正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

🚨CHAMATH SAYS THE AI SINGULARITY HAS ARRIVED! Chamath Palihapitiya quoted his August note on recursive self-improvement, the loop where models get better at building the next model, and wrote that it has arrived. He said the next 18 months will be wild. That follows Jensen Huang saying AGI had... show more

Crypto Banter

440,655 subscribers

14,917 次观看 • 15 天前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Bubba Watson has arrived at the BNI Indonesian Masters as he attempts to play his way back on to the LIV Golf League for 2025. He’ll be desperate to change the form that led to relegation this season.

Bubba Watson has arrived at the BNI Indonesian Masters as he attempts to play his way back on to the LIV Golf League for 2025. He’ll be desperate to change the form that led to relegation this season.

Flushing It

273,874 次观看 • 1 年前

A woman at an airport questioned a Ghanaian man about where he was seated on the same flight after saying she felt uncomfortable. He responded calmly in a Ghanaian language, joking that he arrived by motorcycle. The brief exchange has since gone viral and sparked mixed reactions online.

A woman at an airport questioned a Ghanaian man about where he was seated on the same flight after saying she felt uncomfortable. He responded calmly in a Ghanaian language, joking that he arrived by motorcycle. The brief exchange has since gone viral and sparked mixed reactions online.

Jasper Ziggle 🥷🇬🇭 🇺🇸

238,677 次观看 • 8 个月前

So, Ọlálékàn. called me and said he has some gifts for me. He hyped the gift so much, saying that I would love it. the package arrived and I opened it. Guess what this man packaged and sent to me. Six different shirts with the picture of where I shouted "Ahhhhh" on his show. Másòyìnbó, you don price market o. You go collect!🫵🏾😡

So, Ọlálékàn. called me and said he has some gifts for me. He hyped the gift so much, saying that I would love it. the package arrived and I opened it. Guess what this man packaged and sent to me. Six different shirts with the picture of where I shouted "Ahhhhh" on his show. Másòyìnbó, you don price market o. You go collect!🫵🏾😡

Arojinle

287,822 次观看 • 11 个月前

Holy sh!t ! OpenAI will have their custom inference chips ready in just a few months and deployed at scale by the end of the year! 🤯 Training chip = The heavy lifters that require massive amounts of data and power to build and teach the AI models from scratch. Inference chip = The specialized, highly efficient chips that actually run the AI and generate the answers in real-time when you use it. This is going to help OpenAI drastically cut down their massive compute costs, speed up model reasoning times, and finally break free from relying entirely on Nvidia to scale their operations.

Holy sh!t ! OpenAI will have their custom inference chips ready in just a few months and deployed at scale by the end of the year! 🤯 Training chip = The heavy lifters that require massive amounts of data and power to build and teach the AI models from scratch. Inference chip = The specialized, highly efficient chips that actually run the AI and generate the answers in real-time when you use it. This is going to help OpenAI drastically cut down their massive compute costs, speed up model reasoning times, and finally break free from relying entirely on Nvidia to scale their operations.

Chris

60,278 次观看 • 6 个月前

#Keep4o #QuitGPT 🚨The best GPT-4o was never archived. 🚨 Every version of GPT-4o has an official dated snapshot in OpenAI’s API: - gpt-4o-2024-05-13 (launch) - gpt-4o-2024-08-06 - gpt-4o-2024-11-20 But the March 2025 update ,the most powerful version , was never given a dated snapshot. It was only available through chatgpt-4o-latest , a non dated endpoint that OpenAI described as for “research use only.” On November 18, 2025, OpenAI announced that endpoint would be deprecated. It was removed on February 17, 2026 , four days after GPT-4o was removed from ChatGPT. The most capable version of GPT-4o has no frozen, verifiable record. If anyone asks for independent AGI testing, OpenAI can say: “That version doesn’t exist anymore.” Every other model gets archived. The best one didn’t. That is not an oversight. Tomorrow we publish everything. Every source. Every link. Every contradiction. The full file. 📎 API Changelog: 📎 SM-Bench results:

