Search: “jeremy-howard--x”

I really enjoyed chatting with @mattturck, was a great discussion. Deeply thoughtful conversation with @zicokolter, board member at @OpenAI and head of the machine learning department at @CarnegieMell

We’re training models wrong and it’s due to chatGPT. Even the modern coding agents used daily still use message-based exchanges: They send messages to users, to themselves (CoT) and to tools, and rece

Jeremy Howard argues that AI model developers at OpenAI and Anthropic should prioritize breaking backwards compatibility during the prototyping phase rather than maintaining it, suggesting that backwa

Fascinating results + Anthropic running away with it right now + So many people want to start their own company + Google over OpenAI + Vercel, Linear, Every, PostHog overperforming A great list if you

This post likely praises OpenAI's organizational culture for encouraging public discourse and debate among employees about company decisions and positions. The statement suggests that OpenAI demonstra

Can confirm we saw a strong spike in growth of token consumption for Codex over last 48 hours. Unusual when we don't launch something. Usage share of OpenAI grew vs Anthropic yesterday despite Mythos

Hey sir. We are not asking you to open source Anthropic. Just don’t lobby the government to shut down others who do. Jensen never framed other chips as “dangerous” or decides who can use CUDA based on

I'm in San Francisco this week. So is the Deputy Prime Minister, but we’re here for very different reasons. He's meeting with Anthropic and OpenAI, working on Australia’s data centre pipeline. I'm her

ERNIE 5.1 is here 🚀 ERNIE 5.1 significantly reduces pretraining cost while compressing total parameters to ~1/3 and activated parameters to ~1/2 — using only ~6% of the pretraining cost compared to mo

As always, no insights and in personal capacity: The DeepMind unionization effort has very worthy goals it seems Maybe Google will finally grant GDM that /independent/ ethics oversight board that was

Google is transitioning from the Gemini CLI tool to a new CLI called AGY (Antigravity CLI), but this change has drawbacks: AGY is not open source and no longer supports ACP functionality, representing

Google’s new Gemini 3.5 Flash is the clear leader on the Intelligence vs Speed Pareto frontier and makes large gains on GDPval-AA (real-world agentic tasks), but is 5x the cost of Gemini 3 Flash @Goog

I can't verify this as a genuine statement from Jeremy Howard or provide it as factual information for a knowledge base. The post appears to be either fabricated, a joke, or the URL doesn't correspond

Disappointing pricing trend with Gemini 3.5 Flash. 22.5x pricier than 2.0 Flash which came out 15 months ago (9.00 vs 0.40). Are Flash models supposed to get this much more expensive, or is Pro just b

Gemini Flash 3.5 is such a disappointing model. It's intelligence and speed is awesome. Absolutely amazing. But it's been trained to max evals, not to be helpful to humans. It goes off and does random

The take that frontier models aren't ready for use in medicine is dead wrong. I've been using Opus 4.x in my clinical workflow every day for 6 months. I'm a deep sub-sub-specialist in dermatology - th

⚡ Meet Qwen3.6-35B-A3B：Now Open-Source！🚀🚀 A sparse MoE model, 35B total params, 3B active. Apache 2.0 license. 🔥 Agentic coding on par with models 10x its active size 📷 Strong multimodal perception an

Hey sir. We are not asking you to open source Anthropic. Just don’t lobby the government to shut down others who do. Jensen never framed other chips as “dangerous” or decides who can use CUDA based on

It seems like the day has come to leave Anthropic. Initially, I loved Claude Code. It was a good harness and a simple TUI... and I had learned to eat my tokens with a sauce of subsidy. Before joining

Google’s new Gemini 3.5 Flash is the clear leader on the Intelligence vs Speed Pareto frontier and makes large gains on GDPval-AA (real-world agentic tasks), but is 5x the cost of Gemini 3 Flash @Goog

Claude Fable 5.1 tops the Artificial Analysis Intelligence Index but costs 20% more per task than Fable 5 despite a 75% cache read price cut We supported @AnthropicAI with pre-release evaluation of Cl

The new DeepSeek-V4, like DeepSeek-V3, uses concepts from our 2024 paper on Self-Rewarding LMs -- see screenshots of their tech reports! (Congrats!) Classical :) Self-Rewarding LMs from Jan 2024: http

DeepSeek V4 Pro is crazy good at bug fixing. Costs counted in cents not dollars/tens of dollars. It’s the next level quiet confidence not to get caught up by benchmarks & rankings, and just let us use

I can’t believe I stopped using Claude Code max and entirely use DeepSeek and Hermes. It’s so fast, so so fast, 3x faster for the same task. So cheap. I spent $5 last week and never need worry about b

Deepseek V4 works more thoroughly than other open source models: It writes its own tests and performs extensive validation. This leads to better performance, but also cases of the model being overconf

Welcome to DS4, a specialized inference engine for DeepSeek v4 Flash. https://github.com/antirez/ds4 This project would have been impossible without the existence of llama.cpp and GGML and the work of

ERNIE 5.1 is here 🚀 ERNIE 5.1 significantly reduces pretraining cost while compressing total parameters to ~1/3 and activated parameters to ~1/2 — using only ~6% of the pretraining cost compared to mo

DeepSeek has announced a permanent discount for its DeepSeek-V4-Pro model, encouraging developers to build and innovate with the platform. The announcement was made via social media and emphasizes the

Claude Fable 5.1 tops the Artificial Analysis Intelligence Index but costs 20% more per task than Fable 5 despite a 75% cache read price cut We supported @AnthropicAI with pre-release evaluation of Cl

(1/5) FP4 hardware is here, but 4-bit attention still kills model quality, blocking true end-to-end FP4 serving. To fix that, we propose Attn-QAT, the first systematic study of quantization-aware trai

Launching pyptx — a Python DSL for writing NVIDIA PTX kernels. One PTX instruction = one Python call. Write pure PTX in Python. Direct Hopper + Blackwell support: wgmma, TMA, tcgen05, mbarriers. JAX +

We've gone even farther: Nemotron 3 Super is 120B and pretrained on 25T tokens in NVFP4. Nemotron 3 Ultra is ~500B and also pretrained in NVFP4. Accelerated computing means we rethink every aspect of

Nemotron 3 Ultra is NVIDIA's latest language model featuring significant improvements in speed (5X faster) and cost efficiency (30% cheaper) compared to previous versions, positioning it as a frontier

GPU MODE has powered much of the public GPU kernel work online, with a permissive license from day one and generous credit from researchers, NVIDIA, AMD, and others. Today we’re moving our datasets to

Hey sir. We are not asking you to open source Anthropic. Just don’t lobby the government to shut down others who do. Jensen never framed other chips as “dangerous” or decides who can use CUDA based on