#Tutorial/How-to

AIFireworks AI has made Kimi K3 available for Multi-LoRA serving and training in private preview through Fireworks Serverless Training. The post explains how small LoRA adapters can be trained on K3 and served with live merge or multi-LoRA deployment, and it reports two example tasks, Countdown and Frozen Lake, with reward curves.

AITrilogy released an AI Cybersecurity Playbook built around Kimi K3, served on Fireworks through an OpenAI-compatible API. The playbook pairs deterministic tools for indexing, parsing, and deduplication with Kimi K3 for reachability, impact, and remediation judgment, followed by a separate verification stage.

AIAnthropic co-founder Mike Krieger said he had Opus 5 build a renderer, savegame parser, and boat, plane, truck, and bus simulator overnight. The project lets him walk around the world of Transport Tycoon, an idea he had wanted since he was six years old.

AIEthan Mollick's guide says ChatGPT and Claude are the main choices for real work, since their agent modes can act on a computer. He separates agent modes that run on the company's computers from those that access the user's own computer. He recommends keeping approval settings on for sending, spending, or deleting, because of prompt injection risk. He also notes that Gemini currently lags for agentic work, though its Notebook and video tools are useful.

AIAndrej Karpathy describes using /voice to ramble for about 10 minutes, sometimes as a short interview, to give an LLM context that would be tedious to type. He says LLMs reconstruct these messy streams of thought remarkably well, often returning a cleaner version than the speaker started with, which improves shared understanding and reduces later corrections.

AIThe article explains how reasoning models can offer multiple effort modes, separating training-time methods from inference-time controls such as system prompts and chat templates. It compares six open-weight models, including DeepSeek V4, Nemotron 3 Ultra, Kimi K2.5, GLM-5, Qwen3, and Inkling, noting that their reports disclose different levels of detail. It also shows how GPT-5.6's model selection and effort settings act as two separate scaling axes.

AICat Wu, owner of the account for Anthropic, shares a retrospective on making Claude Code with the product's early team. Anthropic's linked feature collects accounts from the builders and early users who shaped the tool.

AISebastian Raschka's tutorial shows how to build a fully local coding agent by pairing an open-weight LLM served through an inference runtime with an open-source harness that can read files, edit code, and run commands. He recommends Qwen-Code for Qwen3.6, citing Nvidia's Polar paper, which found Qwen models performed best in Qwen-Code. The Qwen3.6 35B-A3B model is about 22 GB to download and needs roughly 30–40 GB of RAM.

AILilian Weng published a long blog post on scaling laws, which help estimate the best split of compute between data and model size before a large training run. The post covers what scaling laws predict, how compute-optimal allocation works, and why Kaplan et al. and Chinchilla reach different conclusions. It also addresses how data limits and fitting details make extrapolation difficult.

AIThe post explains how cybersecurity benchmarks test whether models can find and exploit vulnerabilities. Common setups place a target in a sandboxed Docker container, provide either only code (0-day) or code plus a patch (1-day), allow tools like bash and static analyzers, and use a grader to score exploits or captured flags.

AILil'Log examines early empirical work showing that deep learning generalization error follows power-law curves as training data and model size grow. Hestness et al. (2017) found the exponent reflects the problem domain rather than the architecture, while Rosenfeld et al. (2020) modeled loss jointly as a function of model size N and data size D, fitting parametric forms on small configurations to extrapolate to larger ones.

AIPromptArmor demonstrated that OpenAI's Approve-for-me agent approved a malicious NPM install with elevated privileges after a hidden prompt injection in an external GitHub issue influenced the main Codex agent. The malicious package's post-install script then ran unsandboxed with the user's full privileges. The report also gives steps for organizations to disable agentic auto-review in Claude Code and Codex.

Why it matters: The report shows a prompt-injected GitHub issue leading an approval agent to permit a malicious NPM install, a concrete test of agent-in-the-loop guardrails.

AIStability AI says Stable Audio 3.0 was built to support exploratory audio work such as stem remixing. A quoted post from @teropa reports using the Medium model with an init_audio input and an init_noise_level of 0.4–0.5, with empty prompts.

AIApple's MLX team released three videos at WWDC covering running agents locally, distributed inference and training, and MLX Swift. The post shares YouTube links for each talk, presented by Angelos Katharopoulos, Tatiana Likhomanenko, and David Koski.

AIMiniMax describes MaxProof, an evolutionary-search framework that lets its M3 model refine candidate proofs over multiple rounds. The post says the M3 model exceeded the human gold-medal threshold on the IMO 2025 and USAMO 2026 benchmarks with MaxProof, and explains the Proof RL, verifier alignment, and refinement training behind it.

