to try to attain whatever leadership role she can,” he continues. “If she was a man, she would have been one of the greatest kings that country would have ever experienced, but because she was a woman she had to work within the structure that was there for her. And I think this is the same thing that we have with our Lili character: because of society, she’s limited to how high she can rise.”The player is cast as a member of the Hecate Web, a group of hackers who stand in for Macbeth’s witches, and you begin by accessing Lili’s phone and computer, watching her on CCTV cameras inside her home. The oppression of surveillance and censorship is a key theme. At one point Lili tries to access a YouTube makeup tutorial, only to be blocked by the state’s firewall. The fact that the player takes on an uncomfortable voyeuristic role is intentional. “We’re all part of the problem when it comes to surveillance, when it comes to looking at each other, spying on each other,” says Navid.‘She’s suffocating in the various layers of masks that she must put on’ … Zar Amir Ebrahimi in Lili. Photograph: Ellie SmithYou can bypass it to give her access, however, prompting a memorable scene in which she wraps her head in clingfilm before applying eyeliner and lipstick. “That is her ritual, applying makeup in a world that she cannot do that [in], without her husband knowing or anyone else knowing,” says Vassiliki. “We love it as this sort of allegory, that she’s suffocating in the world and in the various layers of masks that she must put on herself … our Lady Macbeth, our Lili, has an awakening, and all of those tools that we’ve used to hack her become tools that we actually give her to help her take the system down from within.”Lili is scheduled for release in late 2026, and the RSC’s Sarah Ellis says that it might well be turned into a play at some point in the future. Navid says there are already plans for a film version, using some of the same footage shot for the game – a reminder that the boundaries between the worlds of gaming, movies and theatre are increasingly porous. This is unlikely to be the last time the RSC is involved in games.“I was always interested in games, and the convergence of games and theatre in particular,” says Ellis, who originally approached iNK with the idea for this collaboration. “We’ve worked with some of the best Shakespeare scholars … Professor Emma Smith from Oxford has been an absolute keystone in the dramaturgy of this work,” says Ellis. Smith has said that if Shakespeare were alive today, he’d be writing for games, and Ellis agrees: “He was an innovator.”

Addressing Architectural Trade-offs in Language Models As language models scale, balancing expressivity, efficiency, and adaptability becomes increasingly challenging. Transformer architectures dominate due to their strong performance across a wide range of tasks, but they are computationally expensive—particularly for long-context scenarios—due to the quadratic complexity of self-attention. On the other hand, Structured State Space Models (SSMs) offer improved efficiency and linear scaling, yet often lack the nuanced sequence modeling required for complex language understanding. A combined architecture that leverages the strengths of both approaches is needed to support diverse applications across environments. Introducing Falcon-H1: A Hybrid Architecture The Falcon-H1 series, released by the Technology Innovation Institute (TII), introduces a hybrid family of language models that combine Transformer attention mechanisms with Mamba2-based SSM components. This architecture is designed to improve computational efficiency while maintaining competitive performance across tasks requiring deep contextual understanding. Falcon-H1 covers a wide parameter range—from 0.5B to 34B—catering to use cases from resource-constrained deployments to large-scale distributed inference. The design aims to address common bottlenecks in LLM deployment: memory efficiency, scalability, multilingual support, and the ability to handle extended input sequences. Source: https://falcon-lm.github.io/blog/falcon-h1/ Architectural Details and Design Objectives Falcon-H1 adopts a parallel structure where attention heads and Mamba2 SSMs operate side by side. This design allows each mechanism to independently contribute to sequence modeling: attention heads specialize in capturing token-level dependencies, while SSM components support efficient long-range information retention. The series supports a context length of up to 256K tokens, which is particularly useful for applications in document summarization, retrieval-augmented generation, and multi-turn dialogue systems. Model training incorporates a customized microparameterization (μP) recipe and optimized data pipelines, allowing for stable and efficient training across model sizes. The models are trained with a focus on multilingual capabilities. The architecture is natively equipped to handle 18 languages, with coverage including English, Chinese, Arabic, Hindi, French, and others. The framework is extensible to over 100 languages, supporting localization and region-specific model adaptation. Empirical Results and Comparative Evaluation Despite relatively modest parameter counts, Falcon-H1 models demonstrate strong empirical performance: Falcon-H1-0.5B achieves results comparable to 7B-parameter models released in 2024. Falcon-H1-1.5B-Deep performs on par with leading 7B to 10B Transformer models. Falcon-H1-34B matches or exceeds the performance of models such as Qwen3-32B, Llama4-Scout-17B/109B, and Gemma3-27B across several benchmarks. Evaluations emphasize both general-purpose language understanding and multilingual benchmarks. Notably, the models achieve strong performance across both high-resource and low-resource languages without requiring excessive fine-tuning or additional adaptation layers. Source: https://falcon-lm.github.io/blog/falcon-h1/ Deployment and inference are supported through integration with open-source tools such as Hugging Face Transformers. FlashAttention-2 compatibility further reduces memory usage during inference, offering an attractive efficiency-performance balance for enterprise use. Conclusion Falcon-H1 represents a methodical effort to refine language model architecture by integrating