yarapavan

> Because I wasn’t producing value, I felt I wasn’t valuable.

Every new technology tends to grow four layers: capabilities, interfaces, frameworks, and applications.

There does not seem to be a way of skipping or short-cutting around this process. As technologies grow through these stages, they become more accessible to wider audiences. Understanding how these four layers emerge can help in growing a durable and successful business.

1. The first layer, Capabilities, refers to the time when a new technological capability spends time waiting for its power to become accessible.

2. The second layer, Interfaces, emerges on top of the capability and can be an API or any other form of simplifying contract that enables more people to use the technology. The Interfaces layer serves as the democratizer of the Capabilities layer.

3. Third layer, the Frameworks, focuses on utility and asks how the underlying Interfaces layer can be utilized in more effective ways and made even more accessible to an even broader audience. The Frameworks layer tends to be the most opinionated of the stack due to the many definitions of utility and many possible ways to achieve it. The diversity of opinion introduced in the Frameworks layer depends on two factors: the inherent value of the capability and the own opinion of the Interfaces layer. The Frameworks layer becomes the de-facto place where best practices and patterns of applying the capability are developed and stored.

4. Fourth layer, Applications, is where the technology finally faces its users – the consumers of the technological capability. These consumers might be end users who aren’t technology-savvy or another group of developers who are relieved to not have to think about how the technology works on the inside. The pressure toward maximizing utility develops at this layer. Consumer-grade software is serious business and it often takes all available capacity to just stay in the game. The whole reason the Frameworks layer exists is to unlock efficiency and further scale the availability of the technology.

This is exactly what is playing out with the large language models. While ChatGPT is getting all the attention, the actual interesting work is happening at the Frameworks layer that sits on top of the large language model Interfaces layer: the OpenAI, Anthropic, and PaLM APIs.

Techniques - https://www.thoughtworks.com/radar/techniques

Tools - https://www.thoughtworks.com/radar/tools

Platforms - https://www.thoughtworks.com/radar/platforms

Languages & Frameworks - https://www.thoughtworks.com/radar/languages-and-frameworks

- "ChatGPT’s performance in Standard-IE settings is not as good as BERT-based models in most cases. However, ChatGPT achieved excellent accuracy scores in the OpenIE setting, as evaluated by human annotators."

- "ChatGPT could provide high-quality and trustworthy explanations for its predictions. One of the key issues is its tendency towards overconfidence, resulting in low calibration."

- "ChatGPT exhibits a high level of faithfulness to the original text, indicating that its predictions are grounded in the input text."

The new version is a follow-up of the company’s first version, which generally became available last year. With version 2.0, the company added several new features and improvements, such as entirely new Lambda and S3 providers, a significant reduction in LocalStack image size through the separation of LocalStack Community and Pro Docker images, a new Snapshot persistence mechanism, community cloud pods, and cloud pods launchpad.

Furthermore, version 2.0 includes enhancements for developers like improved LocalStack toolings for local cloud development, new LocalStack Developer Hub and Tutorials, and Improved LocalStack Coverage Docs Overview.

The article talks about an American journalist’s experience inside TSMC, a Taiwanese company that is at the center of the global semiconductor industry.

Interesting snippets:

1. The company has been around since 1987 and has grown to become one of the largest semiconductor manufacturers in the world. By revenue, TSMC is the largest semiconductor company in the world. It’s now bigger than Meta and Exxon.

2. TSMC produces 92 percent of the world’s most avant-garde chips. It makes a third of all the world’s silicon chips, notably the ones in iPhones and Macs (quintillion transistors for Apple)

3. TSMC has a reputation for being secretive and mysterious, which has only added to its allure.

4. Perks: employees get a 10 percent discount at Burger King. 10% at other outlets too(?)

5. Two qualities set the TSMC scientists apart: curiosity and stamina. Religion, to the author's surprise, is also common. “Every scientist must believe in God,” Liu says.

6. “If you think about conflicts around Taiwan,” Tooze told Klein, “the global semiconductor industry isn’t just the supply chain. It’s one of humanity’s great technological scientific achievements. Our ability to do this stuff at nanoscale is us up against the face of God, in a sense.” When, later, I recite Tooze’s words about God’s face to Mark Liu, he quietly agrees, but refines the point. “God means nature. We are describing the face of nature at TSMC.”

csvkit/miller - for working with csv, tsvs

rg/fzf - for search and fuzzy finding

The intended audience of this doc is the recently laid-off, or those who suspect they are shortly to be, though a number of others have found it useful outside of that context

1. In January, about eight billion transactions worth nearly $200 billion were carried out on the U.P.I.

2. The value of instant digital transactions in India last year was far more than in the United States, Britain, Germany and France. “Combine the four and multiply by four — it is more than that" per a cabinet minister

3.The system has grown rapidly and is now used by close to 300 million individuals and 50 million merchants

Speedster: Automatically apply the best set of SOTA optimization techniques to achieve the maximum inference speed-up on your hardware. https://github.com/nebuly-ai/nebullvm/blob/main/apps/acceler...

Nos: Automatically maximize the utilization of GPU resources in a Kubernetes cluster through real-time dynamic partitioning and elastic quotas. https://github.com/nebuly-ai/nos

ChatLLaMA: Build faster and cheaper ChatGPT-like training process based on LLaMA architectures. https://github.com/nebuly-ai/nebullvm/tree/main/apps/acceler...

OpenAlphaTensor: Increase the computational performances of an AI model with custom-generated matrix multiplication algorithm fine-tuned for your specific hardware. https://github.com/nebuly-ai/nebullvm/tree/main/apps/acceler...

Forward-Forward: The Forward Forward algorithm is a method for training deep neural networks that replaces the backpropagation forward and backward passes with two forward passes. https://github.com/nebuly-ai/nebullvm/tree/main/apps/acceler...

Do you have any perf numbers, in terms of size and response times? Is there a list of file formats you support? Possible to choose the LLM model as my preference? How does pricing looks like?

Again, great execution and useful tool. Thank you for the launch and good luck!

Companies are looking at other, more cost-effective options, including managed service providers and co-location providers (colos), or even moving those systems to the old server room down the hall. This last group is returning to “owned platforms” largely for two reasons.

First, the cost of traditional compute and storage equipment has fallen a great deal in the past five years or so. If you’ve never used anything but cloud-based systems, let me explain. We used to go into rooms called datacenters where we could physically touch our computing equipment — equipment that we had to purchase outright before we could use it. I’m only half kidding.

When it comes down to renting versus buying, many are finding that traditional approaches, including the burden of maintaining your own hardware and software, are actually much cheaper than the ever-increasing cloud bills.

Second, many are experiencing some latency with cloud. The slowdowns happen because most enterprises consume cloud-based systems over the open internet, and the multi-tenancy model means that you’re sharing processors and storage systems with many others at the same time. Occasional latency can translate into many thousands of dollars of lost revenue a year, depending on what you’re doing with your specific cloud-based AI/ML system in the cloud.