Bamshad Mobasher

Excited to share our latest paper titled - 𝐇𝐨𝐰 𝐭𝐨 𝐂𝐡𝐨𝐨𝐬𝐞 𝐚 𝐓𝐡𝐫𝐞𝐬𝐡𝐨𝐥𝐝 𝐟𝐨𝐫 𝐚𝐧 𝐄𝐯𝐚𝐥𝐮𝐚𝐭𝐢𝐨𝐧 𝐌𝐞𝐭𝐫𝐢𝐜 𝐟𝐨𝐫 𝐋𝐚𝐫𝐠𝐞 𝐋𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐌𝐨𝐝𝐞𝐥𝐬. This work addresses the critical gap in research on setting robust thresholds for LLM evaluation metrics, which are essential for the reliable deployment of LLMs. Drawing inspiration from model risk management practices in regulated industries like finance, we propose a systematic methodology to define these thresholds. The approach involves identifying risks associated with the specific application, understanding stakeholders' risk tolerance, and applying statistically rigorous procedures using ground-truth data.

58

4 Comments

Predictions based on Machine Learning and Statistics often rely on maximal likelihood, which can trap LLMs in biases inherited from the majority opinions of training data. This problem has been substantially addressed through conditional statistics, allowing for mitigation of biases and preventing the cancellation of minority perspectives. https://lnkd.in/gMXKqwyB

7

Often LLMs hallucinate because of semantic uncertainty due to missing factual training data. We propose a method to detect such uncertainties using only one generated output sequence. Super efficient method to detect hallucination in LLMs. See paper on ArXiv: https://lnkd.in/dS4Vxj7t

692

13 Comments

New GAI Textbook Announcement (The textbook is being used by Stanford CS372: AI for precision medicine and psychological disorders) The past two years have been a significant period in the field of AI, during which we have observed substantial strides in generative AI technologies. Having dedicated over 5,000 hours to research and analysis, I have compiled this book to share a collection of findings and hypotheses regarding Large Language Models (LLMs)—detailing their operational mechanisms, and strategies to mitigate biases and reduce instances of hallucination. The central premise is to enhance context awareness and define human objectives with greater precision. By doing so, LLMs can more effectively mirror human linguistic behaviors, utilizing context-specific linguistic features to fulfill intended goals. SocraSynth: Socratic Synthesis with Multiple Large Language Models. https://lnkd.in/giC4474p

25

Today, we are presenting NLPGuard at #CSCW2024 —a framework designed to reduce black-box NLP models’ reliance on protected attributes, enhancing both fairness and compliance. 𝗪𝗵𝘆 𝗜𝘁 𝗠𝗮𝘁𝘁𝗲𝗿𝘀: NLP models are often opaque, making it challenging to know if they depend on sensitive attributes like race or gender. Traditional fairness methods may achieve performance parity across groups but often fail to fully eliminate underlying biases. 𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗡𝗟𝗣𝗚𝘂𝗮𝗿𝗱: NLPGuard reduces NLP models’ dependence on sensitive terms without compromising performance! This framework can be applied to key tasks like toxicity detection, sentiment analysis, and occupation classification. 𝗛𝗼𝘄 𝗡𝗟𝗣𝗚𝘂𝗮𝗿𝗱 𝗪𝗼𝗿𝗸𝘀: Powered by three components, NLPGuard operates as follows: Explainer: Highlights key words in predictions Identifier: Flags words related to protected attributes Moderator: Adjusts training data to mitigate bias 𝗧𝗵𝗲 𝗥𝗲𝘀𝘂𝗹𝘁𝘀? Testing shows that NLPGuard can reduce reliance on sensitive terms by 60% in toxicity detection and by 79% across other tasks—all while maintaining or even improving model accuracy! For example, in occupational classification (like predicting “attorney” or “nurse”), NLPGuard completely removed gender bias, taking the model’s reliance on gender indicators down to 0%. 𝗚𝗲𝘁 𝗦𝘁𝗮𝗿𝘁𝗲𝗱 𝘄𝗶𝘁𝗵 𝗡𝗟𝗣𝗚𝘂𝗮𝗿𝗱! NLPGuard is now open-source, offering a valuable resource for those aiming to comply with evolving regulations in AI ethics. Explore the project and dive into our full study here: Project site: https://lnkd.in/dbfpqPwv Paper: https://lnkd.in/dMs5P4a3 Code: https://lnkd.in/dXgPbRwQ With Salvatore Greco Licia Capra Tania Cerquitelli #ResponsibleAI #NLP #FairnessInAI #AICompliance #CSCW

32

Large Generative Graph Models (LGGMs): A New Class of Graph Generative Model Trained on a Large Corpus of Graphs https://lnkd.in/d7WYXSyX

8

1 Comment

Fine tuning gpt2 on indic-gpt dataset I ran one epoch of fine tuning on an arbitrary dataset of Indic-gpt and managed to train it for language detection with Colab.Have a look: https://lnkd.in/dZJUY-UV #indic-gpt #fine tuning #language detection