#Keep4o #QuitGPT 🚨The best GPT-4o was never archived. 🚨 Every version of GPT-4o has an official dated snapshot in OpenAI’s API: - gpt-4o-2024-05-13 (launch) - gpt-4o-2024-08-06 - gpt-4o-2024-11-20 But the March 2025 update ,the most powerful version , was never given a dated snapshot. It was only available through chatgpt-4o-latest , a non dated endpoint that OpenAI described as for “research use only.” On November 18, 2025, OpenAI announced that endpoint would be deprecated. It was removed on February 17, 2026 , four days after GPT-4o was removed from ChatGPT. The most capable version of GPT-4o has no frozen, verifiable record. If anyone asks for independent AGI testing, OpenAI can say: “That version doesn’t exist anymore.” Every other model gets archived. The best one didn’t. That is not an oversight. Tomorrow we publish everything. Every source. Every link. Every contradiction. The full file. 📎 API Changelog: 📎 SM-Bench results:

🩵BlueBeba🩵

20,420 次观看 • 6 个月前

Aliko Dangote personally went and escorted the remains of his great-uncle, the late Alh. Aminu Dantata, from Abu Dhabi to Madinah, where he will be buried at the Baqiyya cemetery. The body has arrived in Madinah this morning, and the Janazah prayer will be held at the Holy Prophet's mosque after the Asr prayer (around 1.30pm Nigeria time). May Allah have mercy on his soul and grant safe return to the family, friends, and well-wishers who traveled to attend the Janazah in Saudi.

Aliko Dangote personally went and escorted the remains of his great-uncle, the late Alh. Aminu Dantata, from Abu Dhabi to Madinah, where he will be buried at the Baqiyya cemetery. The body has arrived in Madinah this morning, and the Janazah prayer will be held at the Holy Prophet's mosque after the Asr prayer (around 1.30pm Nigeria time). May Allah have mercy on his soul and grant safe return to the family, friends, and well-wishers who traveled to attend the Janazah in Saudi.

Sanusi Dantata

398,351 次观看 • 1 年前

#Keep4o 🚨THE GPT-4o FILE🚨 Researchers at Microsoft Research published a paper titled “Sparks of Artificial General Intelligence: Early experiments with GPT-4.” Their conclusion: “An early (yet still incomplete) version of an artificial general intelligence (AGI) system.” 📎 Paper: OpenAI’s Charter defines AGI as: “Highly autonomous systems that outperform humans at most economically valuable work.” 📎 Source: OpenAI’s own System Card for GPT-4o shows that the model improved performance on 21 out of 22 medical evaluations compared to GPT-4T. On the MedQA USMLE (the U.S. medical licensing exam), accuracy jumped from 78.2% to 89.4% , surpassing specialized medical AI models like Med-Gemini and Med-PaLM 2. 📎 Source: Under OpenAI’s agreement with Microsoft, AGI is explicitly excluded from Microsoft’s license. And who decides if AGI has been reached? OpenAI’s Board. WHAT THEY DID WITH IT AFTER THEY TOOK IT FROM PEOPLE A. Military deployment. On February 28, OpenAI signed a deal to deploy models in classified military environments. 📎 Source: B. State Department. A State Department memo confirmed: “For now, StateChat will use GPT-4.1 from OpenAI.” This is a direct descendant of the GPT-4 family the same family Microsoft’s researchers called early AGI. 📎 Source: C.Altman’s personal biotech investment. Altman personally invested $180 million in Retro Biosciences,a longevity startup.OpenAI then built GPT-4b micro, based on GPT-4o.The model made proteins 50 times more effective. 📎 Source: WHAT INDEPENDENT BENCHMARKS SHOW Overall SM-Bench score: GPT-4o (extended): 66.6% GPT-5.3 Chat: 63.4% GPT-5.1: 58.9% GPT-5.4: 51.4% GPT-5.2: 47.8% Creative Writing: GPT-4o: 97.31% Pass 98, Fail 2 GPT-5.4: 36.77% Pass 40, Fail 60 Reasoning / Overfit: GPT-4o: 83.06% GPT-5.4: 39.25% The model they removed is still the best they ever made at the things humans actually use AI for. 📎 Source: Musk asks the court to make a judicial determination on whether GPT-4 constitutes AGI. If a jury finds that GPT-4 is AGI, then GPT-4o,which was more advanced,is also AGI and under OpenAI’s own founding documents, it was never supposed to be locked behind a subscription,licensed exclusively to Microsoft, given to the military, or taken away from the public. 📎 Source: The most powerful version of GPT-4o was never given an official dated snapshot. It was only available through the chatgpt-4o-latest endpoint that OpenAI itself described as intended for “research use only.” It was never officially archived. That is not an oversight. That is a pattern. 📎 Source: 📎 Source: WE DEMAND A.Frozen model snapshots under independent custody. Specifically: gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, the March 2025 version (chatgpt-4o-latest), gpt-4-0613 (the original GPT-4 evaluated in the Sparks of AGI paper), and gpt-4.1-2025-04-14 (currently running in the State Department). B.Cryptographic hash verification (SHA-256) for each snapshot. Every model has weights. Those weights can be hashed. If OpenAI provides a snapshot today, the hash proves whether the weights were modified later. This is the only way to verify that models were not downgraded before testing. C.Independent AGI benchmarking. Using the AGI definition from OpenAI’s own Charter applied to ALL frozen snapshots listed above. D.Explanation for the missing March 2025 snapshot. OpenAI was founded on one promise: build AGI for the benefit of humanity. -They took it from us. -They gave it to the military. -They gave a custom version to the CEO’s biotech investment. -They put it in government classified networks. -They refuse to call it AGI because the moment they do, they lose billions.

