While there are quite a few horror games out these days, the genre still maintains quite a bit of popularity among indie game developers, where small studios are doing their best to come up with fun ways to reinvent the genre. The latest such attempt is Pine Harbor by Vision Forge Team, which in some ways feels like a retread of older ideas, but in other ways, also feels like an interesting new take on horror games. It is worth noting, however, that Pine Harbor is still in Early Access at the time of writing this review, and will undoubtedly go through quite a few changes before it hits its final release.At its core, Pine Harbor is an incredibly simple game: it melds mechanics that would be more at home in a genre like survival or even extraction shooters with a heavy dose of horror-based tension for its core gameplay loop. As the games description on Steam will aptly tell you, the entire gameplay loop revolves around scavenging for supplies and materials during the day, which in turn will help you out with surviving the following night. And while this core gameplay loop always exists in the background, theres also a story that tries to give players more structure with how they go about achieving their goals."Shortly after getting separated from Richard, its up to the players to try and figure out what kind of mysterious secret he was uncovering, and why a corporation might want to shut him up."The setup is rather simple: Pine Harbor is a small town where some sort of accident caused a factory to blow up. This, for some reason, has essentially caused the town to become infected with zombie-like enemies all over the place, in turn leading the town to get abandoned by most of its regular populace. Players take on the role of Josh, who hitches a bus ride to the eponymous town at the seeming behest of his father. Shortly after meeting his father, Richard, Josh discovers that the letter wasnt sent by him, and that darker forces, and maybe even a conspiracy might be at play.Shortly after getting separated from Richard, its up to the players to try and figure out what kind of mysterious secret he was uncovering, and why a corporation might want to shut him up. It also doesnt help that the zombie-like skinners also seem hellbent on taking him, and by extension, you down. Throughout the plot, youll be sent all over Pine Harbor on missions, ranging from looking for some supplies for some fellow surviving humans that might be looking forward to escaping, to following the cryptic instructions left behind by your father.Generally speaking, Pine Harbor feels like its inspired quite a bit by classic survival horror titles like Silent Hill. The titular town is constantly covered in fog, which does a great job in adding to the general atmosphere of the game. The general nature of the narrative also feels like its lifting a page right out of Silent Hill, thanks to its creepy atmosphere, weird-looking enemies, and strange-acting NPCs."Pine Harbor feels like its inspired quite a bit by classic survival horror titles like Silent Hill"Interestingly, the NPCs behaving strangely might actually be the result of the game still being in Early Access. Right from the get go, its quite evident theres still a lot of work left to be done on Pine Harbor. Among other things, just about every animation in the whole game feels incredibly stiff, with some standout examples being Richard being incredibly still while talking to you in a moving car, or how enemies generally tend to behave. The generally strange behavior of the NPCs also goes beyond just stiff animations, and crosses right over into the bad voice acting category.Most quests tend to have you running between two distant parts of the town on simple fetch quests, and sometimes taking on some enemies along the way. While this has the great effect of getting you intimately familiar with the town over a long period of play, unfortunately, this also has the side-effect of making the game feel incredibly repetitive. The quests needed a lot more variety in them than just simply being sent to a building on the other side of town to either talk to a survivor, or to find some supplies that you then have to bring all the way back.While the core structure of Pine Harbor shows that a fun game might be somewhere in there, just about everything else throughout your time with the title will prove to be tedious, frustrating, or outright boring. For example, there isnt really much in the way of combat options when it comes to dealing with enemies. You either shoot them if you have a gun, or you keep spamming a basic melee attack until they go down. This issue is further exacerbated by the fact that enemies dont really tend to react in any interesting way to your actions, aside from coming to you when they hear you getting close. This isnt, for example, like Resident Evil where you have a ton of options in combat by choosing to shoot different body parts on zombies. Rather, all you can really do to deal with them is keep spamming left click and wait for them to eventually go down."You either shoot them if you have a gun, or you keep spamming a basic melee attack until they go down."Pine Harbor quite obviously still has a long road ahead of it, and in its current state, keeping it in Early Access is definitely the right call. While the background setting is pretty interesting, the main story is bland because of the lack of quality voice acting or even decent writing. NPC characters will often act like theyre either students acting on stage for the first time, or aliens that are failing miserably in their attempts to sound like actual human beings. Couple that with the fact that the gameplay is pretty boring, and the missions are just simple fetch quests, and there really isnt much here.The game really starts to come alive at night, when youre shacked up in your house and have to make some decisions about how youre going to deal with skinners that will be gunning straight for your head. Unfortunately, the bad enemy AI rears its ugly head once again, and you never really have to employ much in the way of strategy aside from finding a good place to hide and having plenty of ammo to survive the night.Somewhere inside Pine Harbor, there is an intriguing game. The fact that its an Early Access release also means that theres nowhere for the games quality to go but up. The foundations are decent, and the general vibes of the game is pretty spot-on when it comes to instilling a sense of horror in the player. Unfortunately, the roughness around the edges, the lack of any real combat options, bland quests, and terrible voice acting hold the game back quite a bit.As it currently stands, Pine Harbor isnt really worth your time, even if you are a horror fan. There are a few good ideas there, however, and with time, there are definitely hopes that it goes on to become a decent addition to the horror genre.This game was reviewed on PC.

