Trending Research
Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small
openai/transformer-debugger
•
•
Research in mechanistic interpretability seeks to explain behaviors of machine learning models in terms of their internal components.
DoRA: Weight-Decomposed Low-Rank Adaptation
NVlabs/DoRA
•
•
By employing DoRA, we enhance both the learning capacity and training stability of LoRA while avoiding any additional inference overhead.
SINDy-RL: Interpretable and Efficient Model-Based Reinforcement Learning
Deep reinforcement learning (DRL) has shown significant promise for uncovering sophisticated control policies that interact in environments with complicated dynamics, such as stabilizing the magnetohydrodynamics of a tokamak fusion reactor or minimizing the drag force exerted on an object in a fluid flow.
MasterWeaver: Taming Editability and Identity for Personalized Text-to-Image Generation
In this work, we present MasterWeaver, a test-time tuning-free method designed to generate personalized images with both faithful identity fidelity and flexible editability.
Perseus: Removing Energy Bloat from Large Model Training
ml-energy/zeus
•
•
Training large AI models on numerous GPUs consumes a massive amount of energy.
QLoRA: Efficient Finetuning of Quantized LLMs
internlm/xtuner
•
•
NeurIPS 2023
Our best model family, which we name Guanaco, outperforms all previous openly released models on the Vicuna benchmark, reaching 99. 3% of the performance level of ChatGPT while only requiring 24 hours of finetuning on a single GPU.
EasySpider: A No-Code Visual System for Crawling the Web
As such, web-crawling is an essential tool for both computational and non-computational scientists to conduct research.
UniTable: Towards a Unified Framework for Table Structure Recognition via Self-Supervised Pretraining
Tables convey factual and quantitative data with implicit conventions created by humans that are often challenging for machines to parse.
HMT: Hierarchical Memory Transformer for Long Context Language Processing
With an additional 0. 5% - 2% of parameters, HMT can easily plug in and augment future LLMs to handle long context effectively.
Zero-Shot Tokenizer Transfer
bminixhofer/zett
•
•
Finally, we show that a ZeTT hypernetwork trained for a base (L)LM can also be applied to fine-tuned variants without extra training.