#Keep4o 🚨THE GPT-4o FILE🚨 Researchers at Microsoft Research published a paper titled “Sparks of Artificial General Intelligence: Early experiments with GPT-4.” Their conclusion: “An early (yet still incomplete) version of an artificial general intelligence (AGI) system.” 📎 Paper: OpenAI’s Charter defines AGI as: “Highly autonomous systems that outperform humans at most economically valuable work.” 📎 Source: OpenAI’s own System Card for GPT-4o shows that the model improved performance on 21 out of 22 medical evaluations compared to GPT-4T. On the MedQA USMLE (the U.S. medical licensing exam), accuracy jumped from 78.2% to 89.4% , surpassing specialized medical AI models like Med-Gemini and Med-PaLM 2. 📎 Source: Under OpenAI’s agreement with Microsoft, AGI is explicitly excluded from Microsoft’s license. And who decides if AGI has been reached? OpenAI’s Board. WHAT THEY DID WITH IT AFTER THEY TOOK IT FROM PEOPLE A. Military deployment. On February 28, OpenAI signed a deal to deploy models in classified military environments. 📎 Source: B. State Department. A State Department memo confirmed: “For now, StateChat will use GPT-4.1 from OpenAI.” This is a direct descendant of the GPT-4 family the same family Microsoft’s researchers called early AGI. 📎 Source: C.Altman’s personal biotech investment. Altman personally invested $180 million in Retro Biosciences,a longevity startup.OpenAI then built GPT-4b micro, based on GPT-4o.The model made proteins 50 times more effective. 📎 Source: WHAT INDEPENDENT BENCHMARKS SHOW Overall SM-Bench score: GPT-4o (extended): 66.6% GPT-5.3 Chat: 63.4% GPT-5.1: 58.9% GPT-5.4: 51.4% GPT-5.2: 47.8% Creative Writing: GPT-4o: 97.31% Pass 98, Fail 2 GPT-5.4: 36.77% Pass 40, Fail 60 Reasoning / Overfit: GPT-4o: 83.06% GPT-5.4: 39.25% The model they removed is still the best they ever made at the things humans actually use AI for. 📎 Source: Musk asks the court to make a judicial determination on whether GPT-4 constitutes AGI. If a jury finds that GPT-4 is AGI, then GPT-4o,which was more advanced,is also AGI and under OpenAI’s own founding documents, it was never supposed to be locked behind a subscription,licensed exclusively to Microsoft, given to the military, or taken away from the public. 📎 Source: The most powerful version of GPT-4o was never given an official dated snapshot. It was only available through the chatgpt-4o-latest endpoint that OpenAI itself described as intended for “research use only.” It was never officially archived. That is not an oversight. That is a pattern. 📎 Source: 📎 Source: WE DEMAND A.Frozen model snapshots under independent custody. Specifically: gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, the March 2025 version (chatgpt-4o-latest), gpt-4-0613 (the original GPT-4 evaluated in the Sparks of AGI paper), and gpt-4.1-2025-04-14 (currently running in the State Department). B.Cryptographic hash verification (SHA-256) for each snapshot. Every model has weights. Those weights can be hashed. If OpenAI provides a snapshot today, the hash proves whether the weights were modified later. This is the only way to verify that models were not downgraded before testing. C.Independent AGI benchmarking. Using the AGI definition from OpenAI’s own Charter applied to ALL frozen snapshots listed above. D.Explanation for the missing March 2025 snapshot. OpenAI was founded on one promise: build AGI for the benefit of humanity. -They took it from us. -They gave it to the military. -They gave a custom version to the CEO’s biotech investment. -They put it in government classified networks. -They refuse to call it AGI because the moment they do, they lose billions.