11

I recently read “Towards Monosemanticity: Decomposing Language Models with Dictionary Learning” by Bricken et al 2023. Published at transformer-circuits.pub In this paper the authors discuss the use of autoencoders to extract interpretable features from a one-layer transformer model. The approach aims to improve the understanding and transparency of artificial neural networks. To do this they use the concept of dictionary learning which is a method used to decompose the activations of the transformer model into a set of basis features that are sparse and interpretable. This involves training sparse autoencoders to identify a dictionary of monosemantic features (or features with a single meaning). These monosemantic features in single units (artificial neurons) reminds me of the “grandmother neuron” theory in neuroscience which proposes that specific neurons respond to specific highly complex stimuli. However, in real brains, it is widely believed that complex representations are encoded by the collective activity of groups of neurons (and not monosemantic). But going back to the paper, I really like the visualization tools and illustration to make such complex networks more transparent (as opposed to black boxes). Paper: https://lnkd.in/gR87pJcs

3

Check out this highly recommended course for those interested in optimizing their models: - Introduction to Quantization - Symmetric and Asymmetric Quantization Modes - Hands-on Implementation from Scratch using PyTorch - covering Per Tensor, Per Channel (Axis), and Per Group techniques - Weights Packing: Exploring 2 or 4 Bits Quantization Methods https://lnkd.in/gMAuWXHm

43

Happy to share my paper “An ensemble deep learning model for fast classification of Twitter spam” coauthored with Prof. Indranil Bose is published in Information & Management. This paper offers novel rich and deep feature extraction and using them in an ensemble neural network model for fast classification of Twitter spam. The paper is available at https://lnkd.in/gBFYqQTQ Happy reading!!!

83

7 Comments

𝐌𝐮𝐬𝐭-𝐫𝐞𝐚𝐝, 𝐟𝐫𝐞𝐞 𝐛𝐨𝐨𝐤 𝐭𝐡𝐚𝐭 𝐩𝐫𝐨𝐯𝐢𝐝𝐞𝐬 𝐚 𝐩𝐚𝐭𝐡𝐰𝐚𝐲 𝐟𝐫𝐨𝐦 𝐆𝐀𝐈 𝐭𝐨 𝐀𝐆𝐈! ​ Yann LeCun has expressed skepticism about the potential of LLMs to achieve Artificial General Intelligence (AGI), citing limitations in memory, planning, and grounding in real-world understanding. However, we argue that 𝗟𝗟𝗠 𝗖𝗼𝗹𝗹𝗮𝗯𝗼𝗿𝗮𝘁𝗶𝘃𝗲 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 (𝗟𝗖𝗜)—particularly with multiple multimodal LLMs working together—offers a promising architecture to address these challenges and progress toward AGI. ​ Two key architectural innovations underlie LCI: ​ 𝐒𝐞𝐩𝐚𝐫𝐚𝐭𝐢𝐨𝐧 𝐨𝐟 𝐔𝐧𝐜𝐨𝐧𝐬𝐜𝐢𝐨𝐮𝐬 𝐏𝐫𝐨𝐜𝐞𝐬𝐬𝐢𝐧𝐠 𝐚𝐧𝐝 𝐂𝐨𝐧𝐬𝐜𝐢𝐨𝐮𝐬 𝐑𝐞𝐚𝐬𝐨𝐧𝐢𝐧𝐠 𝐚𝐧𝐝 𝐏𝐥𝐚𝐧𝐧𝐢𝐧𝐠: Inspired by the human mind’s dual-layer structure, LCI divides its architecture into foundational and adaptive layers. The foundation layer, akin to unconscious processing, leverages extensive datasets to build robust pattern recognition, encoding essential responses much like human instincts. The adaptive layer, mirroring conscious thought, enables rapid adaptation and contextual reasoning. This dual-layer approach allows LLMs to perform complex tasks with foundational knowledge while adjusting to new information on the fly, much as a child learns new concepts within a developed cognitive framework. ​ 𝐃𝐢𝐬𝐭𝐢𝐧𝐜𝐭 𝐑𝐨𝐥𝐞𝐬 𝐟𝐨𝐫 𝐄𝐚𝐜𝐡 𝐋𝐋𝐌-𝐀𝐠𝐞𝐧𝐭: Each LLM-agent is assigned a specific role, such as executive (knowledge processing), legislative (behavior guardrails), and judicial (adapting to cultural norms). ​ Additionally, our validation of the LCI framework has been thoroughly tested through real-world deployment in applications across various domains, including healthcare, sales planning, investment, and debiasing news. These empirical deployments highlight the practical viability and adaptability of the LCI framework, showcasing its ability to enhance reasoning and decision-making through collaborative intelligence. ​ The eleven aphorisms presented in this free book illustrate the philosophical foundation of the LCI framework, backed by empirical studies. Through LCI, we envision a clear pathway to AGI, rooted in collaborative intelligence and flexible, context-aware adaptation. https://lnkd.in/gn74DkNj (free) https://lnkd.in/gszSvfqP ($6.98 printing fee)

