#LoCallLm

A hands-on comparison of TextGen and LM Studio for running local LLMs, with migration steps and honest tradeoffs from real usage.

Day 5 of my 100-experiment local LLM challenge. Re-trained the Day 2 cat-LoRA with 45x the data. It got worse. Curated the data down to 213 truly-my-cat photos and re-trained — it came back to v1 quality. Then tested whether richer captions help further.…

Local LLMs in 2026 work on three hardware lanes: 32-core CPU with 64GB+ RAM hits 10-25 tokens per...

Day 4 of my 100-experiment local LLM challenge. Used CLIP + Qwen2-VL on a DGX Spark to classify and grade 25K iPhone photos. Overall agreement: 84.5%. People detection: 99.2%.

From Dev.to - machinelearning: LLM Quantization Explained: What Q4, Q5, and Q8 Actually Mean for Your GPU

Day 3 of my 100-experiment local LLM challenge. Threw 12 months of credit card data at Qwen 2.5 7B vs 72B on a DGX Spark. Takeaway: aggregate first with pandas, then let the LLM interpret.

Day 2 of my 100-experiment local LLM challenge. LoRA fine-tuning on 22 cat photos with Kohya_ss on a DGX Spark. 13 minutes of training, and "ohwx cat as a chef" now actually gives me my cat as a chef.

From Dev.to - ai: About Sharing Local Inference: A Marketplace for Renting Idle GPUs with an OpenAI-Compatible Backend

Day 1 of my 100-experiment local LLM challenge with NVIDIA DGX Spark. Unboxing, setup, an SSH-permission rabbit hole, and finally — generating cats with ComfyUI.

This is a story about how I built a fully local AI dev setup (and why you should too). Moving to...

From Dev.to - android: Run a Local LLM on Android: What RAM Tier You Need and Which Models Actually Work