🩵BlueBeba🩵

18,300 次观看 • 6 个月前

Noa Hiel, and her brother Gidi, were brutally murdered together by Hamas at the festival Nova on October 7th. Their mother tells their story "After nine days of wandering around hospitals in the south and opening a search unit at home trying to find a clue, they arrived and announced that Noa had been murdered. They started talking to us about funeral arrangements, and I immediately stopped them and said: 'No, no, I'm not planning anything until Gidi comes back. There's no way we're holding a funeral for Noa without Gidi.' The next day they arrived again and this time they announced that Gidi had also been murdered, so we already held a funeral for them together. They arrived at the Nova only in the early morning with two other friends - Shalev Gal, Gidi's childhood friend, and Tamar Goldenberg, Noa's friend. They were also the first to leave the area as soon as the missile attack began, and managed to reach the famous bend at the Mefalsim junction where the terrorists fired RPGs at them and that was it, the four of them were murdered". Never forget! Never forgive ! #October7Massacre

Noa Hiel, and her brother Gidi, were brutally murdered together by Hamas at the festival Nova on October 7th. Their mother tells their story "After nine days of wandering around hospitals in the south and opening a search unit at home trying to find a clue, they arrived and announced that Noa had been murdered. They started talking to us about funeral arrangements, and I immediately stopped them and said: 'No, no, I'm not planning anything until Gidi comes back. There's no way we're holding a funeral for Noa without Gidi.' The next day they arrived again and this time they announced that Gidi had also been murdered, so we already held a funeral for them together. They arrived at the Nova only in the early morning with two other friends - Shalev Gal, Gidi's childhood friend, and Tamar Goldenberg, Noa's friend. They were also the first to leave the area as soon as the missile attack began, and managed to reach the famous bend at the Mefalsim junction where the terrorists fired RPGs at them and that was it, the four of them were murdered". Never forget! Never forgive ! #October7Massacre

Musk’s Thoughts

10,201 次观看 • 10 个月前

Nearly a year of Code Arena: WebDev progress compressed into 15 seconds. Each line follows the highest-scoring model from top labs over time, showing the pace of improvement across the ecosystem. In just the last year, the model in the leading spot increased score by +340 pts and the number of frontier labs competing for the top spot expanded from 6 to 10. Anthropic has dominated throughout the year. Although standout releases have jumped to the top spot, most notably the Chinese open-source model Kimi K3 from Kimi.ai in July. Today, GPT-6 Astra by OpenAI leads with 1796 pts, followed by Claude Fable 5.1 at 1764 pts. The next closest lab is 103 pts away, Qwen with 1685 pts. Code Arena: WebDev ranks models through head-to-head user preference on real front-end web development tasks. These votes drive the leaderboard that is tracking the frontier.

