Register and share your invite link to earn from video plays and referrals.

Search results for GPT3
GPT3 community
One keyword maps to one global community path.
Create community
People
Not Found
Tweets including GPT3
Reading long form prose pre GPT3.5 is such a delight
On the robotics and physical AI VC market Over the last few weeks and months, there’s been a lot of fear in the markets. We believe those concerns are overblown. Let me pose a question - were OpenAI, Anthropic, Cursor, Coreweave, etc good investments to make in 2021/2022? The answer here is pretty clear cut. Were there AI companies from that period that weren’t good investments? Sure, but that’s the nature of venture and a small allocation to the top winners could still more than compensate for many poor picks. Look at this chart. US investments in Robotics has still not even reached pre-ChatGPT levels of investment in AI. At the same time, frontier Physical AI models are already more capable than GPT3.5 at the time ChatGPT came out. Roughly half of Anthropic/OpenAIs revenue comes from software development related use. Software development makes up 2-3% of the global labor market. Robotics represents physical work which is roughly 50% of the global labor market. While I’m very bullish on AI for knowledge work over the next few years, I don’t have a prediction of what happens over the next few weeks and months. Agents have only just started to become really useful for most people and that growth has yet to be realized. Could a continued short term market correction in AI spill over into private Physical AI/robotics? Short term maybe some funding rounds for lower quality companies take longer or don’t get done - there are some pools of capital that cross over. For the companies that otherwise are destined to win, we don’t think there will be much impact. Winners are going to win. Fluctuations in latent demand for automated physical labor for one shouldn’t change much and if they do, matters little given how little existing penetration and superior economics that robotic solutions provide. While I don’t have a strong macro view and am definitely not this bearish, the strongest example could be to look at the Great Depression. Tractors were invented before this period but roughly a fifth of the growth in tractor-owning farms between 1920 and 1954 came during the 1930s. "The Dust Bowl could blacken the sky, but it could not make the horse cheaper than the tractor." Companies that face idiosyncratic circumstances will have idiosyncratic results. Technological advances here are accelerating, not slowing and if companies can grow like weeds then we expect funding to follow even if Anthropic raises at $1.3T instead of $2T in its IPO. Most investors have only started dabbling in this sector and there are mountains of capital looking to get proper exposure. The IPOs of SpaceX and Anthropic are historically large liquidity events that we expect to propel continued downstream VC funding for months and years. Let me clear - there will always be a period where the wheat is separated from the chaff. Picking matters. Some capital has gone into winners but also some capital that has gone into losers and we will see the divergence in outcomes over the next few years. But with the backdrop of agentic robotic breakthroughs, Multimodel models accelerating R&D for physical AI while also proving a new approach of code as policy, and massively scaled data collection, our optimism for the sector is unwavering
Show more
GPT-3 level robot foundation models already exist It’s just not obvious because GPT-3 was only partially successful most of the time. That’s ok for some knowledge work but not for the work we need robots to do GPT-5 level intelligence for robots we could see within a year because research compounds We won’t necessarily have something that looks exactly like the ChatGPT moment for robotics, but the FigureAI livestream was close to that for a lot of people
Show more
Jev gives me early GPT 3.5 vibes. I have so many ideas that suddenly become doable
sometimes I wish I could ask gpt-3 to write me a haiku again
It's been 7 years from GPT-2 to GPT-6, and a lot has changed. The diagram below shows each generation and its key highlights, based on the public information we know. GPT-2 (2019): More like a prototype to validate the potential of the decoder-only transformer. This was a 1.5B-parameter model that only went through pre-training with next-token prediction. The trained model was only capable of continuing text, not answering questions. GPT-3 (2020): The same recipe as GPT-2, just much bigger and trained on more data to test how far scale alone could go. GPT-3 had 175B parameters, more than 100 times the size of GPT-2, and still only went through pre-training. The scale allowed it to show in-context learning capabilities. When shown some examples, the model could follow them. GPT-4 (2023): On top of pre-training, OpenAI fine-tuned it with RLHF (reinforcement learning from human feedback), so it was not just a completion model but could actually answer questions. It was also multimodal, meaning that it could process input images as well as text, and it handled up to 32K tokens of context. GPT-5 (2025): The first GPT generation with built-in reasoning. Given a prompt, tools, and an effort level, the trained model could generate reasoning tokens before the answer. Its context window grew to 400K tokens, with a router in ChatGPT to decide when a prompt needed thinking mode. GPT-6 (2026): This new model has a bigger context window (1.05M), supports async tool calling, and is generally safer. An interesting feature, called mid-turn steering, is that it can take a new instruction while it's thinking.
Show more
An old thread in 2020 talking about GPT 3
@fchollet Constraints in AI models like GPT-3 drive innovation in prompt engineering, revealing new applications in diverse workflows and problem-solving scenarios.
TL;DR • 2019: GPT-2 is "kind of cool" • 2020: a friend gets angry at GPT-3 • Wang sees something new, two years before ChatGPT • 2022: ChatGPT proves him right • 2025: Meta invests $14B+ in Scale; Wang runs Meta's AI • A toy amuses you. A real technology makes you angry
Show more
In our new benchmark AA-PersonalityBench, GPT-4o, Llama 4 Maverick, and GPT-3.5 Turbo, the three oldest models measured, rank highest on the Sycophantic personality trait. Grok 4.7, Claude Fable 5.1, and Inkling rank lowest ➤ @AIatMeta models are highly concentrated near the top: all five Meta models measured rank 65 or higher (Llama 4 Maverick 97, Muse Spark 1.3 87, Muse Glimmer 79, Muse Spark Open 76, Muse Wonder 65) ➤ @GoogleDeepMind models are spread out along the Sycophantic range: Gemini 3.1 Pro Preview ranks 89, Gemini 4 Argon 81, and Gemini 3.8 Flash 28, ranking from the top quartile to almost bottom quartile ➤ @AnthropicAI models skew on the lower end of the Sycophantic range: every Claude model measured ranks 55 or below (Claude Haiku 5.5 55, Claude Opus 5.5 46, Claude Sonnet 5.5 30 at medium effort and 25 at max, Claude Fable 5.1 4) ➤ @SpaceXSI's Grok 4.7 model ranks the lowest of all models measured. Congratulations @elonmusk!
Show more