complementary mechanisms—attention and SSMs—within a unified framework. By doing so, it addresses key limitations in both long-context processing and scaling efficiency. The model family provides a range of options for practitioners, from lightweight variants suitable for edge deployment to high-capacity configurations for server-side applications. Through its multilingual coverage, long-context capabilities, and architectural flexibility, Falcon-H1 offers a technically sound foundation for research and production use cases that demand performance without compromising on efficiency or accessibility. Check out the Official Release, Models on Hugging Face and GitHub Page. All credit for this research goes to the researchers of this project. Also, feel free to follow us on Twitter and don’t forget to join our 95k+ ML SubReddit and Subscribe to our Newsletter. Asif RazzaqWebsite |  + postsBioAsif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.Asif Razzaqhttps://www.marktechpost.com/author/6flvq/Google DeepMind Releases Gemma 3n: A Compact, High-Efficiency Multimodal AI Model for Real-Time On-Device UseAsif Razzaqhttps://www.marktechpost.com/author/6flvq/A Step-by-Step Implementation Tutorial for Building Modular AI Workflows Using Anthropic’s Claude Sonnet 3.7 through API and LangGraphAsif Razzaqhttps://www.marktechpost.com/author/6flvq/Meta Researchers Introduced J1: A Reinforcement Learning Framework That Trains Language Models to Judge With Reasoned Consistency and Minimal DataAsif Razzaqhttps://www.marktechpost.com/author/6flvq/Sampling Without Data is Now Scalable: Meta AI Releases Adjoint Sampling for Reward-Driven Generative Modeling

When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works. Vercel releases first AI model for v0, now in beta David Uzondu Neowin · May 22, 2025 02:18 EDT Google recently showered us with AI goodies, including Gemma 3n, an AI model that's designed to run on low-end devices, like smartphones. Now, Vercel has stepped further into the ring with its own generative UI system, v0, by releasing its very first dedicated model. If you do not know what v0 is, it is a sort of competitor to tools like the recently announced Google Stitch, which also aims to let you describe a user interface and have AI generate the design. The tool first saw the light of day back in 2023 as an invite-only beta, promising to turn natural language into front-end code. The newly available model is dubbed v0-1.0-md, and Vercel states it is specifically designed for building modern web applications. This multimodal model supports both text and image inputs, offers a 128,000-token context window with a 32,000-token output limit, and is priced at $3 per million input tokens and $15 per million output tokens. It offers features like 'auto-fix' for common coding blunders and 'quick edit' for streaming inline changes as they are generated. Crucially, v0-1.0-md uses an OpenAI-compatible API, meaning you can plug it into existing tools like Cursor, Codex, or your own custom applications that already speak OpenAI's language, including Vercel's own AI SDK. It even supports function and tool calls, and promises low-latency streaming responses. Developers can poke around with this new model in the Vercel AI Playground to see how it handles different prompts. Currently, access to the v0 API, and thus the v0-1.0-md model, is in beta, and you will need a Premium or Team plan on Vercel with usage-based billing enabled. To get started, you would grab an API key from v0.dev and then send requests to its POST api.v0.dev/v1/chat/completions endpoint, authenticating with a bearer token. While there are daily message limits around 200 messages and context size constraints that mirror its advertised capabilities, Vercel notes you can request higher limits if you hit those ceilings. If you want to dig into the details or see how to set it up, the official v0 docs on Vercel's site have everything you need, including examples. Tags Report a problem with article Follow @NeowinFeed

Researchers are reimagining how models operate as demand skyrockets for faster, smarter, and more private AI on phones, tablets, and laptops. The next generation of AI isn’t just lighter and faster; it’s local. By embedding intelligence directly into devices, developers are unlocking near-instant responsiveness, slashing memory demands, and putting privacy back into users’ hands. With mobile hardware rapidly advancing, the race is on to build compact, lightning-fast models that are intelligent enough to redefine everyday digital experiences. A major concern is delivering high-quality, multimodal intelligence within the constrained environments of mobile devices. Unlike cloud-based systems that have access to extensive computational power, on-device models must perform under strict RAM and processing limits. Multimodal AI, capable of interpreting text, images, audio, and video, typically requires large models, which most mobile devices cannot handle efficiently. Also, cloud dependency introduces latency and privacy concerns, making it essential to design models that can run locally without sacrificing performance. Earlier models like Gemma 3 and Gemma 3 QAT attempted to bridge this gap by reducing size while maintaining performance. Designed for use on cloud or desktop GPUs, they significantly improved model efficiency. However, these models still required robust hardware and could not fully overcome mobile platforms’ memory and responsiveness constraints. Despite supporting advanced functions, they often involved compromises limiting their real-time smartphone usability. Researchers from Google and Google DeepMind introduced Gemma 3n. The architecture behind Gemma 3n has been optimized for mobile-first deployment, targeting performance across Android and Chrome platforms. It also forms the underlying basis for the next version of Gemini Nano. The innovation represents a significant leap forward by supporting multimodal AI functionalities with a much lower memory footprint while maintaining real-time response capabilities. This marks the first open model built on this shared infrastructure and is made available to developers in preview, allowing immediate experimentation. The core innovation in Gemma 3n is the application of Per-Layer