The hottest Machine Learning Substack posts right now

And their main takeaways
Category
Top Business Topics

Building LLM-powered Apps: What You Need to Know

Gradient Flow • 519 implied HN points • 06 Apr 23
🕹 Technology Machine Learning
  1. Developers can now create AI-powered applications without deep machine learning knowledge, opening up opportunities for rapid experimentation and innovation.
  2. Building custom large language models (LLMs) is becoming more accessible through startups offering resources for model fine-tuning or training from scratch.
  3. Integration of custom LLMs with third-party services, utilizing knowledge bases, and serving models efficiently are key areas of focus for developers in the AI application space.

Use interpretability to improve and debug your ML model

Mindful Modeler • 279 implied HN points • 05 Dec 23
🕹 Technology Machine Learning
  1. Identify target leakage using feature importance to prevent accidental data pre-processing errors that leak target information into features.
  2. Debug your model by utilizing ML interpretability to spot errors in feature coding, such as incorrect signs on feature effects.
  3. Gain insights for feature engineering by understanding important features, and know which ones to focus on for creating new informative features.

Practical Prompt Engineering (Part One)

Deep (Learning) Focus • 373 implied HN points • 01 May 23
🕹 Technology Machine Learning
  1. LLMs are powerful due to their generic text-to-text format for solving a variety of tasks.
  2. Prompt engineering is crucial for maximizing LLM performance by crafting detailed and specific prompts.
  3. Techniques like zero and few-shot learning, as well as instruction prompting, can optimize LLM performance for different tasks.

The Sequence AI of the Week #781: The Amazing GLM 4.7

TheSequence • 28 implied HN points • 31 Dec 25
🕹 Technology Machine Learning
  1. GLM-4.7 is built to act like an "employee" rather than a chatty companion, prioritizing reliable task execution over conversational flair.
  2. Its architecture—mixing a mixture-of-experts design with a "Preserved Thinking" approach—is optimized for long-context loops, terminal error recovery, and stateful reasoning to handle real-world workflows.
  3. As an open-weight model focused on engineering and autonomous workflows, it’s positioned to become a standard choice for software development and task automation in 2026.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

Infinite Context Length 🤯

Sector 6 | The Newsletter of AIM • 99 implied HN points • 18 Apr 24
🕹 Technology Machine Learning
  1. Meta has introduced MEGALODON, a new neural architecture that allows for infinite context length in AI, making it more efficient than previous models.
  2. With developments from Microsoft, Google, and Meta, the focus will shift away from which model has the highest context length, as all will likely have infinite capabilities soon.
  3. The upcoming Llama-3 model is expected to continue this trend by also supporting infinite context length, enhancing its utility in various applications.

But have you considered writing your own evaluation metric?

Mindful Modeler • 299 implied HN points • 21 Nov 23
🕹 Technology Machine Learning
  1. Consider writing your own evaluation metric in machine learning to better align with your specific goals and domain knowledge.
  2. Off-the-shelf metrics like mean squared error come with assumptions that may not always fit your model's needs, so customizing metrics can be beneficial.
  3. Communication with domain experts and incorporating domain knowledge into evaluation metrics can lead to more effective model performance assessments.

GPTs Are Maxed Out

The Algorithmic Bridge • 647 implied HN points • 11 Nov 24
🕹 Technology Machine Learning
  1. AI companies are hitting limits with current models. Simply making AI bigger isn't creating better results like it used to.
  2. The upcoming models, like Orion, may not meet the high expectations set by previous versions. Users want more dramatic improvements and are getting frustrated.
  3. A new approach in AI may focus on real-time thinking, allowing models to give better answers by taking a bit more time, though this could test users' patience.

RAG Survey & Available Research

Cobus Greyling on LLMs, NLU, NLP, chatbots & voicebots • 39 implied HN points • 27 Jun 24
🕹 Technology Machine Learning
  1. Retrieval-Augmented Generation (RAG) mixes retrieval methods with learning systems to help large language models use real-time data.
  2. RAG can enhance the accuracy of language models by incorporating current information, avoiding wrong answers that might come from outdated knowledge.
  3. The framework of RAG includes steps like pre-retrieval, retrieval, post-retrieval, and generation, each contributing to better outputs in language processing tasks.

The Tech Buffet #20: How To deploy a Cloud Function That Summarizes Youtube Videos

The Tech Buffet • 139 implied HN points • 11 Mar 24
🕹 Technology Machine Learning
  1. Cloud Functions are a serverless way to run your code on Google Cloud without managing servers. You pay only for what you use, making it cost-effective.
  2. You can build a Cloud Function to summarize YouTube videos by extracting their transcripts and using AI to create concise summaries. This is done using Python libraries like youtube-transcript-api and langchain.
  3. Testing your Cloud Function locally is a great way to ensure it works before deploying it. You can use tools like Postman to check the API responses easily.

