borderland
Test-Time Scaling in Reasoning Models Is Not Effective for Knowledge-Intensive Tasks Yet
Zhao, James Xu, Hooi, Bryan, Ng, See-Kiong
Test-time scaling increases inference-time computation by allowing models to generate long reasoning chains, and has shown strong performance across many domains. However, in this work, we show that this approach is not yet effective for knowledge-intensive tasks, where high factual accuracy and low hallucination rates are essential. We conduct a comprehensive evaluation of test-time scaling using 12 reasoning models on two knowledge-intensive benchmarks. Our results reveal that increasing test-time computation does not consistently improve accuracy and, in many cases, it even leads to more hallucinations. We then analyze how extended reasoning affects hallucination behavior. We find that reduced hallucinations often result from the model choosing to abstain after thinking more, rather than from improved factual recall. Conversely, for some models, longer reasoning encourages attempts on previously unanswered questions, many of which result in hallucinations. Case studies show that extended reasoning can induce confirmation bias, leading to overconfident hallucinations. Despite these limitations, we observe that compared to non-thinking, enabling thinking remains beneficial. Code and data are available at https://github.com/XuZhao0/tts-knowledge
- North America > Canada > Ontario > Toronto (0.15)
- Europe > Austria > Vienna (0.14)
- Europe > United Kingdom (0.14)
- (5 more...)
'New Tales from the Borderlands' is a perfect spiritual successor
The game looks great, and one of my favorite parts was just watching the characters move in scenes. The dialogue and movement always match perfectly, with Octavio's loosey-goosey style countered by his sister's quick anxious movements and Fran's stillness in her hoverchair. It's clear the development team and cast put a lot of effort into the motion capture performances. On top of that, the settings are always covered in little touches that just make the world feel developed. As for the dialogue itself, if you know Borderlands, then you already know what it's like: dark humor and serious conversation all blend together in a soup, each standing out in its own moment without muddying the overall flavor profile.
'New Tales from the Borderlands' is more Gearbox than Telltale
It's hard to put a finger on what, exactly, as the action plays out during a virtual demo presented by Gearbox, the studio handling development after Telltale, the spinoff's originator, died in 2018 and then (sort of) came back the next year. The setup is reminiscent of the original "Tales from the Borderlands" while starring a new cast: Anu, Octavio and Fran, three lovable losers who'd likely be background NPCs in a standard first-person shooter "Borderlands" game, bumble, stumble and stutter their way through the machinations of evil capitalists and a planetary invasion on the worst day of their lives. There are jokes and quips aplenty. There are dialogue choices, quick-time events and minigames. Octavio, the main male lead, exudes a certain-to-be-humbled cocksureness that's reminiscent of previous "Tales" lead Rhys, except streetwise instead of corporate.
Dungeon crawler or looter shooter? Nine video game genres explained
This term is a portmanteau derived from two beloved games that arrived on the Nintendo Entertainment System in the mid-1980s, Metroid and Castlevania, and is usually applied to 2D games in which the world is explorable in all directions (as opposed to classic platform games, in which you go from left to right). There are usually secret rooms and areas that can only be accessed once you've found some key or item later on, so players have to mentally map their progress and backtrack when necessary. In this way a good metroidvania world is like a story, with tension, foreshadowing, plants, payoffs and surprise reveals built into the very foundations. Try: Hollow Knight, Axiom Verge, Ori and the Blind Forest. One of the most popular indie game genres, the term roguelike comes from the 1980 game Rogue, originally developed by coders Michael Toy and Glenn Wichman.
The Joy and Misery of Hunting for Video Game Trophies
I'm burning rubber as I approach the motel parking lot. The tires squeal in protest as I spin the handlebars hard, waiting for the motorbike to start drifting before I hit the nitro. I'm doing a donut, not too tight, just clearing the parked cars. Surely this time I'm going to make it. I only need five seconds, four, three, two … and a zombie slams into me out of nowhere, sending me sprawling onto the asphalt.
The Eight Best Local Multiplayer Games To Play With Family On Thanksgiving
If your Thanksgivings are anything like mine over the years, there's a bit of dead time throughout the afternoon/evening where cousins, nieces, nephews, brothers and sisters need to be kept occupied. As such, in my house that meant it was often time to bust out the video game consoles. There are less and less local multiplayer games that don't rely on passing a controller around the room. And you'll be lucky if you even own 2-4 controllers in the first place, given how rare local multiplayer can be these days. But great local multiplayer experiences do still exist this console generation, and I wanted to highlight eight games in particular that might work well for this circumstance.
- Information Technology > Artificial Intelligence > Games (0.73)
- Information Technology > Communications > Social Media (0.51)