Embeddings (PLE), a method that drastically reduces RAM usage. While the raw model sizes include 5 billion and 8 billion parameters, they behave with memory footprints equivalent to 2 billion and 4 billion parameter models. The dynamic memory consumption is just 2GB for the 5B model and 3GB for the 8B version. Also, it uses a nested model configuration where a 4B active memory footprint model includes a 2B submodel trained through a technique known as MatFormer. This allows developers to dynamically switch performance modes without loading separate models. Further advancements include KVC sharing and activation quantization, which reduce latency and increase response speed. For example, response time on mobile improved by 1.5x compared to Gemma 3 4B while maintaining better output quality. The performance metrics achieved by Gemma 3n reinforce its suitability for mobile deployment. It excels in automatic speech recognition and translation, allowing seamless speech conversion to translated text. On multilingual benchmarks like WMT24++ (ChrF), it scores 50.1%, highlighting its strength in Japanese, German, Korean, Spanish, and French. Its mix’n’match capability allows the creation of submodels optimized for various quality and latency combinations, offering developers further customization. The architecture supports interleaved inputs from different modalities, text, audio, images, and video, allowing more natural and context-rich interactions. It also performs offline, ensuring privacy and reliability even without network connectivity. Use cases include live visual and auditory feedback, context-aware content generation, and advanced voice-based applications. Several Key Takeaways from the Research on Gemma 3n include: Built using collaboration between Google, DeepMind, Qualcomm, MediaTek, and Samsung System LSI. Designed for mobile-first deployment. Raw model size of 5B and 8B parameters, with operational footprints of 2GB and 3GB, respectively, using Per-Layer Embeddings (PLE). 1.5x faster response on mobile vs Gemma 3 4B. Multilingual benchmark score of 50.1% on WMT24++ (ChrF). Accepts and understands audio, text, image, and video, enabling complex multimodal processing and interleaved inputs. Supports dynamic trade-offs using MatFormer training with nested submodels and mix’n’match capabilities. Operates without an internet connection, ensuring privacy and reliability. Preview is available via Google AI Studio and Google AI Edge, with text and image processing capabilities. In conclusion, this innovation provides a clear pathway for making high-performance AI portable and private. By tackling RAM constraints through innovative architecture and enhancing multilingual and multimodal capabilities, researchers offer a viable solution for bringing sophisticated AI directly into everyday devices. The flexible submodel switching, offline readiness, and fast response time mark a comprehensive approach to mobile-first AI. The research addresses the balance of computational efficiency, user privacy, and dynamic responsiveness. The result is a system capable of delivering real-time AI experiences without sacrificing capability or versatility, fundamentally expanding what users can expect from on-device intelligence. Check out the Technical details and Try it here. All credit for this research goes to the researchers of this project. Also, feel free to follow us on Twitter and don’t forget to join our 95k+ ML SubReddit and Subscribe to our Newsletter. Asif RazzaqWebsite |  + postsBioAsif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.Asif Razzaqhttps://www.marktechpost.com/author/6flvq/A Step-by-Step Implementation Tutorial for Building Modular AI Workflows Using Anthropic’s Claude Sonnet 3.7 through API and LangGraphAsif Razzaqhttps://www.marktechpost.com/author/6flvq/Meta Researchers Introduced J1: A Reinforcement Learning Framework That Trains Language Models to Judge With Reasoned Consistency and Minimal DataAsif Razzaqhttps://www.marktechpost.com/author/6flvq/Sampling Without Data is Now Scalable: Meta AI Releases Adjoint Sampling for Reward-Driven Generative ModelingAsif Razzaqhttps://www.marktechpost.com/author/6flvq/Google AI Releases MedGemma: An Open Suite of Models Trained for Performance on Medical Text and Image Comprehension

Emma Stone and Dave McCary are offloading their Texas compound for $26.5 million. The La La Land star and her writer-producer husband bought the property in Tarrytown, one of the oldest enclaves in the Lone Star state. The suburb is located minutes away from downtown Austin. After quietly tying the knot in 2020 and welcoming their first child the following year, the family of three moved from California to the historic West Austin estate, which holds the title of most expensive real estate purchase in their portfolio. The pair has decided to part ways with the abode after investing years into restoring the property.The home was completely updated by the award-winning architecture firm Cuppett Kilpatrick. For the project, the Austin-based firm took inspiration from “the gracious nature of the surrounding community,” per the official listing. The four-bedroom, three-and-a-half bathroom pad is tucked away on an oak-shaded street. The main house, constructed with Georgia-style masonry, spans 10,000 square feet. A detached garage is outfitted with a screening room and an entertainment space above. Elsewhere on the 1.24-acre plot is a two-bedroom guesthouse, a screened-in breezeway looking out onto the pool, and a garden reminiscent of a countryside retreat.A landscaped garden area in the yard of the home. Photo: Travis Wayne Baker/Twist ToursStone and McCary tapped local architects Koch McIntyre Construction to reimagine the interiors of the traditional-style house. Quaint features like herringbone oak, brick floors, and creamy marble used for fireplace surrounds and counters are incorporated in every room. The estate boasts five spacious living areas and two dining rooms flooded with natural light. Other amenities include arched doorways, leaded-glass windows, a sunlit solarium, a wood-paneled library, and a playroom. Construction on the property is not yet finished, though it is slated for completion in the summer. The listing is currently the most expensive in the city.