A new chapter on generalization

Mindful Modeler • 99 implied HN points • 16 Apr 24
🔬 Science Machine Learning
  1. Many COVID-19 classification models based on X-ray images during the pandemic were found to be ineffective due to various issues like overfitting and bias.
  2. Generalization in machine learning goes beyond just low test errors and involves understanding real-world complexities and data-generating processes.
  3. Generalization of insights from machine learning models to real-world phenomena and populations is a challenging process that requires careful consideration and assumptions.

TinyStories

Cobus Greyling on LLMs, NLU, NLP, chatbots & voicebots • 39 implied HN points • 26 Jun 24
🕹 Technology Machine Learning
  1. Phi-3 is a small language model that uses a special dataset called TinyStories. This dataset was designed to help the model create more varied and engaging stories.
  2. TinyStories uses simple vocabulary suitable for young children, focusing on quality over quantity. The stories generated are meant to be both understandable and entertaining.
  3. Training the Phi-3 model with TinyStories can be done quickly and allows for easier fine-tuning. This helps smaller organizations use advanced language models without needing huge resources.

Data Science Weekly - Issue 513

Data Science Weekly Newsletter • 359 implied HN points • 21 Sep 23
🕹 Technology Machine Learning
  1. There's a new newsletter focusing on AI safety in China, showing that the country is more invested in AI safety than many think.
  2. A podcast discusses how startups can run better AI models without needing to upgrade their hardware—a big challenge in the field.
  3. An online event is coming up for those looking to secure data science jobs in big tech, focusing on interview strategies and market insights.

Machine learning interpretability from first principles

Mindful Modeler • 359 implied HN points • 26 Sep 23
🕹 Technology Machine Learning
  1. Machine learning models can be understood as mathematical functions that can be broken down into simpler parts
  2. Interpretation methods address the behavior of these simplified components to enhance model interpretability
  3. Techniques like Permutation Feature Importance (PFI), SHAP values, and Accumulated Local Effect Plots use decomposition to explain the importance of features in prediction models

Understand AI or die trying

Technically • 43 implied HN points • 04 Dec 25
🕹 Technology Machine Learning
  1. Understanding how AI works is crucial to using it effectively. If you learn the basics, you can make AI a powerful tool instead of letting it take over your job.
  2. Many people use AI tools lazily and don’t take the time to understand how they work. This can lead to getting replaced if you’re not careful with your AI usage.
  3. There are resources available to help you learn about AI, and it's important to use them. The more you know, the better you can leverage AI in your work.

Data Science Weekly - Issue 537

Data Science Weekly Newsletter • 139 implied HN points • 07 Mar 24
🕹 Technology Machine Learning
  1. The newsletter shares valuable links about Data Science, AI, and Machine Learning each week. It's a great way to keep updated on the latest in the field.
  2. There are interesting articles highlighting statistical analyses and practical guides, like building GPU clusters at home. These resources help both beginners and experienced practitioners learn more.
  3. The newsletter also encourages people to participate in AI-related events and offers resources for job seekers. This can help you connect with others and grow your career.

Data Science Weekly - Issue 517

Data Science Weekly Newsletter • 339 implied HN points • 19 Oct 23
🕹 Technology Machine Learning
  1. Data science, AI, and ML are rapidly evolving fields, with new technologies and techniques emerging frequently. Staying updated through news and articles can help professionals keep their skills relevant.
  2. Fine-tuning large language models (LLMs) is a growing demand in the job market. Many companies are now looking for experience with LLMs alongside traditional skills like Python and SQL.
  3. Understanding different data visualization goals, like storytelling versus exploration, is important for effectively communicating data insights. This can improve how data is presented in reports and analyses.

Quick reflection on AI in 2024

Gonzo ML • 504 implied HN points • 02 Jan 25
🕹 Technology Machine Learning
  1. In 2024, AI is focusing on test-time compute, which is helping models perform better by using new techniques. This is changing how AI works and interacts with data.
  2. State Space Models are becoming more common in AI, showing improvements in processing complex tasks. People are excited about new tools like Bamba and Falcon3-Mamba that use these models.
  3. There's a growing competition among different AI models now, with many companies like OpenAI, Anthropic, and Google joining in. This means more choices for users and developers.

Data Science Weekly - Issue 509

Data Science Weekly Newsletter • 399 implied HN points • 25 Aug 23
🕹 Technology Machine Learning
  1. Each week, a newsletter shares important links and articles about data science, machine learning, and AI. It's a good way to keep updated on new happenings in the field.
  2. The newsletter features articles on various topics, including programming, AI forecasting, and data management practices. These articles are meant to help both newcomers and experienced professionals.
  3. Job listings and training resources are also provided, helping readers find opportunities and learn new skills beneficial for their careers in data science.

