| |
| 🔗 Stories, Tutorials & Articles |
| |
|
| |
| March 20 ChatGPT outage: Here’s what happened |
| |
| |
ChatGPT had to be taken offline early this week because of a bug in an open source library that allowed users to see the chat history of other active users.
ChatGPT Plus subscribers may have had their personal information, including their name, email address, payment information, and last four digits of their credit card exposed.
The bug has been patched, and technical details will be published after an investigation. |
|
| |
|
| |
|
| |
| The best orchestration tool for MLOps: a real story about difficult choices ✅ |
| |
| |
The article discusses the use of orchestrators in automating machine learning processes and ensuring the reproducibility of results.
The three main orchestrators discussed are Kubeflow Pipelines, Airflow, and Prefect. The article compares the advantages and disadvantages of each orchestrator and concludes that the choice of orchestrator depends on the specific project requirements.
Kubeflow was the preferred choice for the authors due to its CI/CD configuration and containerization advantages. However, Prefect is recommended for small projects or teams with a limited budget, and Airflow is ideal for running a variety of tasks using a mature and developed tool. |
|
| |
|
|
| |
|
| |
| Speed up your model training with Vertex AI |
| |
| |
Learn how to speed up training of PyTorch and Hugging Face models using a reduction server through Vertex AI feature, optimizing bandwidth and latency to distribute training across multiple Nvidia GPUs.
The all-reduce based algorithm synchronously averages gradients across multiple devices, reducing training time and costs.
The post dives into a detailed description of datum parallelism and reduction server's architecture and impact on training throughput.
>>Check out the accompanying notebook and video to learn more! |
|
| |
|
| |
|
| |
| Unlocking Real-time Predictions with Shopify's Machine Learning Platform |
| |
| |
Shopify has enhanced its machine learning platform, Merlin, to include an end-to-end machine learning workflow. Merlin Online Inference is a flexible, robust, and low-latency solution that enables Shopify to deploy machine learning models in real-time at scale.
With the growing importance of close user-face product integration, it has become necessary to utilize machine learning models with low latency, which poses many different use cases and requirements.
Shopify uses its internal service ecosystem to make sure the service deployment is scalable and that the Merlin Online Inference service's health is observed. |
|
| |
|
|
| |
|
| |
| Lessons Learned Building Products Powered by Generative AI |
| |
| |
Buzzfeed shares its experience in incorporating generative AI product (like OpenAI's GPT) last year.
Buzzfeed says AI will bring a new era of creativity that opens up fresh content formats and novel ways for content creators to interact with audiences.
Buzzfeed staff used OpenAI playground to understand how language models work and to gain skills in creating an effective prompt. Through this, they learned moderation is essential to building guardrails around the prompt. |
|
| |
|
| |
|
| |
| 5 ways GPT-4 outsmarts ChatGPT ✅ |
| |
| |
OpenAI has released its latest language model, GPT-4, which has several new features that set it apart from previous versions like ChatGPT and GPT-3.5.
The five biggest differences are that GPT-4:- can understand and process images,
- is harder to trick,
- has a longer memory,
- is more multilingual,
- has different "personalities" that can be customized to suit specific needs.
GPT-4 is expected to be available to developers soon via API and will be featured on OpenAI's paid service, ChatGPT Plus. |
|
| |
|
| |