In this tutorial, we provide a practical guide for implementing LangGraph, a streamlined, graph-based AI orchestration framework, integrated seamlessly with Anthropic’s Claude API. Through detailed, executable code optimized for Google Colab, developers learn how to build and visualize AI workflows as interconnected nodes performing distinct tasks, such as generating concise answers, critically analyzing responses, and automatically composing technical blog content. The compact implementation highlights LangGraph’s intuitive node-graph architecture. It can manage complex sequences of Claude-powered natural language tasks, from basic question-answering scenarios to advanced content generation pipelines. from getpass import getpass import os anthropic_key = getpass("Enter your Anthropic API key: ") os.environ["ANTHROPIC_API_KEY"] = anthropic_key print("Key set:", "ANTHROPIC_API_KEY" in os.environ) We securely prompt users to input their Anthropic API key using Python’s getpass module, ensuring sensitive data isn’t displayed. It then sets this key as an environment variable (ANTHROPIC_API_KEY) and confirms successful storage. import os import json import requests from typing import Dict, List, Any, Callable, Optional, Union from dataclasses import dataclass, field import networkx as nx import matplotlib.pyplot as plt from IPython.display import display, HTML, clear_output We import essential libraries for building and visualizing structured AI workflows. It includes modules for handling data (json, requests, dataclasses), graph creation and visualization (networkx, matplotlib), interactive notebook display (IPython.display), and type annotations (typing) for clarity and maintainability. try: import anthropic except ImportError: print("Installing anthropic package...") !pip install -q anthropic import anthropic from anthropic import Anthropic We ensure the anthropic Python package is available for use. It attempts to import the module and, if not found, automatically installs it using pip in a Google Colab environment. After installation, it imports the Anthropic client, essential for interacting with Claude models via the Anthropic API. 4o @dataclass class NodeConfig: name: str function: Callable inputs: List[str] = field(default_factory=list) outputs: List[str] = field(default_factory=list) config: Dict[str, Any] = field(default_factory=dict) This NodeConfig data class defines the structure of each node in the LangGraph workflow. Each node has a name, an executable function, optional inputs and outputs, and an optional config dictionary to store additional parameters. This setup allows for modular, reusable node definitions for graph-based AI tasks. class LangGraph: def __init__(self, api_key: Optional[str] = None): self.api_key = api_key or os.environ.get("ANTHROPIC_API_KEY") if not self.api_key: from google.colab import userdata try: self.api_key = userdata.get('ANTHROPIC_API_KEY') if not self.api_key: raise ValueError("No API key found") except: print("No Anthropic API key found in environment variables or Colab secrets.") self.api_key = input("Please enter your Anthropic API key: ") if not self.api_key: raise ValueError("Please provide an Anthropic API key") self.client = Anthropic(api_key=self.api_key) self.graph = nx.DiGraph() self.nodes = {} self.state = {} def add_node(self, node_config: NodeConfig): self.nodes[node_config.name] = node_config self.graph.add_node(node_config.name) for input_node in node_config.inputs: if input_node in self.nodes: self.graph.add_edge(input_node, node_config.name) return self def claude_node(self, name: str, prompt_template: str, model: str = "claude-3-7-sonnet-20250219", inputs: List[str] = None, outputs: List[str] = None, system_prompt: str = None): """Convenience method to create a Claude API node""" inputs = inputs or [] outputs = outputs or [name + "_response"] def claude_fn(state, **kwargs): prompt = prompt_template for k, v in state.items(): if isinstance(v, str): prompt = prompt.replace(f"{{{k}}}", v) message_params = { "model": model, "max_tokens": 1000, "messages": [{"role": "user", "content": prompt}] } if system_prompt: message_params["system"] = system_prompt response = self.client.messages.create(**message_params) return response.content[0].text node_config = NodeConfig( name=name, function=claude_fn, inputs=inputs, outputs=outputs, config={"model": model, "prompt_template": prompt_template} ) return self.add_node(node_config) def transform_node(self, name: str, transform_fn: Callable, inputs: List[str] = None, outputs: List[str] = None): """Add a data transformation node""" inputs = inputs or [] outputs = outputs or [name + "_output"] node_config = NodeConfig( name=name, function=transform_fn, inputs=inputs, outputs=outputs ) return self.add_node(node_config) def visualize(self): """Visualize the graph""" plt.figure(figsize=(10, 6)) pos = nx.spring_layout(self.graph) nx.draw(self.graph, pos, with_labels=True, node_color="lightblue", node_size=1500, arrowsize=20, font_size=10) plt.title("LangGraph Flow") plt.tight_layout() plt.show() print("\nGraph Structure:") for node in self.graph.nodes(): successors = list(self.graph.successors(node)) if