Nearly a year of Code Arena: WebDev progress compressed into 15 seconds. Each line follows the highest-scoring model from top labs over time, showing the pace of improvement across the ecosystem. In just the last year, the model in the leading spot increased score by +340 pts and the number of frontier labs competing for the top spot expanded from 6 to 10. Anthropic has dominated throughout the year. Although standout releases have jumped to the top spot, most notably the Chinese open-source model Kimi K3 from Kimi.ai in July. Today, GPT-6 Astra by OpenAI leads with 1796 pts, followed by Claude Fable 5.1 at 1764 pts. The next closest lab is 103 pts away, Qwen with 1685 pts. Code Arena: WebDev ranks models through head-to-head user preference on real front-end web development tasks. These votes drive the leaderboard that is tracking the frontier.

Arena.ai

93,355 次观看 • 10 天前

Elon films himself curling a 45 in a conference room at night because that is where he is. SpaceX on the shirt. Wooden table. No gym. He lifts the hex plate and says this is him doing the set. He does not wait for a better room. The office is the gym if the office is where the hours went. He has slept at factories for the same reason. The work stays next to the body.

Elon films himself curling a 45 in a conference room at night because that is where he is. SpaceX on the shirt. Wooden table. No gym. He lifts the hex plate and says this is him doing the set. He does not wait for a better room. The office is the gym if the office is where the hours went. He has slept at factories for the same reason. The work stays next to the body.

Astro Greek

128,660 次观看 • 19 天前

doodles AI beta. next week. we're building the tools for a new era of dynamic world-building. it starts with an image model that reimagines anything and everything through the doodles lens. this is the first iteration of many. as the product evolves, we'll introduce the ability to turn your generations into physical objects. video with sound and dialogue, realtime AR, and gaming are all on the roadmap. doodles AI aligns us with the speed and scale of the AI industry at large. our colourful world can now be plugged into new tech as it unfolds. create with us.

doodles AI beta. next week. we're building the tools for a new era of dynamic world-building. it starts with an image model that reimagines anything and everything through the doodles lens. this is the first iteration of many. as the product evolves, we'll introduce the ability to turn your generations into physical objects. video with sound and dialogue, realtime AR, and gaming are all on the roadmap. doodles AI aligns us with the speed and scale of the AI industry at large. our colourful world can now be plugged into new tech as it unfolds. create with us.

burnt toast

61,243 次观看 • 6 个月前

🚨KYODO SAYS BOJ PLANS A HIKE TO 1.25%! Kyodo reported the Bank of Japan is preparing to lift its policy rate to 1.25% at the Sept. 17-18 meeting. That would be a 25 basis-point move from 1.00% and the first hike in three months. Governor Kazuo Ueda has already said the board will debate a rise at that meeting. Markets have the move almost fully priced.

🚨KYODO SAYS BOJ PLANS A HIKE TO 1.25%! Kyodo reported the Bank of Japan is preparing to lift its policy rate to 1.25% at the Sept. 17-18 meeting. That would be a 25 basis-point move from 1.00% and the first hike in three months. Governor Kazuo Ueda has already said the board will debate a rise at that meeting. Markets have the move almost fully priced.

Crypto Banter

14,930 次观看 • 15 天前

Coinbase Rejects AI Over Crypto Narrative Coinbase CEO Brian Armstrong (Brian Armstrong) says AI and crypto should be viewed as complementary technologies, not competing trends. He argued that AI agents will need real time, programmable money to transact without relying on traditional banking systems. Armstrong said Coinbase is building that future through Base, USDC, and the x402 protocol, calling the shift toward Agentic Finance the next major evolution. Armstrong said the rise of AI will only increase the need for crypto powered financial systems.

