『Vanishing Gradients』のカバーアート

Vanishing Gradients

Vanishing Gradients

著者： Hugo Bowne-Anderson
無料で聴く

A podcast for people who build with AI. Long-format conversations with people shaping the field about agents, evals, multimodal systems, data infrastructure, and the tools behind them. Guests include Jeremy Howard (fast.ai), Hamel Husain (Parlance Labs), Shreya Shankar (UC Berkeley), Wes McKinney (creator of pandas), Samuel Colvin (Pydantic) and more.

hugobowne.substack.comHugo Bowne-Anderson 科学
エピソード
  • The Rise of the AI Scientist

    The Rise of the AI Scientist
    2026/09/04
    “If you feel that a product is hard to eval, or you feel like, ‘I don’t even know how to eval this,’ it’s a strong smell that your product isn’t good.”— Hamel Husain, on AI product designAn AI data agent tells you last quarter’s net revenue. It doesn’t show the metric definition, source tables, filters, query, intermediate calculations, or assumptions. You can’t trust the answer without asking a data scientist to reproduce it.Hamel Husain argues that the eval problem is evidence of bad product design. If the user can’t inspect the work well enough to decide whether the answer is right, another scoring pipeline won’t rescue the experience. The fix begins by exposing the evidence and checks a domain expert actually uses.Follow that problem far enough and you arrive at a broader argument: AI isn’t killing data science. It’s creating more noisy, black-box systems that need hypotheses, experimentation, search expertise, and judgment. Hamel suggests that the people doing this work may eventually be called AI scientists.This episode connects those two ideas. Building AI products people can verify and understanding whether those products work are becoming part of the same job.“AI has made data science way more valuable than ever before, because now you have way more data and way more noisy signals that you need to reason about and debug.”— Hamel Husain, on the rise of the AI scientistYou can also find the full episode on Spotify, Apple Podcasts, and YouTube.👉 Want to build and evaluate AI agents that work in production? Hamel Husain and Shreya Shankar’s AI Evals For Engineers & PMs begins Sep 6, 2026. The course takes you from instrumenting an agent and inspecting traces through validated evaluators, regression testing, red teaming, and improving accuracy, latency, and cost. Vanishing Gradients viewers save 25% with the code hugo-2026. 👈In This Episode* The data agent everyone is building, and why a net-revenue answer without definitions, calculations, provenance, or uncertainty only creates more work.* Hamel’s case that AI has made data science more valuable by producing more traces, more nondeterministic output, and more noisy systems to understand.* How agents can learn from human annotations, improve sampling, and help validate LLM judges without taking human understanding out of the loop.* “Forget evals. Inspect ten traces.” What teams learn by starting with real failures instead of an evaluation framework.* Three AI products redesigned around the expert’s actual process: verifying a financial number, reviewing a workers’ compensation case, and adapting a trusted lesson plan.* Why generic skills have an upper limit, when sharing the shape of a skill works better, and how Hamel turns browser actions into a reusable API.* Shared human-agent canvases, WebMCP, and notebook-like interfaces for preserving evidence, experiments, and state during long-running work.* Your agent has 5,000 trace dimensions. How do hypotheses, exploratory analysis, and dimensionality reduction reveal which signals matter?* RAG is search, the right retrieval metric depends on the product, and choosing that metric still requires human judgment.* What an AI scientist might actually do: form hypotheses, choose analytical tools, design experiments, inspect failures, and decide whether an AI system is working.Resources* “It’s Hard to Eval” Is a Product Smell* The Revenge of the Data Scientist* Hamel’s reverse-engineered-site skill* Hamel’s guide to evaluating agentic workflows* Automating repetitive work at OpenAI with Codex* How Evals Are Central to Harness Engineering* AI Evals For Engineers & PMsListen or WatchYou can also find the full episode on Spotify, Apple Podcasts, and YouTube.👉 Want to build and evaluate AI agents that work in production? Hamel Husain and Shreya Shankar’s AI Evals For Engineers & PMs begins Sep 6, 2026. The course takes you from instrumenting an agent and inspecting traces through validated evaluators, regression testing, red teaming, and improving accuracy, latency, and cost. Vanishing Gradients viewers save 25% with the code hugo-2026. 👈How You Can Support Vanishing GradientsVanishing Gradients is an independent podcast, workshop series, blog, and newsletter about what people are building with AI and what survives contact with real users.* Become a paid subscriber* Share this episode with someone building an AI product* Subscribe to the Vanishing Gradients YouTube channel* Browse upcoming workshops Get full access to Vanishing Gradients at hugobowne.substack.com/subscribe
    続きを読む 一部表示
    1 時間 11 分
  • If Developers Build on Chinese Open-Weight Models, Who Leads AI?