9

LLMs are powerful but prone to 'hallucinations'—false yet plausible outputs. In our #NeurIPS2024 paper, we introduce a compute-efficient method for detecting hallucinations in single responses using hidden states, attention maps, and output probabilities. Our approach achieves >45x speedups over traditional methods while significantly boosting detection accuracy across diverse datasets. Link to the paper: https://lnkd.in/eBWNXRkx Joint work with Gaurang Sriramanan Siddhant Bharti Vinu Sankar Sadasivan Shoumik Saha Priyatham Kattakinda

309

2 Comments

🚨ALERT: Innovation in artificial neural networks🚨 Published earlier today: “SWAG: A Novel Neural Network Architecture Leveraging Polynomial Activation Functions for Enhanced Deep Learning Efficiency” Kudos to Dr. Saeid Safaei, my last doctoral student at the The University of Georgia. This work is significant because it breaks a dogma in artificial neural network (ANN) literature. In 1993, Leshno et al. reported that an ANN “… can approximate any continuous function to any degree of accuracy if and only if the network's activation function is not a polynomial”. This myth became legend, the legend became dogma. We showed that not only it is possible to use polynomials… it is desirable. Our method with simple architectures outperforms traditional ANNs with complex architectures. It could have profound implications to decrease energy demand for AI through the development of photonic chips. https://lnkd.in/gQAtcNEa

23

8 Comments

How Much Research Is Being Written by Large Language Models? 17.5% (that is assumed to have been detected) https://buff.ly/3VdqVv0 This will mean all future papers on average will be referencing 3 to 5 papers containing LLM generated content significant error rates.

1

👉🏼 Expansive data, extensive model: Investigating discussion topics around LLM through unsupervised machine learning in academic papers and news 🤓 Hae Sun Jung 👇🏻 https://lnkd.in/ef_MF54v 🔍 Focus on data insights: - Utilized data from Web of Science and LexisNexis to explore topic modeling methods for Large Language Models (LLMs). - Evaluated various topic modeling approaches such as LDA, NMF, CTM, and BERTopic based on diversity and coherence metrics. - BERTopic demonstrated superior performance in diversity and coherence across both platforms. 💡 Main outcomes and implications: - News articles covered a wide range of topics related to LLM applications, focusing on specialized domains. - Research papers concentrated on technical aspects of LLM technology. - Insights gained can help understand future challenges and opportunities for LLMs in different sectors. 📚 Field significance: - Enhances understanding of how LLMs are utilized in academic and news contexts. - Provides insights for enterprises leveraging LLMs for service delivery. 🗄️: #topicmodeling #unsupervisedlearning #machinelearning #datainsights

1

AI Safety is an important issue, but the current solution with RLHF suffers from several shortcomings including subjective feedback, feedback hacking, and the forgetting effect. This work, which will be presented at NeurIPS this week, proposes a paradigm shift: using three LLM modules to perform checks and balances to represent knowledge, legislative, and judicial functions. The paper delicts the details. https://lnkd.in/gXthyE9w

9

OpenCoder: An Open-Source Code LLM Redefining Model Efficiency 🌟 The OpenCoder team has launched a game-changer in the world of code language models. Unlike many LLMs that focus on sheer size, OpenCoder brings something unique to the table: a data-focused training approach. By putting quality before volume, OpenCoder’s 1.5B and 8B models rival some of the biggest players, like DeepSeekCoder and QwenCoder, with only a fraction of their token count. The core strength of OpenCoder lies in its meticulous data processing pipeline and two-stage fine-tuning. With a training dataset composed of 2.5 trillion tokens, 90% raw code and 10% code-related web data, OpenCoder refines every aspect of its corpus. Advanced filtering and deduplication strategies, applied to 607 programming languages, optimize model efficiency and precision, ensuring high-quality data drives its training. Key Features: - Complete Training Transparency: Open-source datasets and training pipelines enable reproducibility and further research. - Instruction Fine-Tuning: Two-stage tuning sharpens both core CS concepts and real-world coding skills, prepping OpenCoder for production-level code generation. - Available on Ollama: Accessibility in deployment means more avenues for experimentation and improvement. OpenCoder's data-driven strategy showcases the potential of a leaner approach. By focusing on quality data rather than quantity alone, OpenCoder achieves file-level deduplication that outperforms traditional repo-level methods, making it both efficient and robust. 🚀 Read more here: https://lnkd.in/eEnG3yJA #OpenSource #MachineLearning #CodeLLM #AI #ModelEfficiency #DataQuality #Innovation #CodeGeneration #OpenCoder #TechNews #AIResearch #Programming #LargeLanguageModels #ArtificialIntelligence

1