# aquin (audio feeds) — RSS Amplifier Recent posts from the 1 feeds in the RSS Amplifier directory that cover aquin. Page: Feed: --- ## [We Built Minds Before We Understood Them](https://aquinf03.substack.com/p/we-built-minds-before-we-understood) _2026-07-09 · Aquin Labs · Aquin Labs_ Every generation of technology has a moment where building stops being the hardest part. ## [Activation Steering for Bias Reduction](https://aquinf03.substack.com/p/activation-steering-for-bias-reduction) _2026-07-07 · Aquin Labs · Aquin Labs_ Activation Steering Can Reduce Bias, But Only If You're Honest About Its Limits ## [Training Sparse Autoencoders](https://aquinf03.substack.com/p/training-sparse-autoencoders) _2026-06-24 · Aquin Labs · Aquin Labs_ Dictionary learning over transformer activations: what the trainer sees, how feature quality scales with data, and how a trained SAE plugs into inspection, steering, and benchmarks. ## [Simulating Training](https://aquinf03.substack.com/p/simulating-training) _2026-06-01 · Aquin Labs · Aquin Labs_ No gradient-descent loop runs. Four analytical passes: SAE baseline, gradient landscape, LiSSA influence scoring, and NTK-linearised weight prediction ## [Aquin DevKit: SDK & CLI](https://aquinf03.substack.com/p/aquin-devkit-sdk-and-cli) _2026-05-23 · Aquin Labs · Aquin Labs_ https://pypi.org/project/aquin ## [Structuring Social Data for AI](https://aquinf03.substack.com/p/structuring-social-data-for-ai) _2026-05-22 · Aquin Labs · Aquin Labs_ Vivly.in × Aquin Labs Experiment Study ## [QLoRA Fine-Tuning of LLaMA 3.2-1B Instruct on a Healthcare Domain Dataset](https://aquinf03.substack.com/p/qlora-fine-tuning-of-llama-32-1b) _2026-05-21 · Aquin Labs · Aquin Labs_ Open-Source Repo ## [Embedding Models](https://aquinf03.substack.com/p/embedding-models) _2026-05-21 · Aquin Labs · Aquin Labs_ Geometry inspection, retrieval evaluation, fine-tuning monitoring, and embedding diff across checkpoints. Load any sentence-transformers compatible encoder and get the full picture of embedding space. ## [Transformers & LLMs](https://aquinf03.substack.com/p/transformers-and-llms) _2026-05-21 · Aquin Labs · Aquin Labs_ Aquin supports the full transformer family: dense LLMs and Mixture-of-Experts models. Every tool in the platform is architecture-aware from the moment you load a model. ## [The Security System](https://aquinf03.substack.com/p/the-security-system) _2026-05-21 · Aquin Labs · Aquin Labs_ Adversarial risk detection across the model checkpoint and the boundary between model versions. ## [Refer a Friend, Earn Rewards (Sponsored)](https://crawlproof.com/a/g38GEvSOrmvj) _2026-05-21 · **Sponsored**_ Earn rewards when a friend signs up for the Graphite Business Card ## [The Training Inspect System](https://aquinf03.substack.com/p/the-training-inspect-system) _2026-05-20 · Aquin Labs · Aquin Labs_ Real-time signal detection, behavioral before/after comparison, and per-layer feature diffs: everything the loss curve doesn't show you. ## [The Attribution System](https://aquinf03.substack.com/p/the-attribution-system) _2026-05-20 · Aquin Labs · Aquin Labs_ Seven tools that answer two questions: how did the model produce this output, and is the output actually correct? ## [The Eval System](https://aquinf03.substack.com/p/the-eval-system) _2026-05-20 · Aquin Labs · Aquin Labs_ Three behavioral evals that go beyond accuracy measuring whether a model answers consistently, what it quietly avoids, and where its knowledge runs out. ## [Benchmarks](https://aquinf03.substack.com/p/benchmarks) _2026-05-19 · Aquin Labs · Aquin Labs_ How Aquin scores and validates SAE features, and how the Benchmark Builder lets you measure model capability without leaving your inspection session.