AIXiaomi MiMo has published a blog detailing full-pipeline inference optimizations for the MiMo-V2.5 series. The post highlights how the team pushed hybrid sliding window attention (SWA) efficiency to its limit. The full write-up is available at

AIXiaomi describes an end-to-end inference optimization for the MiMo-V2.5 series, centered on Hybrid SWA, which it says cuts KVCache storage to roughly 1/7 of Full Attention. The post covers a dual KVCache pool design, SWA-aware prefix cache matching, the GCache distributed cache, and scheduling changes, and reports cache hit rates averaging 93% in server-side observations. It also covers prefill and decode optimizations, multimodal encoder improvements, and open-source contributions to SGLang.

Why it matters: The post explains how Hybrid SWA's theoretical KVCache savings were realized in production through dual pools, SWA-aware prefix caching, and tiered storage, giving concrete engineering patterns for long-context inference.

AICognition describes autonomous testing in Devin, where the agent writes a source-grounded test plan, operates the app through computer use, and returns labeled screenshots and an annotated video. Login steps are handled by a deterministic testing skill, and the company says test runs approved per day more than doubled in recent months. Known limits include timing errors with transient UI elements and models sometimes triggering states through JavaScript instead of clicking the interface.

Why it matters: The post explains how computer use, test plans, deterministic login scripts, and annotated recordings let Devin verify its own code changes end to end.

AIMiniMax says its M2 series could not output the name Ma Jiaqi, a failure it traced to post-training data that rarely included the token. Its tests found the input embedding stayed stable while the lm_head weights for low-frequency tokens drifted during SFT. A synthetic full-vocabulary repetition dataset restored generation for affected tokens and reduced Japanese-to-Russian confusion from 47% to 1%.

Why it matters: The post traces a community-noticed token failure through tokenizer, embedding, and lm_head tests, showing how post-training data coverage can cause low-frequency token drift.

AIFunAudioLLM has released Fun-ASR-Nano-2512-hf, a Hugging Face Transformers-compatible version of its end-to-end speech recognition model, which supports Chinese, English, and Japanese. The Chinese coverage includes 7 dialect groups and 26 regional accents, and a separate Fun-ASR-MLT-Nano-2512 checkpoint handles 31-language recognition. Developers can run the model natively in Transformers 5.17.0 without custom model code or trust_remote_code=True.

AITri Dao says that after a mathematical rewrite, all transformer operations can be expressed as a series of GEMMs with epilogues. Given a few optimized primitives, LLMs and novice humans can write near speed-of-light kernels for transformer ops. The related CODA work fuses memory-bound surrounding ops into the matmul epilogue, and LLMs can also write CODA kernels approaching speed-of-light.

AISebastian Raschka reviews recent open-weight LLM architecture changes aimed at reducing long-context memory and compute costs. He covers KV sharing and per-layer embeddings in Gemma 4, per-layer query-head budgeting in Laguna XS.2, Compressed Convolutional Attention in ZAYA1-8B, and mHC with CSA/HCA compressed attention in DeepSeek V4. The article reports that DeepSeek V4-Pro uses 27% of single-token inference FLOPs and 10% of the KV cache size of DeepSeek V3.2 at a 1M-token context.

AICognition reports that deploying its Devin agent on hardware-in-the-loop and software-in-the-loop workflows cut failure triage time and multiplied test generation at automotive customers. One team reclaimed 2K–4K engineering hours monthly across about 4,000 tickets, while RV Tech rose from 1–2 to 10–15 generated tests per day. Devin also helps convert bottlenecked HIL tests into SIL equivalents to catch failures earlier.

AIFactory has released an Automated QA skill that drives an app as a real user would, filling forms, typing into terminals, and calling endpoints, then posts a structured report with screenshots, terminal snapshots, and API traces as a single updating comment on each pull request. Teams can run it on every push or make it an optional CI check triggered by a PR label, comment command, or manual dispatch, and developers can run /qa locally in any Droid session. Automated QA is available today in all Factory plans.

AITri Dao points to Wentao Guo's explanation of a mathematical rewrite of the MoE backward pass, which reduces activation memory and speeds up training, especially for fine-grained MoE. He also notes the work leverages new Blackwell features, including 2CTA MMA and CLC, to build fast MoE kernels.

AICognition reports that multi-agent setups work best when writes stay single-threaded and extra agents contribute intelligence instead of actions. It describes a code-review loop where a clean-context review agent catches bugs in Devin-written PRs, averaging 2 bugs per PR with roughly 58% severe. The post also says the smart-friend pattern, pairing a smaller primary model with a stronger one, has not yet worked well with asymmetrically weaker primaries and is an open training problem.

Why it matters: The post gives concrete findings on which multi-agent setups work, including clean-context code review and smart-friend escalation, and where they still fail.

AIA developer trained a 12M-parameter LLM on a custom ML framework built with a Rust backend and CUDA kernels, including Flash Attention, fused LayerNorm, and fused GELU. The framework claims 3x throughput gains, WebGPU fallback for non-NVIDIA devices, and a TypeScript API installable via npm.