successors: print(f" {node} → {', '.join(successors)}") else: print(f" {node} (endpoint)") print() def _get_execution_order(self): """Determine execution order based on dependencies""" try: return list(nx.topological_sort(self.graph)) except nx.NetworkXUnfeasible: raise ValueError("Graph contains a cycle") def execute(self, initial_state: Dict[str, Any] = None): """Execute the graph in topological order""" self.state = initial_state or {} execution_order = self._get_execution_order() print("Executing LangGraph flow:") for node_name in execution_order: print(f"- Running node: {node_name}") node = self.nodes[node_name] inputs = {k: self.state.get(k) for k in node.inputs if k in self.state} result = node.function(self.state, **inputs) if len(node.outputs) == 1: self.state[node.outputs[0]] = result elif isinstance(result, (list, tuple)) and len(result) == len(node.outputs): for i, output_name in enumerate(node.outputs): self.state[output_name] = result[i] print("Execution completed!") return self.state def run_example(question="What are the key benefits of using a graph-based architecture for AI workflows?"): """Run an example LangGraph flow with a predefined question""" print(f"Running example with question: '{question}'") graph = LangGraph() def question_provider(state, **kwargs): return question graph.transform_node( name="question_provider", transform_fn=question_provider, outputs=["user_question"] ) graph.claude_node( name="question_answerer", prompt_template="Answer this question clearly and concisely: {user_question}", inputs=["user_question"], outputs=["answer"], system_prompt="You are a helpful AI assistant." ) graph.claude_node( name="answer_analyzer", prompt_template="Analyze if this answer addresses the question well: Question: {user_question}\nAnswer: {answer}", inputs=["user_question", "answer"], outputs=["analysis"], system_prompt="You are a critical evaluator. Be brief but thorough." ) graph.visualize() result = graph.execute() print("\n" + "="*50) print("EXECUTION RESULTS:") print("="*50) print(f"\n🔍 QUESTION:\n{result.get('user_question')}\n") print(f"📝 ANSWER:\n{result.get('answer')}\n") print(f"✅ ANALYSIS:\n{result.get('analysis')}") print("="*50 + "\n") return graph The LangGraph class implements a lightweight framework for constructing and executing graph-based AI workflows using Claude from Anthropic. It allows users to define modular nodes, either Claude-powered prompts or custom transformation functions, connect them via dependencies, visualize the entire pipeline, and execute them in topological order. The run_example function demonstrates this by building a simple question-answering and evaluation flow, showcasing the clarity and modularity of LangGraph’s architecture. def run_advanced_example(): """Run a more advanced example with multiple nodes for content generation""" graph = LangGraph() def topic_selector(state, **kwargs): return "Graph-based AI systems" graph.transform_node( name="topic_selector", transform_fn=topic_selector, outputs=["topic"] ) graph.claude_node( name="outline_generator", prompt_template="Create a brief outline for a technical blog post about {topic}. Include 3-4 main sections only.", inputs=["topic"], outputs=["outline"], system_prompt="You are a technical writer specializing in AI technologies." ) graph.claude_node( name="intro_writer", prompt_template="Write an engaging introduction for a blog post with this outline: {outline}\nTopic: {topic}", inputs=["topic", "outline"], outputs=["introduction"], system_prompt="You are a technical writer. Write in a clear, engaging style." ) graph.claude_node( name="conclusion_writer", prompt_template="Write a conclusion for a blog post with this outline: {outline}\nTopic: {topic}", inputs=["topic", "outline"], outputs=["conclusion"], system_prompt="You are a technical writer. Summarize key points and include a forward-looking statement." ) def assembler(state, introduction, outline, conclusion, **kwargs): return f"# {state['topic']}\n\n{introduction}\n\n## Outline\n{outline}\n\n## Conclusion\n{conclusion}" graph.transform_node( name="content_assembler", transform_fn=assembler, inputs=["topic", "introduction", "outline", "conclusion"], outputs=["final_content"] ) graph.visualize() result = graph.execute() print("\n" + "="*50) print("BLOG POST GENERATED:") print("="*50 + "\n") print(result.get("final_content")) print("\n" + "="*50) return graph The run_advanced_example function showcases a more sophisticated use of LangGraph by orchestrating multiple Claude-powered nodes to generate a complete blog post. It starts by selecting a topic, then creates an outline, an introduction, and a conclusion, all using structured Claude prompts. Finally, a transformation node assembles the content into a formatted blog post. This example demonstrates how LangGraph can automate complex, multi-step content generation tasks using modular, connected nodes in a clear and executable flow. print("1. Running simple question-answering example") question = "What are the three main advantages of using graph-based AI architectures?" simple_graph = run_example(question) print("\n2. Running advanced blog post creation example") advanced_graph = run_advanced_example() Finally, we trigger the execution of both defined LangGraph workflows. First, it runs the simple question-answering example by passing a predefined question to the run_example() function. Then, it initiates the more advanced blog post generation workflow using run_advanced_example(). Together, these calls demonstrate the practical flexibility of LangGraph, from basic prompt-based interactions to multi-step content automation using Anthropic’s Claude API. In conclusion, we have implemented LangGraph integrated with Anthropic’s Claude API, which illustrates the ease of designing modular AI workflows that leverage powerful language models in structured, graph-based pipelines. Through visualizing task flows and separating responsibilities among nodes, such as question processing, analytical evaluation, content outlining, and assembly, developers gain practical experience in building maintainable, scalable AI systems. LangGraph’s clear node dependencies and Claude’s sophisticated language capabilities provide an efficient solution for orchestrating complex AI processes, especially for rapid prototyping and execution in environments like Google Colab. Check out the Colab Notebook. All credit for this research goes to the researchers of this project. Also, feel free to follow us on Twitter and don’t forget to join our 95k+ ML SubReddit and Subscribe to our Newsletter. Asif RazzaqWebsite |  + postsBioAsif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.Asif Razzaqhttps://www.marktechpost.com/author/6flvq/Meta Researchers Introduced J1: A Reinforcement Learning Framework That Trains Language Models to Judge With Reasoned Consistency and Minimal DataAsif Razzaqhttps://www.marktechpost.com/author/6flvq/Sampling Without Data is Now Scalable: Meta AI Releases Adjoint Sampling for Reward-Driven Generative ModelingAsif Razzaqhttps://www.marktechpost.com/author/6flvq/Google AI Releases MedGemma: An Open Suite of Models Trained for Performance on Medical Text and Image ComprehensionAsif Razzaqhttps://www.marktechpost.com/author/6flvq/NVIDIA Releases Cosmos-Reason1: A Suite of AI Models Advancing Physical Common Sense and Embodied Reasoning in Real-World Environments

Google CEO Sundar Pichai speaks during a Google I/O conference. Justin Sullivan/Getty Images 2025-05-21T22:01:46Z Save Saved Read in app This story is available exclusively to Business Insider subscribers. Become an Insider and start reading now. Have an account? Google announced over two dozen new AI updates at its I/O developer conference. It's impressive, though some of the new products seem to overlap significantly. Google's approach could lose to more focused rivals as tech races to build an "everything app." Attending Google's I/O developer conference is like being doused with a firehose of new AI announcements.At I/O's keynote event on Tuesday, Business Insider counted at least two dozen new models, features, and updates."We are shipping faster than ever," Google CEO Sundar Pichai boasted onstage.Indeed. But it's starting to get a little confusing. For one, some of the launches seem to overlap with each other. Launching so many AI products in such a short timeframe is impressive, and it can also feel scatterbrained.AI Mode allows you to chat with Google as you browse the web, creating a more conversational search experience. Don't confuse it with Gemini in Chrome, which allows you to ask Gemini questions while you browse.With Gemini Live, you can point your phone at whatever you want and talk to the AI assistant about it. Don't mistake it for Search Live, which allows you to chat with Search about whatever your phone sees.Project Mariner is an experimental AI agent that can take actions like booking tickets. Gemini's upcoming Agent Mode also has agentic capabilities, like helping users find just the right Zillow listing.Not all the new tools seemed that similar. Google launched an impressive new AI filmmaking tool called Flow, powered by its new model Veo 3.Google also touted updates to an entirely separate AI model family from Gemini called Gemma which, incidentally, can help decipher how dolphins talk to each other — that's DolphinGemma.Multiple Googlers that Business Insider spoke with at I/O used a single word to describe Google's current rate of shipping: "intense."Google's approach complicates its own vision of building a single, universal AI assistant. (That mission has its own name, too: Project Astra.)OpenAI is also moving fast towards this goal and appears intent on launching a dedicated device to run it, given its recent purchase of Apple designer Jony Ive's hardware startup.Google risks building so many overlapping AI products that it will be tough to compete with a single, more stand-alone solution, such as an AI-native phone.No one's counting Google out, though. The tech giant has become an undeniable AI leader, inventing much of the core research behind the current boom and successfully launching transformational technology like Waymo. Time will tell whether Google's more sprawling approach wins out.Google did not immediately respond to a request for comment from Business Insider.

