# Main Content

- **Adversarial policies** trained against the state-of-the-art Go AI, **KataGo**, achieved a **>97% win rate** by exploiting its vulnerabilities rather than through superior gameplay.

- **MLC-LLM enables AMD GPUs** to achieve **80%** of the performance of NVIDIA's RTX 4090 and **94%** of the RTX 3090 Ti for Llama2-7B/13B models, utilizing **ROCm** and Vulkan for broader device compatibility, including SteamDeck.

- **LoRA (Low-Rank Adaptation)** enhances parameter-efficient fine-tuning by inserting low-rank matrices as adapters, significantly reducing the number of trainable parameters while maintaining model performance.

- **Cerebrum** is a novel framework that integrates the **Hodgkin-Huxley neuron model** with **Graph Neural Networks** to simulate and infer synaptic connectivity in large-scale brain networks, enhancing our understanding of neural dynamics.

- **Contextual Backpropagation Loops (CBLs)** enhance neural networks by iteratively refining features through **top-down feedback**, mimicking human cognitive processes to improve understanding and decision-making.

- **JaVAD** is a **custom conv-based voice activity detector** that outperforms existing open-source VADs, utilizing a **sliding window approach** over mel-spectrograms for enhanced speed and efficiency.

- **Neural networks with fixed hidden layer sizes but arbitrary depth can approximate complex functions**, as established by various theorems in neural network theory, particularly those addressing depth's role in representation power.

- **Foundation models (FMs) are set to transform Artificial Life (ALife) research** by automating the discovery of lifelike simulations, significantly reducing reliance on manual design and trial-and-error methods.

- **OREO** introduces a novel **offline RL** method that enhances **multi-step reasoning** in **large language models** by employing **soft Bellman equations** and **preference optimization** for improved credit distribution across reasoning steps.

- **Recent advancements** in handling **implicit feedback** in recommender systems focus on **Logistic Matrix Factorization** and **Collaborative Filtering**, which enhance recommendation accuracy by leveraging vast amounts of user interaction data.  
  [Logistic Matrix Factorization for Implicit Feedback Data](https://stanford.edu/~rezab/nips2014workshop/submits/logmat.pdf)  
  |  
  [Collaborative Filtering for Implicit Feedback Datasets](http://yifanhu.net/PUB/cf.pdf).

- **InkSight** effectively teaches Vision-Language Models to convert handwritten images into digital ink, achieving **87% validity** and **67% indistinguishability** from human-generated ink in blind studies.

- **NVIDIA's AI advancements** this year showcase a shift from basic interactions to **dynamic, agentic AI**, enabling complex problem-solving and personalized user experiences through tools like ChatRTX and AnythingLLM.

- **New consistency metric**: A novel **arbitrage-based consistency metric** evaluates LLM forecasters by measuring the logical coherence of their predictions on related questions, such as conflicting probabilities for political outcomes.

- **KunServe** introduces a **parameter-centric memory management** approach that selectively drops replicated model parameters, optimizing GPU memory usage during high-load scenarios without compromising ongoing request performance.

- **XY-Serve** introduces a novel **end-to-end production LLM-serving system** that optimizes workload variability through a unique abstraction mechanism, enhancing efficiency on AI accelerators.