This article details the rediscovery of the source code for AM and EURISKO, two groundbreaking AI programs created by Douglas Lenat in the 1970s and early 80s. AM autonomously rediscovered mathematical concepts, while EURISKO excelled in VLSI design and even defeated human players in the Traveller RPG. Lenat had previously stated he no longer possessed the code, but it was found archived on SAILDART, the original Stanford AI Laboratory backup data, and in printouts at the Computer History Museum. The code was password protected until Lenat's passing, and has now been made available on Github.
The New Stack encourages its readers to contribute to Towards Data Science, a leading platform for data science and AI. Recognizing the increasing convergence of cloud infrastructure, DevOps, and AI engineering, the article invites practitioners to share their experiences with building and deploying AI systems. Successful TDS submissions are technically detailed, timely, and specific. Authors can also benefit from editorial support, promotion, and potential payment opportunities, while building their reputation within the AI community.
RFC 9457 defines a standardized format for communicating error details in HTTP API responses, known as "problem details."
- Supersedes RFC 7807,
- Core fields include 'type', 'title', 'detail' and 'instance'.
- Generally paired with 4xx/5xx HTTP status codes.
- Content types `application/problem+xml`, `application/problem+json`
Example:
```
{
"type": "https://example.com/probs/invalid-input",
"title": "Invalid Input",
"status": 400,
"detail": "The provided 'age' value must be a positive integer.",
"instance": "/users/123",
"age": -5
}
```
- Supersedes RFC 7807,
- Core fields include 'type', 'title', 'detail' and 'instance'.
- Generally paired with 4xx/5xx HTTP status codes.
- Content types `application/problem+xml`, `application/problem+json`
Example:
```
{
"type": "https://example.com/probs/invalid-input",
"title": "Invalid Input",
"status": 400,
"detail": "The provided 'age' value must be a positive integer.",
"instance": "/users/123",
"age": -5
}
```
This essay argues that the economics of context engineering expose a gap in the Brynjolfsson-Hitzig framework that changes its practical implications: for how enterprises build with AI, which firms centralize successfully, and whether the AI economy will be as centralized as their framework suggests. It explores how the cost and effort required to make knowledge usable by AI—context engineering—creates a bottleneck that prevents complete centralization, preserving the importance of local knowledge and human judgment. The article discusses the implications for SaaS companies, knowledge workers, and the future of work in an AI-driven economy, predicting that those who invest in context engineering capabilities will see the highest ROI.
The article details “autoresearch,” a project by Karpathy where an AI agent autonomously experiments with training a small language model (nanochat) to improve its performance. The agent modifies the `train.py` file, trains for a fixed 5-minute period, and evaluates the results, repeating this process to iteratively refine the model. The project aims to demonstrate autonomous AI research, focusing on a simplified, single-GPU setup with a clear metric (validation bits per byte).
* **Autonomous Research:** The core concept of AI-driven experimentation.
* **nanochat:** The small language model used for training.
* **Fixed Time Budget:** Each experiment runs for exactly 5 minutes.
* **program.md:** The file containing instructions for the AI agent.
* **Single-File Modification:** The agent only edits `train.py`.
* **Autonomous Research:** The core concept of AI-driven experimentation.
* **nanochat:** The small language model used for training.
* **Fixed Time Budget:** Each experiment runs for exactly 5 minutes.
* **program.md:** The file containing instructions for the AI agent.
* **Single-File Modification:** The agent only edits `train.py`.
Timer-S1 is a scalable Mixture-of-Experts time series model with 8.3B parameters that uses serial scaling and novel TimeMoE blocks to improve long-term forecasting accuracy.