    If Developers Build on Chinese Open-Weight Models, Who Leads AI?
    2026/08/03
    “It would be sad if local models were not an option and there were only proprietary models. It’s good to have alternatives. Competition is good for business.”— Sebastian Raschka, on open-weight AIKimi K3’s weights landed about an hour before Hugo Bowne-Anderson and Sebastian Raschka went live. Sebastian had already updated his architecture diagram. That speed captures his approach to the current model wave: wait until the weights exist, run the model in the harness where it will actually work, then inspect the architecture closely enough to understand what changed.The conversation arrived during a larger fight over who supplies the models underneath global software. Three days earlier, twenty-five companies including NVIDIA, Meta, Microsoft, Hugging Face, and IBM published Open Weights and American AI Leadership. Their argument closely matches Sebastian’s practical case for local models: open weights create competition, reduce dependence on a single provider, and let organizations choose a model at the right capability and cost.Update: Four days after we recorded, DeepSeek released V4 Flash 0731, a re-post-trained API model for agentic coding. Developers are already reporting that it can debug multi-project codebases and stay on task across very long contexts.You can find the full episode on Spotify, Apple Podcasts, and YouTube.👉 The next cohort of Master Agentic Data Science starts August 4. Hugo is teaching it with Thomas Wiecki and Luca Fiaschi from PyMC Labs. You will build agents that explore data, run predictive and causal workflows, challenge one another’s conclusions, and produce analysis humans can inspect and reproduce. Podcast listeners get 20% off with code MADSAGC20. Join the cohort. 👈In This Episode* Kimi K3’s weights landed about an hour before we went live. Sebastian had already updated his architecture diagram, then walked us through how Kimi Delta Attention and its mixture-of-experts design turn Kimi Linear’s research architecture into a production model.* When a local model is already good enough. Sebastian separates everyday coding and private workloads from problems where the strongest available model, extra reasoning time, and higher cost are justified.* Why the model and harness must be evaluated together. The same open-weight model behaves differently across Qwen Code, Claude Code, Codex, Pi, and other harnesses because context handling, compaction, tool conventions, and post-training all affect the result.* The harness should choose the model and its reasoning effort. Sebastian describes a router that uses the task, conversation history, context, and previous loops to make that decision automatically.* How long agent contexts are changing model architecture. Sebastian moves through grouped-query attention, multi-head latent attention, sparse attention, mixture-of-experts, DeltaNet, and Mamba-style hybrid layers.* The implementation does not lie. A from-scratch model can be checked against open reference weights tensor by tensor, and the first divergence exposes the missing RoPE scaling detail, RMSNorm variant, or architectural assumption that prose left out.* When fine-tuning earns its cost. A specialized task performed thousands of times a day creates different economics from a task performed eight times, so the decision depends on quality, volume, latency, and token cost.* Sebastian keeps the creative work and gives agents the tedious checks. He uses Codex and Claude to check links, enforce editorial conventions, and compare notebook code with the corresponding book manuscript.About Sebastian RaschkaSebastian Raschka is an independent AI researcher, author, and educator focused on large language models, reasoning models, deep learning, and practical machine-learning systems. He created Ahead of AI and is the author of Build a Reasoning Model (From Scratch) and Build a Large Language Model (From Scratch).Resources* Build a Reasoning Model (From Scratch): podcast listeners get 45% off the ebook or print edition with code 45VGraschka2.* Sebastian Raschka’s LLM architecture gallery* Kimi K3 in the architecture gallery* Build a Large Language Model (From Scratch)* Ahead of AI* Using Local Coding Agents* Open Weights and American AI Leadership* Anthropic’s position on open-weight modelsListen or WatchYou can find the full episode on Spotify, Apple Podcasts, and YouTube.👉 The next cohort of Master Agentic Data Science starts August 4. Hugo is teaching it with Thomas Wiecki and Luca Fiaschi from PyMC Labs. You will build agents that explore data, run predictive and causal workflows, challenge one another’s conclusions, and produce analysis humans can inspect and reproduce. Podcast listeners get 20% off with code MADSAGC20. Join the cohort. 👈How You Can Support Vanishing GradientsVanishing Gradients is a podcast, workshop series, blog, and newsletter focused on what you can build with AI right now. More than 70 episodes with expert practitioners ...
    続きを読む 一部表示
    1 時間 18 分
  • Four Months Inside a Production AI Agent: What Real Users Changed

