Join us

ContentUpdates and recent posts about Unsloth..
ย Activity
@alihenryofficial started using tool SAP Commerce Cloud , 1ย month, 2ย weeks ago.
ย Activity
@alihenryofficial started using tool Google Publisher Tag , 1ย month, 2ย weeks ago.
Story
@laura_garcia shared a post, 1ย month, 2ย weeks ago
Software Developer, RELIANOID

๐—ฒ๐—ป๐—ฎ๐—ฏ๐—น๐—ฒ ๐—ฆ๐—ฒ๐—ฐ๐˜‚๐—ฟ๐—ฒ ๐—•๐—ผ๐—ผ๐˜ ๐—ถ๐—ป ๐—ฅ๐—˜๐—Ÿ๐—œ๐—”๐—ก๐—ข๐—œ๐—— ๐—˜๐—ป๐˜๐—ฒ๐—ฟ๐—ฝ๐—ฟ๐—ถ๐˜€๐—ฒ ๐—˜๐—ฑ๐—ถ๐˜๐—ถ๐—ผ๐—ป

๐—ฆ๐—ฒ๐—ฐ๐˜‚๐—ฟ๐—ฒ ๐—•๐—ผ๐—ผ๐˜ + ๐—ฅ๐—˜๐—Ÿ๐—œ๐—”๐—ก๐—ข๐—œ๐—— = ๐˜ด๐˜ต๐˜ณ๐˜ฐ๐˜ฏ๐˜จ๐˜ฆ๐˜ณ ๐˜ต๐˜ณ๐˜ถ๐˜ด๐˜ต from the very first instruction executed. Here's a practical guide on ๐—ต๐—ผ๐˜„ ๐˜๐—ผ ๐—ฒ๐—ป๐—ฎ๐—ฏ๐—น๐—ฒ ๐—ฆ๐—ฒ๐—ฐ๐˜‚๐—ฟ๐—ฒ ๐—•๐—ผ๐—ผ๐˜ ๐—ถ๐—ป ๐—ฅ๐—˜๐—Ÿ๐—œ๐—”๐—ก๐—ข๐—œ๐—— ๐—˜๐—ป๐˜๐—ฒ๐—ฟ๐—ฝ๐—ฟ๐—ถ๐˜€๐—ฒ ๐—˜๐—ฑ๐—ถ๐˜๐—ถ๐—ผ๐—ป using the standard shim + MOK approach. ๐Ÿ” Whatโ€™s inside: - Why Secure Boot canโ€™t be enabled on first install - Step-by-step MOK enr..

ย Activity
@sanjayjoshi gave ๐Ÿพ to How To Make a Fast Dynamic Language Interpreter , 1ย month, 2ย weeks ago.
ย Activity
Story Trending
@laura_garcia shared a post, 1ย month, 2ย weeks ago
Software Developer, RELIANOID

๐—›๐—ฎ๐—ฐ๐—ธ ๐—ฆ๐—ฝ๐—ฎ๐—ฐ๐—ฒ ๐—–๐—ผ๐—ป ๐Ÿฎ๐Ÿฌ๐Ÿฎ๐Ÿฒ

๐Ÿš€ ๐—›๐—ฎ๐—ฐ๐—ธ ๐—ฆ๐—ฝ๐—ฎ๐—ฐ๐—ฒ ๐—–๐—ผ๐—ป ๐Ÿฎ๐Ÿฌ๐Ÿฎ๐Ÿฒ ๐Ÿ“ Kennedy Space Center ๐Ÿ“… May 6โ€“9, 2026 ๐™’๐™๐™š๐™ง๐™š ๐™˜๐™ฎ๐™—๐™š๐™ง๐™จ๐™š๐™˜๐™ช๐™ง๐™ž๐™ฉ๐™ฎ ๐™ข๐™š๐™š๐™ฉ๐™จ ๐™จ๐™ฅ๐™–๐™˜๐™š ๐™ž๐™ฃ๐™ฃ๐™ค๐™ซ๐™–๐™ฉ๐™ž๐™ค๐™ฃ. Hack Space Con is not your typical event โ€” itโ€™s where cybersecurity, aerospace, and advanced technologies converge to shape the future of security beyond Earth. ๐Ÿ” ๐—ช๐—ต๐—ฎ๐˜ ๐˜๐—ผ ๐—ฒ๐˜…๐—ฝ๐—ฒ๐—ฐ๐˜: - Hands-on techn..

HACKSPACECON2026_florida_RELIANOID
Link
@varbear shared a link, 1ย month, 2ย weeks ago
FAUN.dev()

A Couple Million Lines of Haskell: Production Engineering at Mercury

Mercury runs ~2M lines ofHaskellin production. They choseTemporalto replace cron and DB-backed state machines. Durable workflows replace brittle coordination. They open-sourced aHaskellSDK forTemporal, wired inOpenTelemetryhooks, and pushed records-of-functions plus domain-error types... read more ย 

A Couple Million Lines of Haskell: Production Engineering at Mercury
Link
@varbear shared a link, 1ย month, 2ย weeks ago
FAUN.dev()

Agentic Coding is a Trap

AI-driven coding agents are the hot new trend, but beware of the trade-offs: increased complexity, skills atrophy, vendor lock-in, and fluctuating costs. Only skilled developers can spot issues in the vast lines of generated code, but paradoxically, AI tools are impacting critical thinking skills ne.. read more ย 

Agentic Coding is a Trap
Link
@varbear shared a link, 1ย month, 2ย weeks ago
FAUN.dev()

When upserts don't update but still write: Debugging Postgres performance at scale

The Datadog team introduced a new upsert query to track inactive hosts, but it unexpectedly increased disk writes and WAL syncs due to row locking. By digging into Postgres's Write-Ahead Logging (WAL) and rewriting the query using a Common Table Expression (CTE), they avoided unnecessary overhead an.. read more ย 

Unsloth is an open-source toolkit for training and fine-tuning large language models faster and with less memory than a standard Hugging Face stack. Its core library replaces PyTorch's default autograd with custom backpropagation kernels written in OpenAI's Triton language, which is where most of its speed and memory savings come from. It supports LoRA, QLoRA, full fine-tuning, reinforcement learning, pretraining, and 4-bit, 16-bit, and FP8 training, across more than 500 text, vision, audio, and embedding models.

The practical draw is hardware reach. QLoRA workflows in Unsloth let you fine-tune an 8B model on a single 12 GB consumer GPU, and the project headlines roughly 2x faster training with about 70 percent less VRAM versus baseline implementations, though the exact figures vary by model, GPU, and config. A 2026 update added faster mixture-of-experts training, with models like Qwen3-30B-A3B fine-tunable on about 17.5 GB of VRAM. It runs on NVIDIA (including Blackwell and DGX Spark), AMD, and Intel GPUs, with free Colab and Kaggle notebooks for trying it without local hardware.

It fits cleanly into the local-AI workflow. Unsloth integrates with Hugging Face transformers and TRL, and uses llama.cpp to save and run models, exporting to GGUF for Ollama or LM Studio as well as safetensors. As of 2026 it also ships Unsloth Studio, a local no-code GUI that covers the full lifecycle from dataset creation to training to running and comparing GGUF and safetensors models, with tool-calling, web search, and an OpenAI-compatible API, all running offline on Mac and Windows, with the core library under the Apache 2.0 license.