We introduce Timer-S1, a strong Mixture-of-Experts (MoE) time series foundation model with 8.3B total parameters, 0.75B activated parameters for each token, and a context length of 11.5K. To overcome the scalability bottleneck in existing pre-trained time series foundation models, we perform Serial Scaling in three dimensions: model architecture, dataset, and training pipeline. Timer-S1 integrates sparse TimeMoE blocks and generic TimeSTP blocks for Serial-Token Prediction (STP), a generic training objective that adheres to the serial nature of forecasting. The proposed paradigm introduces serial computations to improve long-term predictions while avoiding costly rolling-style inference and pronounced error accumulation in the standard next-token prediction. Pursuing a high-quality and unbiased training dataset, we curate TimeBench, a corpus with one trillion time points, and apply meticulous data augmentation to mitigate predictive bias. We further pioneer a post-training stage, including continued pre-training and long-context extension, to enhance short-term and long-context performance. Evaluated on the large-scale GIFT-Eval leaderboard, Timer-S1 achieves state-of-the-art forecasting performance, attaining the best MASE and CRPS scores as a pre-trained model. Timer-S1 will be released to facilitate further research.
We introduce Timer-S1, a strong Mixture-of-Experts (MoE) time series foundation model with 8.3B total parameters, 0.75B activated parameters for each token, and a context length of 11.5K. To overcome the scalability bottleneck in existing pre-trained time series foundation models, we perform Serial Scaling in three dimensions: model architecture, dataset, and training pipeline. Timer-S1 integrates sparse TimeMoE blocks and generic TimeSTP blocks for Serial-Token Prediction (STP), a generic training objective that adheres to the serial nature of forecasting. The proposed paradigm introduces serial computations to improve long-term predictions while avoiding costly rolling-style inference and pronounced error accumulation in the standard next-token prediction. Pursuing a high-quality and unbiased training dataset, we curate TimeBench, a corpus with one trillion time points, and apply meticulous data augmentation to mitigate predictive bias. We further pioneer a post-training stage, including continued pre-training and long-context extension, to enhance short-term and long-context performance. Evaluated on the large-scale GIFT-Eval leaderboard, Timer-S1 achieves state-of-the-art forecasting performance, attaining the best MASE and CRPS scores as a pre-trained model. Timer-S1 will be released to facilitate further research.
This article explores the power of causal inference as a method for quantifying the impact of actions and improving decision-making, particularly in comparison to traditional A/B testing. It details how causal inference can provide a deeper understanding of customer behavior by estimating the individual impact of treatments (like marketing campaigns) and addressing the limitations of A/B testing, such as treating customer variability as noise and requiring large sample sizes. The article highlights applications in marketing, product recommendations, and customer retention, emphasizing benefits like customer segmentation, more precise estimates, and real-time learning. Ultimately, it argues that embracing causal inference can lead to more effective testing, optimized customer experiences, and shorter test cycles.
MIT researchers developed a new approach that rethinks how a classic method, known as Bayesian optimization, can be used to solve problems with hundreds of variables. In tests on realistic engineering-style benchmarks, like power-system optimization, the approach found top solutions 10 to 100 times faster than widely used methods.
Their technique leverages a foundation model trained on tabular data that automatically identifies the variables that matter most for improving performance, repeating the process to hone in on better and better solutions. The researchers’ tabular foundation model does not need to be constantly retrained as it works toward a solution, increasing the efficiency of the optimization process.
The technique also delivers greater speedups for more complicated problems, so it could be especially useful in demanding applications like materials development or drug discovery. The research will be presented at the International Conference on Learning Representations.
Their technique leverages a foundation model trained on tabular data that automatically identifies the variables that matter most for improving performance, repeating the process to hone in on better and better solutions. The researchers’ tabular foundation model does not need to be constantly retrained as it works toward a solution, increasing the efficiency of the optimization process.
The technique also delivers greater speedups for more complicated problems, so it could be especially useful in demanding applications like materials development or drug discovery. The research will be presented at the International Conference on Learning Representations.
Qwen3-Coder-Next is an 80-billion-parameter language model that activates only 3 billion parameters during inference, achieving strong coding capabilities through agentic training with verifiable task synthesis and reinforcement learning. It is an open-weight model specialized for coding agents, and both base and instruction-tuned versions are released to support research and real-world coding agent development.
Learn how to build a simple semantic search engine using sentence embeddings and nearest neighbors, focusing on the limitations of keyword-based search and leveraging large language models for semantic understanding.