    Four Months Inside a Production AI Agent: What Real Users Changed
    2026/07/25
    When ML/AI Engineer William Horton last joined me, Maven Assistant had reached its first external users the day before. The healthcare AI agent was available to 20 percent of Maven Clinic’s users, and the team had deliberately withheld answers about benefits. A wrong response could shape a decision involving $15,000 of fertility coverage, and the evals had not earned the right to ship it.Four months later, Maven Assistant is available to 100 percent of users, benefits answering is live, and weekly conversation volume has grown by roughly ten times. Real usage also overturned part of the roadmap. The team had invested heavily in provider search and appointment tools, but 50 to 60 percent of early conversations were basic health questions such as whether someone could eat tuna while pregnant.Production changed the engineering system too. An emergency guardrail told someone already in the ER to go to the ER. Zendesk content told people already using the Maven app to open the app. A newer model failed an upcoming-appointments eval because it correctly noticed that the mocked appointments were in the past.William explains how Maven turns those failures into deterministic tests, LLM judges, synthetic negatives, and manual review. He also walks through the move from Gemini Flash models toward newer OpenAI models, what GPT-5.6 and Fable mean for a production agent, why model upgrades can make old prompt instructions obsolete, and why open-weight models still have to justify their GPU, infrastructure, and engineering costs.“If anybody tells you that they’ve got their evaluations so good that they can just swap a model and know, with no manual review, that it’s going to be better, that person is probably lying, or they work at one of three places in the world.”— William Horton, Staff Machine Learning Engineer, Maven ClinicYou can also find the full episode on Spotify, Apple Podcasts, and YouTube.👉 Want to build agents from the ground up? Registration is open for Build AI Agents from First Principles, a live workshop on the loops, tools, context, harnesses, and engineering decisions behind useful AI agents. You’ll learn how to design agent systems from first principles, with enough structure to decide which harness patterns your product actually needs. Sign up today with code production10 for 10% off. 👈In This Episode* What changed between 20 percent and 100 percent rollout. Benefits answering cleared its release bar, Maven Assistant reached the remaining users, and weekly conversation volume grew by roughly ten times.* Why real usage beat the original roadmap. The complex provider and appointment agents received less traffic than expected, while 50 to 60 percent of early conversations were basic health questions.* How production failures enter the evaluation system. William estimates that the deterministic tool-use layer now involves more than 1,000 test scenarios, while clinical quality, empathy, and completeness still require judges, human calibration, and manual work.* When the model is right and the eval is broken. GPT-5.6 Terra rejected an “upcoming” appointment that had already happened, exposing a bad test fixture that the previous model had accepted.* What happens when GPT-5.6 or Fable arrives. A model swap can remove old prompt instructions, add new behavioral failures, change latency and cost, or reveal that yesterday’s harness is constraining a more capable model.* Why Maven moved beyond Gemini Flash 2.5. William discusses adopting newer OpenAI models, keeping real-time chat on smaller models and low reasoning settings, and changing the model without simultaneously rewriting the prompt.* The economics of open-weight models. A smaller self-hosted model still needs an always-on GPU, infrastructure, and engineering attention that could otherwise go into the product.* What William would rebuild today. Provider search and appointment booking probably belong in one agent, and model experiments should begin before the original choice hardens into the architecture.Start With the First EpisodeWilliam first joined Vanishing Gradients the day after Maven Assistant reached external users. In Building an Enterprise AI Agent for Healthcare, he explains the original architecture, how failures become regression cases, why deterministic checks should come before LLM judges, and how the consequence of a wrong answer sets the release bar.Resources* Maven Clinic* Building an Enterprise AI Agent for Healthcare* Stop Overengineering Your Agent Harness* Build AI Agents from First Principles* All Vanishing Gradients workshops👉 Want to build agents from the ground up? Registration is open for Build AI Agents from First Principles, a live workshop on the loops, tools, context, harnesses, and engineering decisions behind useful AI agents. You’ll learn how to design agent systems from first principles, with enough structure to decide which harness patterns your product actually needs. Sign up today...
    続きを読む 一部表示
    1 時間 5 分
adbl_web_anon_alc_button_suppression_t1
まだレビューはありません