Biology and the evolution of AI tools

A Biologist's Guide to Life • 16 implied HN points • 17 Jan 26
🕹 Technology Machine Learning
  1. Major technological shifts mirror biological evolution: replication and innovation create new forms and disruptive functions that reshape systems over time.
  2. AI is a major economic transition driven by internet-scale data and modern neural networks, automating many digital tasks; its future will be shaped by competition for compute and users, technical advances like model compression, and cultural and legal responses.
  3. Individuals can adapt by learning to use AI as a practical sidekick to upskill and build new things, while being careful not to share sensitive information.

We need to do something about AI now

Philosophy bear • 486 implied HN points • 05 Jan 25
🕹 Technology Machine Learning
  1. AI is rapidly advancing and could soon take over many jobs, which might lead to massive unemployment. We need to pay attention and prepare for these changes.
  2. There's a real fear that AI could create a huge gap between a rich elite and the rest of society. We shouldn't just accept this as a given; instead, we should work towards solutions.
  3. To protect our rights and livelihoods, we need to build movements that unite people concerned about AI's impact on jobs and society. It's important to act before it’s too late.

DeepSeek moment

Gonzo ML • 441 implied HN points • 27 Jan 25
🕹 Technology Machine Learning
  1. DeepSeek is a game-changer in AI, trained models at a much lower cost compared to its competitors like OpenAI and Meta. This makes advanced technology more accessible.
  2. They released new models called DeepSeek-V3 and DeepSeek-R1, which offer impressive performance and reasoning capabilities similar to existing top models. These require advanced setups but show promise for future development.
  3. Their multimodal model, Janus-Pro, can work with both text and images, and it reportedly outperforms popular models in generation tasks. This indicates a shift toward more versatile AI technologies.

Report: OpenAI Spends Millions a Year Miscounting the R's in 'Strawberry'

The Algorithmic Bridge • 573 implied HN points • 22 Nov 24
🕹 Technology Machine Learning
  1. OpenAI has spent a lot of money trying to fix an issue with counting the letter R in the word 'strawberry.' This problem has caused a lot of confusion among users.
  2. The CEO of OpenAI thinks the problem is silly but feels it's important to address because users are concerned. They are also looking into redesigning how their models handle letter counting.
  3. Some employees joked about extreme solutions like eliminating red fruits to avoid the R issue. They are also thinking of patches to improve letter counting, but it's clear they have more work to do.

The Best Skillsets to Learn in 2024 for Generative AI

Rod’s Blog • 238 implied HN points • 15 Dec 23
🕹 Technology Machine Learning
  1. Generative AI is a rapidly evolving field creating novel content like images, text, music, etc., with real-world applications from enhancing creativity to helping solve problems.
  2. To succeed in generative AI, you need skills like mathematics and statistics, programming, data science, knowledge of generative AI methods, and creativity in your specific domain.
  3. To learn generative AI in 2024, leverage online courses, books, blogs, tools, and engage in communities and events dedicated to this field.

Unlocking the Future of Efficient AI Model Deployment

Gradient Flow • 339 implied HN points • 07 Sep 23
🕹 Technology Machine Learning
  1. Deep learning plays a key role in various industries, from healthcare to finance, with applications like computer vision and natural language processing being pervasive.
  2. Efficient AI model deployment involves crucial stages of model development, including domain-specific model refinement, and model optimization to ensure lightweight and fast models compatible with target hardware.
  3. Tools like Ivy are emerging to streamline the deployment of trained models, optimizing them for real-world use through techniques like enhanced graph representations, operator fusion, and quantization.

Data Science Weekly - Issue 514

Data Science Weekly Newsletter • 339 implied HN points • 29 Sep 23
🕹 Technology Machine Learning
  1. Data science involves a mix of techniques for analyzing and visualizing data which can help make informed decisions.
  2. Learning about advanced customer segmentation methods can enhance how businesses understand and target their customers.
  3. There are various roles in data-related careers beyond just being a data scientist, so it's good to explore different paths.

Data Science Weekly - Issue 519

Data Science Weekly Newsletter • 299 implied HN points • 03 Nov 23
🕹 Technology Machine Learning
  1. Companies are increasingly sharing their advanced AI models openly, which can help them improve and build better products. This open sharing can lead to a more cooperative tech environment.
  2. Data science job applications are extremely competitive, with many positions receiving thousands of applicants within a day. This shows a high interest and demand in the data science field.
  3. Exploring advanced tools and frameworks in AI can be complex, but understanding how they work can help in building effective applications, especially in question-answering systems.