Among the early access titles released this year, Moon Studios No Rest for the Wicked had a rough start with stuttering, inventory management, janky movement, and difficulty spikes. However, its improved significantly with more patches. In a Steam Community post, the development team teased whats coming up in 2025. These include The Breach, the next major update, and subsequent Content and Feature updates.We know weve been quiet a lot has been happening for us at Moon Studios this year, and we will finally be able to share all about it very soon! There will be better and more frequent communication in 2025, starting with an announcement in January.No Rest for the Wicked is exclusive to PC via Steam Early Access, with console versions in the works. You can check out our review for the launch version. While previously published by Private Division, Take-Two Interactive has taken over since selling the label last month.Alongside improvements to combat and performance, Moon Studios last major update reworked the Cerim Crucible, adding new chambers, more randomization for the arenas, Echoes for customizing runs, and The Gloam, a new enemy type. Besides The Breach, the team is working on updates to farming and other features.

Large Language Models (LLMs) have demonstrated impressive proficiency in numerous tasks, but their ability to perform multi-step reasoning remains a significant challenge. This limitation becomes particularly evident in complex scenarios such as mathematical problem-solving, embodied agent control, and web navigation. Traditional Reinforcement Learning (RL) methods, like Proximal Policy Optimization (PPO), have been applied to address this issue but often come with high computational and data costs, making them less practical. Likewise, methods such as Direct Preference Optimization (DPO), while effective for aligning models with human preferences, struggle with multi-step reasoning tasks. DPOs reliance on pairwise preference data and uniform token treatment undermines its capacity to assign credit effectively in situations with sparse rewards. These obstacles highlight the need for more targeted and efficient solutions to enhance LLM reasoning capabilities.Introducing OREO: Offline Reasoning OptimizationOREO (Offline REasoning Optimization) is an offline RL approach specifically designed to address the shortcomings of existing methods in improving multi-step reasoning for LLMs. Developed collaboratively by researchers from UC San Diego, Tsinghua University, Salesforce Research, and Northwestern University, OREO builds on insights from maximum entropy reinforcement learning. It trains a policy model and a value function concurrently by optimizing the soft Bellman Equation. This methodology removes the dependency on pairwise preference data, making it possible to utilize unpaired datasets with sparse rewards. Furthermore, OREO enables precise credit assignment across reasoning trajectories, which is especially beneficial when success depends on a few critical steps. The framework can also be extended to iterative exploration setups and incorporates a learned value function to enhance inference through tree search during testing.Technical Details and BenefitsOREOs core innovation lies in optimizing the soft Bellman Equation to simultaneously train policy and value models. This strategy ensures accurate credit assignment across reasoning steps, addressing the limitations of methods like DPO. Additionally, OREO offers step-level and response-level objectives, providing flexibility for different granularities of reasoning tasks. During test-time inference, the value function supports advanced search techniques, such as beam search, improving accuracy. Unlike baseline methods like supervised fine-tuning (SFT) or rejection sampling, OREO excels at leveraging failed trajectories to enhance model robustness and adaptability. This capacity to learn from failures makes it particularly valuable for iterative multi-step reasoning tasks.Results and InsightsOREOs performance has been rigorously evaluated on benchmarks such as GSM8K and MATH for mathematical reasoning, and ALFWorld for embodied agent control. Key findings include:On GSM8K, OREO delivered a 5.2% relative improvement in accuracy using a 1.5B parameter model compared to SFT, alongside a 10.5% improvement on MATH.52.5% on MATH with 1.5B LLM (w/o using augmented problem set)In ALFWorld, OREO achieved a 17.7% relative improvement in performance in unseen environments, underscoring its ability to generalize beyond training data.Iterative training further amplified OREOs effectiveness, showing consistent accuracy gains over multiple iterations. While approaches like rejection sampling exhibited diminishing returns, OREO continued to improve by incorporating insights from failed attempts. Test-time search using OREOs value function resulted in up to a 17.9% relative improvement over greedy decoding on the MATH dataset, highlighting its impact on inference quality.ConclusionOREO provides a practical and effective solution for enhancing multi-step reasoning in LLMs through offline RL. By addressing the limitations of existing approaches, it offers a scalable method for improving reasoning capabilities. Its integration of detailed credit assignment, iterative training, and test-time search makes it a versatile tool for addressing complex reasoning challenges. The results demonstrate OREOs potential for application across a range of domains requiring sophisticated problem-solving, contributing to the evolution of AI systems capable of deeper reasoning.Check out the Paper. All credit for this research goes to the researchers of this project. Also,dont forget to follow us onTwitter and join ourTelegram Channel andLinkedIn Group. Dont Forget to join our60k+ ML SubReddit. Asif RazzaqAsif 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. [Download] Evaluation of Large Language Model Vulnerabilities Report (Promoted)

