muthuishere
- Karma
- 0
- Created
- ()
- Submissions
- 0
- Show HN: Llama CPU Benchmarks (deemwar-products.github.io)
-
Most local LLM setups are not portable. This bundles model, embeddings, chunks and metadata into a single export. The exported package runs fully in the browser without internet or install. Trying to solve…
-
I like using Excalidraw for diagrams, especially the hand-drawn style. But I don’t like starting from a blank canvas. I usually spend 10–15 minutes placing boxes and arrows before I even get to the actual thinking. I…
-
https://deemwar.com/. Helping startups to grow
-
I run a very small VPS to host demos for my open source work. Traffic is minimal (maybe 10–20 users), so I assumed no one really cared. After checking the logs, I realized that assumption was wrong. Even with almost no…
-
Docker Desktop on macOS often sits in a VM that constantly sucks CPU. For years I tolerated it—until today, when running just two Node.js apps + one Python app pegged my system, froze the UI. I switched to Colima and…
-
You build tools. Shell scripts. Docker commands. Things you want to run from anywhere. But every OS has different rules: Mac/Linux: edit .bashrc, .zshrc, mess with PATH, source it Windows: create .bat, set up %PATH%,…
-
Encrypt in Java, decrypt in the browser (or seed JS-encrypted payloads into Java). Sounds straightforward. It’s not. You hit landmines like: Mismatched IV/tag defaults (Java Crypto API vs WebCrypto) Conflicting key…