RAG Implementations Are Becoming More Agent-Like

Cobus Greyling on LLMs, NLU, NLP, chatbots & voicebots • 99 implied HN points • 08 Apr 24
🕹 Technology Machine Learning
  1. RAG implementations are changing to become more like agents, which means they can make better decisions and adapt to different situations.
  2. The structure of prompts is really important now; it’s not just about adding data, but about crafting the prompts to improve how they perform.
  3. Agentic RAG allows for complex tasks by using multiple tools together, making it capable of handling detailed questions that standard RAG cannot.

SpatialBench: real world tasks for spatial agents

LatchBio • 54 implied HN points • 13 Nov 25
🕹 Technology Machine Learning
  1. SpatialBench offers a set of 98 evaluation packs to measure how well spatial agents perform on real tasks, helping to compare different technologies effectively.
  2. The evaluations are designed from actual tasks scientists face, making them useful to assess real-world analysis abilities in biology.
  3. There's a need for specialized tools and resources in biology since standard coding methods don’t easily translate to biological analysis tasks.

Scaling realities

Democratizing Automation • 562 implied HN points • 14 Nov 24
🕹 Technology Machine Learning
  1. Scaling in AI is technically effective, but the improvements visible to users are slowing down.
  2. There is a need for more specialized AI models, as bigger models may not always be the solution for current limits.
  3. There's still a lot of potential for new AI products and capabilities, which could unlock significant value in the future.

A Pragmatic View of Uncertainty in Machine Learning

Mindful Modeler • 359 implied HN points • 06 Jun 23
🕹 Technology Machine Learning
  1. Machine learning models have uncertainty in predictions, categorized into aleatoric and epistemic uncertainty.
  2. Defining and distinguishing between aleatoric and epistemic uncertainty is a complex task influenced by deterministic and random factors.
  3. Conformal prediction methods capture both aleatoric and epistemic uncertainty, providing prediction intervals reflecting model uncertainty.

Are AlphaFold's new results a miracle or a mirage?

Oleg’s Substack • 37 HN points • 24 Jun 24
🔬 Science Machine Learning
  1. AlphaFold 3 can predict how drug-like molecules bind to proteins better than existing programs without needing a 3D structure of the target.
  2. Data redundancy in scientific datasets can impact the performance and interpretation of machine learning models.
  3. AlphaFold 3's occasional missed obvious insights, like atoms overlapping, raises questions about its learning methods and performance.

The Sequence Opinion #778: After Scaling: The Era of Research and New Recipes for Frontier AI

TheSequence • 28 implied HN points • 25 Dec 25
🕹 Technology Machine Learning
  1. Scaling up transformers with more data and compute drove past AI gains, but that straightforward path is hitting limits because high-quality pretraining data and scaling efficiency are finite.
  2. The field is shifting to an "age of research" where diverse experiments and new ideas, not just bigger models, will determine future breakthroughs.
  3. Progress will come from a toolbox of new recipes — like souped-up pretraining, novel architectures, and improved fine-tuning — that turn compute into faster learning, better adaptation, and fewer odd model failures.

In The Context Of Long Context

Adjacent Possible • 553 implied HN points • 21 Nov 24
🕹 Technology Machine Learning
  1. A new AI feature can turn a whole book into a fun audio conversation, making learning more engaging. This feature has caught a lot of attention online and even received media coverage.
  2. The ability of the AI to handle large amounts of text—up to 1.5 million words—makes it much more useful for users, allowing for better, more detailed interactions.
  3. Long context models can help organizations make better decisions by recalling important documents and past experiences, adding a new kind of intelligence to team discussions.

Data Science Weekly - Issue 522

Data Science Weekly Newsletter • 259 implied HN points • 23 Nov 23
🕹 Technology Machine Learning
  1. This newsletter shares weekly interesting links and updates in data science, AI, and machine learning. It's a great way to stay informed about new developments in these fields.
  2. There's a focus on practical tools and techniques for improving data science work, like using cloud processing for large datasets and methods for fine-tuning AI models effectively.
  3. The newsletter also highlights job opportunities and resources for those looking to enter or advance in the data science industry. It's beneficial for anyone looking to grow their career in this area.

RAG Glossary

The Tech Buffet • 179 implied HN points • 21 Jan 24
🕹 Technology Machine Learning
  1. Retrieval Augmented Generation (RAG) helps AI answer questions and generate content. It combines searching through documents with generating relevant answers.
  2. Using RAG can be tricky, especially in production environments. Adjustments may be needed to improve reliability and performance.
  3. Different indexing methods can optimize how RAG retrieves information. This can make it more efficient and effective in finding the right data.