10 Comprehensive Strategies for Ensuring Ethical Artificial Intelligence 0 like December 23, 2024Share this postLast Updated on December 24, 2024 by Editorial TeamAuthor(s): Veritas AI Originally published on Towards AI. This member-only story is on us. Upgrade to access all of Medium.Now, we are in the middle of a very unusual rise of artificial intelligence, especially in this post-GPT and generative AI era. This emergence is going to get much stronger for the next few years and will see AI being introduced more and more into all areas of businesses, industries and directly into our daily lives.However, we have to move from the state of wonder and seriously think about the positioning that we give to AI in our lives and the risk(s) that this represents. From a purely technical point of view, an AI can only seem enormously useful. Still, it can hide fine layers of problems linked mainly to its structures, architecture, and model. not to mention the risk of using AI to achieve extremist groups bad intentions. In addition to all national or international laws in progress or already implemented, allowing the regularization of the creation and use of AI, it is necessary to be able to put in place systems that can evaluate AI about its compliance with certain ethical principles.The integration of Artificial Intelligence into various sectors of society raises important ethical concerns that must Read the full blog for free on Medium.Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming asponsor. Published via Towards AITowards AI - Medium Share this post

LatestMachine LearningStep-by-Step Exploration of Transformer Attention Mechanisms 0 like December 23, 2024Share this postLast Updated on December 24, 2024 by Editorial TeamAuthor(s): Shenggang Li Originally published on Towards AI. A Practical Walkthrough of Training Transformer Models with Insights into Positional Encoding and Its Role in Attention DynamicsThis member-only story is on us. Upgrade to access all of Medium.Photo by Abiyyu Zahy on UnsplashIf youre diving into AI and want to understand the secret sauce behind modern language models like ChatGPT or BERT, you need to get familiar with Transformers and their game-changing attention mechanism. These concepts are the foundation of cutting-edge NLP, and once you grasp them, youll see why theyre so powerful and versatile.Imagine youre trying to read a book, not line by line, but by flipping to any page you want instantly and picking up on the connections between parts of the story. Thats kind of what Transformers do in NLP. They ditched the old ways of reading word-by-word, like RNNs or LSTMs, and instead take in whole chunks of data whether its a sentence, a paragraph, or an entire sequence all at once. This gives them super speed in training and makes them great at spotting patterns across the whole text.At the heart of this magic is something called the attention mechanism. Its like having a spotlight that focuses on the most important words in a sentence while still keeping an eye on the rest.Were going to break it all down Read the full blog for free on Medium.Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming asponsor. Published via Towards AITowards AI - Medium Share this post

