Creating programs to correctly manipulate data is a difficult task, as the underlying programming languages and APIs can be challenging to learn for many users who are not skilled programmers. Large language models (LLMs) demonstrate remarkable potential for generating code from natural language, but in the data manipulation domain, apart from the natural language (NL) description of the intended task, we also have the dataset on which the task is to be performed, or the "data context". Existing approaches have utilized data context in a limited way by simply adding relevant information from the input data into the prompts sent to the LLM. In this work, we utilize the available input data to execute the candidate programs generated by the LLMs and gather their outputs. We introduce semantic reranking, a technique to rerank the programs generated by LLMs based on three signals coming the program outputs: (a) semantic filtering and well-formedness based score tuning: do programs even generate well-formed outputs, (b) semantic interleaving: how do the outputs from different candidates compare to each other, and (c) output-based score tuning: how do the outputs compare to outputs predicted for the same task. We provide theoretical justification for semantic interleaving. We also introduce temperature mixing, where we combine samples generated by LLMs using both high and low temperatures. We extensively evaluate our approach in three domains, namely databases (SQL), data science (Pandas) and business intelligence (Excel's Power Query M) on a variety of new and existing benchmarks. We observe substantial gains across domains, with improvements of up to 45% in top-1 accuracy and 34% in top-3 accuracy.

That is what SynctacticAI is: a platform-as-a-service, dedicated to AI and Machine learning.

Self-driving cars were once thought of as a far-off, and maybe even impossible, concept, but they're here now. At the end of November, General Motors announced its plan to launch a fleet of driverless cars -- without backup drivers -- across several major U.S. cities, beginning in 2019. In doing so, the auto industry signified that it's prepared to lead a dramatic shift in how both humans and commercial goods move from place to place. Of course, powering this initiative is data -- and likely exabytes of it. Driverless vehicles depend on data for everything from communicating their position on the road, to calculating speed and braking distances, to recognizing traffic signals and upcoming hazards in their path. While the idea of AVs whisking us away to our destinations with little to no effort does sound appealing on its own, an even greater prospect comes when you consider how such information will impact the travel experience itself.

Raviv Turner is Co-Founder and CEO at CaliberMind. On this episode, he and I discuss the challenges B2B marketers face in understanding the lengthy buyer's journey. It's long, it's complex, there are more people involved, and there is a ton of data spread across both structured and unstructured sources, making it difficult for B2B sales teams to accelerate their revenue goals. Raviv's career path includes being an entrepreneur, a product designer, a marketing technologist, and a Former Israel Signal Intelligence Community Officer. He applied the knowledge he learned about Big Data while working as an Intelligence Officer into his work building software products.