BillionPhotos.com - stock.adobe. News Google I/O: LLM capabilities power agentic AI search As Google strives to make AI universal, it is starting to integrate agentic AI into Google Search to fast-track purchasing on websites By Cliff Saran, Managing Editor Published: 21 May 2025 17:00 Google has taken steps to advance artificial intelligence (AI) language models closer to what it calls “world models”, as it tries to make them more useful and universal. The company used its annual developer event, Google I/O, to showcase the Gemini 2.5 large language model (LLM), new application programming interfaces (APIs) and programming tools and agentic AI functionality built into Google’s internet search engine.  Gemini is Google’s primary AI engine, but it offers several others including Gemma 3n, a small language model for mobile devices. Demis Hassabis, CEO of Google Deepmind, said: “Our ultimate vision is to transform the Gemini app into a universal AI assistant that will perform everyday tasks for us, take care of our mundane admin and surface delightful new recommendations – making us more productive and enriching our lives.” Hassabis said the company was beginning to develop new AI capabilities, following on from work on a research prototype called Project Astra, which explored concepts such as video understanding, screen sharing and memory. “Over the past year, we’ve been integrating capabilities like these into Gemini Live for more people to experience today.” Google has been working to make its main AI model, Gemini, a world model. With Gemini 2.5 Pro, Hassabis said the model can make plans and imagine new experiences by understanding and simulating aspects of the world. Hassabis said the progress the company has made is based on training AI agents to master complex games such as Go and StarCraft, with its Genie 2 software able to generate 3D-simulated interactive worlds. According to Hassabis, Gemini is making use of this work in how it handles world knowledge and reasoning to represent and simulate natural environments. Other examples include Veo, Google’s AI-based video content generator, which Hassabis said has a deep understanding of “intuitive physics”. As it strives to make its AI more useful, the company has released a Gemini 2.5-powered feature called AI Mode, on its North American internet search site, to provide more in-depth querying than just what is possible with the AI Overview functionality currently available. An agentic AI feature called Project Mariner is also now part of AI Mode, which Google said can help people searching the internet get tasks done quicker. As an example, Google said a query to find affordable tickets would use AI Mode to look across multiple websites, analysing hundreds of potential ticket options with real-time pricing and inventory, and handle the work of filling in forms. “AI Mode will present ticket options that meet your exact criteria, and you can complete the purchase on whichever site you prefer, saving you time while keeping you in control,” Google said. Another agentic AI feature uses AI Mode to fast-track browsing and purchases on websites, with the entire payment process automated using Google Pay. To support software developers, Google has integrated Gemini 2.5 Pro into the native code editor of Google AI Studio, which it said would help programmers prototype faster. It has also released a beta version of Jules, an asynchronous code agent, which works directly with a software developer’s GitHub repositories. Google said users can ask Jules to take on tasks such as version upgrades, writing tests, updating features and bug fixes. Read more about Google AI models Gemini vs. ChatGPT – what’s the difference: ChatGPT took the early lead among AI-generated chatbots before Google answered with Gemini. While ChatGPT and Gemini perform similar tasks, there are differences. Google Gemini 2.5 Pro explained – Everything you need to know: Google’s latest multimodal model – Gemini 2.5 Pro – entered the AI race with enhanced reasoning and improved performance across coding, math and science benchmarks. In The Current Issue: UK critical systems at risk from ‘digital divide’ created by AI threats UK at risk of Russian cyber and physical attacks as Ukraine seeks peace deal Standard Chartered grounds AI ambitions in data governance Download Current Issue Microsoft entices developers to build more Windows AI apps – Cliff Saran's Enterprise blog Red Hat launches llm-d community & project – Open Source Insider View All Blogs

Large language models are now being used for evaluation and judgment tasks, extending beyond their traditional role of text generation. This has led to “LLM-as-a-Judge,” where models assess outputs from other language models. Such evaluations are essential in reinforcement learning pipelines, benchmark testing, and system alignment. These judge models rely on internal chain-of-thought reasoning, mirroring human judgment processes. Unlike conventional reward models that provide direct scores, these models simulate thoughtful evaluation, making them better suited for complex tasks such as math problem-solving, ethical reasoning, and user intent interpretation. Their ability to interpret and validate responses across languages and domains enhances automation and scalability in language model development. However, current AI judgment systems face issues with inconsistency and shallow reasoning. Many rely on basic metrics or static annotations, which are inadequate for evaluating subjective or open-ended prompts. A common problem is position bias, where the order of answers affects the final decision, compromising fairness. Also, collecting human-annotated data at