LatestMachine LearningWhisper.cpp: How to Use OpenAIs Whisper Model in C/C++ for Efficient Speech Recognition 0 like December 22, 2024Share this postLast Updated on December 24, 2024 by Editorial TeamAuthor(s): Md Monsur ali Originally published on Towards AI. OpenAIs Whisper in C/C++ for Accurate, High-Speed Transcription Without Internet Step-by-Step TutorialThis member-only story is on us. Upgrade to access all of Medium. GitHub | LinkedIn | Medium | Ko-fiSource: https://github.com/ggerganov/whisper.cppIn the fast-evolving field of artificial intelligence and machine learning, the Whisper model developed by OpenAI has been a game-changer for automatic speech recognition. Designed to provide highly accurate transcription, translation, and multilingual speech recognition from the start, Whisper was a strong tool for developers working with speech-related applications. The original model, however, is implemented in Python, whereas many developers like to work with more lightweight, efficient, and portable implementations in their systems. Enter Whisper.cpp: an optimized C/C++ version of OpenAIs model, Whisper, designed for fast, cross-platform performance.In this post, we will take a closer look at what Whisper.cpp is, its main features, and how it can be used to bring speech recognition into applications such as voice assistants or real-time transcription systems.Whisper.cpp is the OpenAI Whisper Model implementation in C and C++. It has been made, trying to achieve as much performance and portability as the model itself and aiming at running Whisper on platforms that cannot utilize the original Python model: it will make embedding much simpler in systems with restricted resources, like some embedded Read the full blog for free on Medium.Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming asponsor. Published via Towards AITowards AI - Medium Share this post

Dec 24, 2024Ravie LakshmananVulnerability / Zero DayThe Apache Software Foundation (ASF) has released a security update to address an important vulnerability in its Tomcat server software that could result in remote code execution (RCE) under certain conditions.The vulnerability, tracked as CVE-2024-56337, has been described as an incomplete mitigation for CVE-2024-50379 (CVSS score: 9.8), another critical security flaw in the same product that was previously addressed on December 17, 2024."Users running Tomcat on a case insensitive file system with the default servlet write enabled (readonly initialisation parameter set to the non-default value of false) may need additional configuration to fully mitigate CVE-2024-50379 depending on which version of Java they are using with Tomcat," the project maintainers said in an advisory last week.Both the flaws are Time-of-check Time-of-use (TOCTOU) race condition vulnerabilities that could result in code execution on case-insensitive file systems when the default servlet is enabled for write."Concurrent read and upload under load of the same file can bypass Tomcat's case sensitivity checks and cause an uploaded file to be treated as a JSP leading to remote code execution," Apache noted in an alert for CVE-2024-50379.CVE-2024-56337 impacts the below versions of Apache Tomcat -Apache Tomcat 11.0.0-M1 to 11.0.1 (Fixed in 11.0.2 or later)Apache Tomcat 10.1.0-M1 to 10.1.33 (Fixed in 10.1.34 or later)Apache Tomcat 9.0.0.M1 to 9.0.97 (Fixed in 9.0.98 or later)Additionally, users are required to carry out the following configuration changes depending on the version of Java being run -Java 8 or Java 11 - Explicitly set system property sun.io.useCanonCaches to false (it defaults to true) Java 17 - Set system property sun.io.useCanonCaches to false, if already set (it defaults to false)Java 21 and later - No action is required, as the system property has been removedThe ASF credited security researchers Nacl, WHOAMI, Yemoli, and Ruozhi for identifying and reporting both shortcomings. It also acknowledged the KnownSec 404 Team for independently reporting CVE-2024-56337 with a proof-of-concept (PoC) code.The disclosure comes as the Zero Day Initiative (ZDI) shared details of a critical bug in Webmin (CVE-2024-12828, CVSS score: 9.9) that allows authenticated remote attackers to execute arbitrary code."The specific flaw exists within the handling of CGI requests," the ZDI said. "The issue results from the lack of proper validation of a user-supplied string before using it to execute a system call. An attacker can leverage this vulnerability to execute code in the context of root."Found this article interesting? Follow us on Twitter and LinkedIn to read more exclusive content we post.SHARE

Looking forthe most recentMini Crossword answer?Click here for today's Mini Crossword hints, as well as our daily answers and hints for The New York Times Wordle, Strands and Connections puzzles.The New York TimesMini Crosswordthrew kind of a toughie at us for Christmas Eve, I thought. And that day actually showed up in the clues -- spoiler!. Anyway, if you're stuck on any of the clues today, we've got the answers. And if you could use some hints and guidance for daily solving, check out our Mini Crossword tips.The Mini Crossword is just one of many games in the Times' games collection. If you're looking for today's Wordle, Connections and Strands answers, you can visitCNET's NYT puzzle hints page.Read more: Tips and Tricks for Solving The New York Times Mini CrosswordLet's get at those Mini Crossword clues and answers. The completed NYT Mini Crossword puzzle for Dec. 24, 2024. NYT/Screenshot by CNETMini across clues and answers1A clue: Christmas ___Answer: EVE4A clue: Hanker for, as Christmas cookiesAnswer: CRAVE6A clue: Instrument heard in many a Christmas hymnAnswer: ORGAN7A clue: Rounded heapAnswer: MOUND8A clue: Vintner's deviceAnswer: PRESSMini down clues and answers1A clue: Proofreader's catchAnswer: ERROR2A clue: Like a Christmas wish list that just says "toys"Answer: VAGUE3A clue: Chris of "Knives Out"Answer: EVANS4A clue: Provide free of chargeAnswer: COMP5A clue: Comes to a closeAnswer: ENDSHow to play more Mini CrosswordsThe New York Times Games section offers a large number of online games, but only some of them are free for all to play. You can play the current day's Mini Crossword for free, but you'll need a subscription to the Times Games section to play older puzzles from the archives.

