Issue #2 - Friday February 11, 2022
What’s New @ NeuML publishes interesting content covering our open source projects, services and insights. The scheduled frequency is weekly to monthly.
Here we go!
💭 Tutorials
The following new txtai tutorial was published this week.
Push notifications with workflows
This notebook demonstrates how to build workflow notifications with txtai and Slack.
📄 Articles
Here is a list of recent articles relevant to our projects we found interesting.
Review of OpenAI GPT-3 Embeddings (by Nils Reimers)
Great empirical article by the author of Sentence Transformers, the main vectorization method used by txtai.PromptBERT (arXiv)
This research paper proposes a novel way to generate prompt-based sentence embeddings.Post message to Slack with cURL (slack.com)
Great tutorial on getting started with posting a message to Slack via the API. This article was used to help build the “Push notifications with workflows” tutorial above.SHAP (SHapley Additive exPlanations) (github.com/slundberg/shap)
If you want to learn more about model explainability, check out this library. The GitHub README is extremely informative and the examples easy to follow.
💡 Roadmap
As mentioned last week, work continues on txtai 4.2 with a focus on container-based and serverless workflows.
The recent and upcoming releases of txtai are incorporating functionality started in paperai, codequestion and tldrstory. This is getting close to being a reality. For example, paperai report queries are possible with workflows. Workflows can also solve most of the indexing and labeling found in tldrstory. While these libraries won’t go away, they will be refactored to eliminate duplicate code.
🔎 Where to find NeuML
In addition to this newsletter, NeuML can be found in the following places:
🏈 Super Bowl
Who do you have winning? No machine learning here - Bengals 28 Rams 21