scale is costly and time-consuming, limiting the generalizability of these models. Several existing approaches have addressed these challenges, but with limited success. Systems like EvalPlanner and DeepSeek-GRM rely on human-labeled data or rigid training schemes, which limit adaptability across task types. Others, like DeepSeek-R1, depend on distillation from large models but perform poorly on ambiguous prompts. Static datasets and offline tuning strategies hinder dynamic reasoning, while newer methods using score formatting or structured prompts have shown minimal accuracy improvements. Despite larger datasets and models, performance gains in traditional systems have stalled. Researchers from Meta’s GenAI and FAIR teams introduced J1 to address the above limitations. J1 trains judgment models through a reinforcement learning-based framework, making them capable of learning through verifiable reward signals. The team used synthetic data to create high-quality and low-quality responses to a prompt, transforming subjective tasks into verifiable pairwise judgments. This synthetic dataset included 22,000 preference pairs, split between 17,000 prompts from the WildChat corpus and 5,000 mathematical queries. These were used to train two versions of J1: J1-Llama-8B and J1-Llama-70B, initialized from the Llama-3.1-8B-Instruct and Llama-3.3-70B-Instruct base models, respectively. The models were trained using Group Relative Policy Optimization (GRPO), a reinforcement algorithm that eliminates the need for critic models and accelerates convergence. At the training strategy’s core is position-agnostic learning, where both (x, a, b) and (x, b, a) input formats are used in training to prevent position bias. Also, consistency-based rewards are applied only when the model delivers correct verdicts across both answer orderings. This structure allows the judge to be fair and reliable regardless of prompt or answer order. The training framework supports multiple variations: models can output final verdicts, numeric scores for each answer, or both. A pointwise judging variant is included, which evaluates single responses using scores from 0 to 10. These formats make J1 a versatile and generalizable system capable of judging various tasks. The results obtained using the J1 models reveal substantial performance improvements over existing systems. On the widely used Preference Proxy Evaluations (PPE) benchmark, J1-Llama-70B achieved an overall accuracy of 69.6%, outperforming models trained with over ten times more data. In contrast, models like DeepSeek-GRM-27B and EvalPlanner-Llama-70B scored 67.2% and 65.6%, respectively. Even the smaller J1-Llama-8B model exceeded baseline systems like EvalPlanner-Llama-8B, scoring 62.2% versus 55.5%. J1 also showed top-tier performance on other critical benchmarks such as RewardBench, RM-Bench, JudgeBench, and FollowBenchEval, demonstrating robust generalization across verifiable and subjective tasks. These improvements are not just marginal but significant, considering the limited training data used in J1 compared to the expansive datasets in other models. Several Key Takeaways from the Research on J1: J1 is trained using 22,000 synthetic preference pairs, including 17K from WildChat and 5K from MATH tasks. The training uses GRPO, which streamlines RL by avoiding the need for separate critic models. It introduces position-agnostic learning, reducing position bias through consistency-based rewards. Two main model variants, J1-Llama-8B and J1-Llama-70B, were trained on modest data but outperformed large-scale models. J1-Llama-70B scored 69.6% on PPE, exceeding DeepSeek-GRM-27B (67.2%) and EvalPlanner-Llama-70B (65.6%). Supports multiple judgment formats: pairwise with verdicts, pairwise with scores, and pointwise scores. Surpasses models distilled from DeepSeek-R1 and OpenAI’s o1-mini on several tasks. Demonstrates that reasoning quality, not just dataset size, is critical for accurate judgments. J1’s framework makes it a generalist judge applicable to verifiable and non-verifiable tasks. In conclusion, the J1 approach fundamentally redefines how judgment models are trained and evaluated. Synthetic data and reinforcement learning bypass the traditional need for costly annotations while promoting fair, logical, and consistent evaluations. This work illustrates that reasoning-driven judging can outperform larger models that rely heavily on data volume and static alignment techniques. It also validates the notion that judgment models should be thinkers first, and scorers second. With performance that rivals and often surpasses state-of-the-art systems, J1 sets a new benchmark in training LLM-as-a-Judge systems. Check out the Paper. All credit for this research goes to the researchers of this project. Also, feel free to follow us on Twitter and don’t forget to join our 95k+ ML SubReddit and Subscribe to our Newsletter. Asif RazzaqWebsite |  + postsBioAsif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.Asif Razzaqhttps://www.marktechpost.com/author/6flvq/Sampling Without Data is Now Scalable: Meta AI Releases Adjoint Sampling for Reward-Driven Generative ModelingAsif Razzaqhttps://www.marktechpost.com/author/6flvq/Google AI Releases MedGemma: An Open Suite of Models Trained for Performance on Medical Text and Image ComprehensionAsif Razzaqhttps://www.marktechpost.com/author/6flvq/NVIDIA Releases Cosmos-Reason1: A Suite of AI Models Advancing Physical Common Sense and Embodied Reasoning in Real-World EnvironmentsAsif Razzaqhttps://www.marktechpost.com/author/6flvq/A Step-by-Step Coding Guide to Efficiently Fine-Tune Qwen3-14B Using Unsloth AI on Google Colab with Mixed Datasets and LoRA Optimization