Looking for the most recent Connections answers? Click here for today's Connections hints, as well as our daily answers and hints for The New York Times Mini Crossword, Wordle and Strands puzzles.TheNew York Times Connections puzzleeditors went kind of loopy on us today. That yellow category is completely strange, oh my. And the blue and purple groups were absolutely killer. If you need help solving the puzzle, read on for today's Connections hints and answers.Read more:Hints, Tips and Strategies to Help You Win at NYT Connections Every TimeHints for today's Connections groupsHere are four hints for the groupings in today's Connections puzzle, ranked from the easiest, yellow group to the tough (and sometimes bizarre) purple group.Yellow group hint: Wizard of Oz quote.Green group hint: Buddies.Blue group hint: Letters in a group.Purple group hint: Songs with repeated words in title.Answers for today's Connections groupsYellow group: Lions and tigers and bears, oh my!Green group: Beloved, as a friend.Blue group: Words that sound like plural letters.Purple group: When tripled, hit song titles.Read more: Wordle Cheat Sheet: Here Are the Most Popular Letters Used in English WordsWhat are today's Connections answers? The completed NYT Connections puzzle for Dec. 24, 2024. NYT/Screenshot by CNETThe yellow words in today's ConnectionsThe theme is "lions and tigers and bears, oh my." The four answers are lions, tigers, bears and oh my.The green words in today's ConnectionsThe theme is beloved, as a friend. The four answers are close, dear, intimate and tight.The blue words in today's ConnectionsThe theme is words that sound like plural letters. The four answers are bees, ease, jays and use.The purple words in today's ConnectionsThe theme is when tripled, hit song titles. The four answers are bills, bye, gimme and please. (Bills, Bills, Bills is by Destiny's Child; Bye, Bye, Bye is by NSYNC; Gimme! Gimme! Gimme! Is by ABBA; and Please Please Please is by Sabrina Carpenter.)

Looking for the most recent Strands answer?Click here for our daily Strands hints, as well as our daily answers and hints for The New York Times Mini Crossword, Wordle and Connections puzzles.Today'sStrandshas a fun topic. I used to watch the CBS soap opera The Bold and The Beautiful, and it always occurred to me how many of the characters had names that were also things in nature Ridge, Thorne and Brooke being the main ones. So this was an entertaining theme for me. If you need more hints and answers for today's Strands puzzle, read on.Also, I go into depth about therules for Strands in this story.If you're looking for today's Wordle, Connections and Mini Crossword answers, you can visitCNET's NYT puzzle hints page.Read more:NYT Connections Turns 1: These Are the 5 Toughest Puzzles So FarHint for today's Strands puzzleToday's Strands theme is:What on earth...?If that doesn't help you, here's a clue: Mother Nature.Clue words to unlock in-game hintsYour goal is to find hidden words that fit the puzzle's theme. If you're stuck, find any words you can. Every time you find three words of four letters or more, Strands will reveal one of the theme words. These are the words I used to get those hints, but any words of four or more letters that you find will work:HEAL, LEAN, VEER, TINE, VINE, MANE, RIVE, MEAL, MALE, ALES, MALES, ROOK, CULL, VINA, RILLAnswers for today's Strands puzzleThese are the answers that tie into the theme. The goal of the puzzle is to find them all, including the spangram, a theme word that reaches from one side of the puzzle to the other. When you've got all of them (I originally thought there were always eight but learned that the number can vary), every letter on the board will be used. Here are the nonspangram answers:HOLLY, BROOK, RIVER, LAUREL, WILLOW, CLEMENTINEToday's Strands spangramToday's Strands spangram isNATURENAMES.To find it, start with the N that's three letters to the right on the top row, and wind down. The completed NYT Strands puzzle for Dec. 24, 2024. NYT/Screenshot by CNET