The Daily Index — Wednesday, May 29, 2024 · TLDR Index

ARTICLES & TUTORIALS

TLDR Web Dev

TLDR Web Dev

WHAT WE LEARNED FROM A YEAR OF BUILDING WITH LLMS (PART I)

This is a comprehensive collection of ideas that helps dev work with LLMs better in production. For example, RAG (Retrieval-Augmented Generation) is great at improving LLM performance and is preferred over fine-tuning for adding new knowledge to a model's context. There are tips on prompting models better, such as using JSON or XML to structure inputs and outputs. There are also guidelines on evaluating and monitoring LLM I/O properly in areas where LLMs are in a production-level pipeline.

oreilly.com · 37 min

TLDR Web Dev

HOW LARGE IS A 160 BIT NUMBER?

Dolt, a version controlled database, uses cryptographic checksums to address data blocks. While this might lead to potential collisions where different data blocks have the same checksum, the chance of collision is extremely low due to the combo of SHA-512, a strong hash function, and a 20-byte key, which provides a collision resistance of 2^80. This means that even with billions of objects stored, the probability of a collision is small and negligible.

dolthub.com · 9 min

TLDR Web Dev

THE DISAPPOINTMENT FRONTIER

The disappointment frontier is the gap between a team's expectations and reality. This gap can grow when managers shield their teams from reality or overpromise on deliverables. Managing disappointment is a core leadership skill, so great managers acknowledge the disappointment frontier and actively work to shrink it.

theengineeringmanager.substack.com · 12 min

TLDR Web Dev

HIRING (AND MANAGING) CRACKED ENGINEERS

PostHog prioritizes hiring and retaining "cracked" engineers who are self-motivated and optimistic. The key to motivating and managing cracked engineers properly is to create an environment where they can collaborate with like-minded individuals, not have to deal with unnecessary bureaucracy, and can just focus on the work and the vision.

newsletter.posthog.com · 7 min

TLDR Web Dev

LLAMA 3-V: MATCHING GPT4-V WITH A 100X SMALLER MODEL AND 500 DOLLARS

Llama3-V is an open-source multimodal model created by Stanford ML researchers that's able to keep up with GPT4-Vision. The researchers accomplished this by combining the language capabilities of Llama 3 with a vision encoder from SigLIP, using a projection block for alignment, and prepending image tokens to textual tokens.

links.tldrnewsletter.com · 7 min

TLDR Web Dev

7 TIPS TO CRUSH YOUR ONBOARDING FROM AN APPLE STAFF ENGINEER

When onboarding, create a clear 30/60/90-day plan to set expectations and goals, network a lot, and pair with an onboarding mentor if possible.

open.substack.com · 10 min

TLDR Web Dev

HOW LONG WILL IT TAKE TO BUILD AN APP LIKE SPOTIFY WITH 3 DEVS? (REDDIT THREAD)

A very basic Spotify clone would probably take a few months to build, but actually building “all of Spotify” would take decades.

links.tldrnewsletter.com

TLDR Web Dev

ASK HN: WHY DO GAMES COMPANIES NOT RELEASE SOURCE CODE FOR OLD GAMES? (HACKER NEWS THREAD)

Companies may not have old source code anymore or it may be unclear who actually owns the source code and not worth a company's time to figure it out.

news.ycombinator.com