Christopher Dancy’s Post

Its that time of the year again, lots to read and digest here. I am specifically interested in reading about the risks, bias, policy and governance along with the benefits from accelerated scientific progress. Thanks to Stanford Institute for Human-Centered Artificial Intelligence (HAI) https://lnkd.in/eWrAnJdN

https://lnkd.in/exCsMuJQ ”This book strives to fulfill a three-pronged mission: first, to lay before you in nonmathematical language the intellectual content of the Causal Revolution and how it is affecting our lives as well as our future; second, to share with you some of the heroic journeys, both successful and failed, that scientists have embarked on when confronted by critical cause-effect questions. Finally, returning the Causal Revolution to its womb in artificial intelligence, I aim to describe to you how robots can be constructed that learn to communicate in our mother tongue—the language of cause and effect.”

⌛ DistilSenti-Net: Context Distilled Sentiment Analysis Model Why does context distillation matter in Sentiment Analysis? Context distillation in text helps extract the important parts of a sentence, producing more accurate predictions than using the hard prompt. This sentiment analysis model consists of 42M parameters in total, which is relatively a small text classification model. During the first stage of model training, there were indications of overfitting due to the complex architecture. In the second attempt, a different model architecture was tried but still resulted in overfitting, although this was monitored with early stopping. In the third attempt, the model architecture was simplified, leading to much better results than the previous experiments. The final model was a hybrid combination of Vanilla RNN and LSTM. I would like to share some key takeaways from this experiment: - The most important factor for producing accurate results is the quality of the data. Collect high-quality data and perform necessary preprocessing steps, such as checking for skewness and kurtosis, to ensure a proper data distribution. Note: Generally, the data should follow a Gaussian distribution. - Sometimes, a simple model architecture can solve the problem effectively. However, the optimal architecture depends on the specific problem and the data available. - Experiment with different hyperparameters and model architectures. Note: Do not limit your model training to a small number of epochs. 🔗 (HF) DistilSenti-Net: https://lnkd.in/ebaE59RM #deep_learning #machine_learning #nlp #sentiment_analysis

Embeddings, a new idea on human machine alignment! So much highgrade thinkfluencing this week! I argue here that the concept of embedding (understood in the mathematical and computer science sense) provides a general way of understanding the relation between generative A.I., written language and semiotics, and animal cognition, when understood recursively. I propose this framing as an application of cultural science, and suggest that this offers a new way to understand the alignment problem between humans and increasingly intelligent machines. https://lnkd.in/gVeEuWNU

With all the attention given to generative AI, here is a challenge, open to humans and machines: Compose a poem in the manner of John Donne’s 'A Valediction Forbidding Mourning' using compass and rule to convey the beauty of Euler’s equation, 0 = e^iπ + 1. (^ denotes exponentiation) Hint: start at the origin. From my book, 'Buddhism and Computing: How to Flourish in the Age of Algorithms' (Mud Pie Books).

Discover the latest advancements in depth estimation with the new paper, "Depth Anywhere: Enhancing 360° Monocular Depth Estimation via Perspective Distillation and Unlabeled Data Augmentation." This research introduces innovative techniques to improve the accuracy and reliability of 360° monocular depth estimation. The paper provides a detailed analysis of perspective distillation and the use of unlabeled data augmentation, making it a valuable resource for developers and researchers in computer vision and AI. The findings highlight significant improvements in depth estimation performance, addressing key challenges in the field. For those interested in the technical aspects and practical applications of depth estimation, this paper is a must-read. It offers insights and solutions that contribute meaningfully to ongoing advancements in the field. Read the full paper here: [https://lnkd.in/dPmg_5G3]

#AI is not cheap its needs money to make and it brings money when its made. #opensouce is a gift, that does not mean, integrating #ai comes with "Zero-cost" #LLM #realitycheck , this model will further be fine-tuned on 1. maths 2. sql and custom dataset. #DM for any other daataset you would want to see . based model #mistral instrcut v2. dataset: #orca pairs fine-tuning : DPO https://lnkd.in/gaSjVifd

Businesses were taken by surprise with AI, and have been reminded about the importance of preparing for technical disruption. As interest in AI calms, decision makers need to ensure they don’t overlook quantum. "Quantum-Inspired Soft Computing Algorithms for Intelligent Data Science" offers a timely and comprehensive exploration of this transformative field. It promises to equip readers with the knowledge and practical insights needed to leverage the power of quantum-inspired algorithms for groundbreaking data science advancements. Do submit your quality chapters to the upcoming River Publishers book titled "Quantum-Inspired Soft Computing Algorithms for Intelligent Data Science " #scopus #quantumcomputing #softcomputing #datascience #artificialintelligence Submit your abstract and full-chapters to : [email protected]

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

As the new school year begins, it's the perfect time to dive into some thought-provoking reads that can broaden your understanding of technology and artificial intelligence. Five books to make you think, laugh, and understand the tech world we are in