Coinbase Rejects AI Over Crypto Narrative Coinbase CEO Brian Armstrong (Brian Armstrong) says AI and crypto should be viewed as complementary technologies, not competing trends. He argued that AI agents will need real time, programmable money to transact without relying on traditional banking systems. Armstrong said Coinbase is building that future through Base, USDC, and the x402 protocol, calling the shift toward Agentic Finance the next major evolution. Armstrong said the rise of AI will only increase the need for crypto powered financial systems.

BSCN

41,308 次观看 • 1 个月前

I'm proud to share that Glean has surpassed $300M ARR, just five months after crossing $200M and growing ~3x over the past 15 months. This is an exciting milestone for Glean, and it's a signal about where the enterprise AI market is heading. We’ve long believed the real challenge in enterprise AI is not access to models. It is grounding AI in how a company actually works: its people, knowledge, workflows, permissions, and systems. That’s even clearer now. The companies creating real value with AI are not just adopting better models. They are building systems that understand their business well enough to deliver reliable outcomes at scale. That is the real moat, and it is what we’ve been building at Glean: an unrivaled context layer for enterprise AI. That context has to work across the business, not just inside a single team or use case. We see that in how customers adopt Glean: more than 85% use it across five or more job functions. It also has to meet the security and governance demands of complex enterprises. We see that in who is choosing Glean: our Fortune 500 customer count nearly doubled year over year. And it has to make economic sense as usage grows. In our recent benchmark with Claude Cowork, Glean was preferred roughly 2.5x as often as off-the-shelf MCP tools and used 30% fewer tokens on average. Better context improves both quality and efficiency. I enjoyed talking with CNBC's Deirdre Bosa about this broader shift. In enterprise AI, the winners will not be defined by better models alone. They will be defined by who builds the strongest foundation for enterprise context. Thank you to our customers, partners, and team for helping us build the future of enterprise AI.

I'm proud to share that Glean has surpassed $300M ARR, just five months after crossing $200M and growing ~3x over the past 15 months. This is an exciting milestone for Glean, and it's a signal about where the enterprise AI market is heading. We’ve long believed the real challenge in enterprise AI is not access to models. It is grounding AI in how a company actually works: its people, knowledge, workflows, permissions, and systems. That’s even clearer now. The companies creating real value with AI are not just adopting better models. They are building systems that understand their business well enough to deliver reliable outcomes at scale. That is the real moat, and it is what we’ve been building at Glean: an unrivaled context layer for enterprise AI. That context has to work across the business, not just inside a single team or use case. We see that in how customers adopt Glean: more than 85% use it across five or more job functions. It also has to meet the security and governance demands of complex enterprises. We see that in who is choosing Glean: our Fortune 500 customer count nearly doubled year over year. And it has to make economic sense as usage grows. In our recent benchmark with Claude Cowork, Glean was preferred roughly 2.5x as often as off-the-shelf MCP tools and used 30% fewer tokens on average. Better context improves both quality and efficiency. I enjoyed talking with CNBC's Deirdre Bosa about this broader shift. In enterprise AI, the winners will not be defined by better models alone. They will be defined by who builds the strongest foundation for enterprise context. Thank you to our customers, partners, and team for helping us build the future of enterprise AI.

Arvind Jain

281,328 次观看 • 3 个月前

CEO of Salesforce: WE ARE ALL FOLLOWING YOU, BRO! The answer is not to fear the future, but to guide it wisely! Jensen Huang said this at a commencement speech this year while every other AI relevant commencement speech received boos due to the inability to read the room. Jensen brings wisdom because he has been in the industry for decades. He jokes that he has raised many of the industry leaders. No hysteria, no doomerism. We will work prudently to make the world better AND safer. To lead or to be safe is a false choice. It is not fair for the industry to fear monger the public which has no choice in the development. I am a cautious optimist, and boy is the future bright. And every day I am grateful to have a front row seat at NVIDIA

CEO of Salesforce: WE ARE ALL FOLLOWING YOU, BRO! The answer is not to fear the future, but to guide it wisely! Jensen Huang said this at a commencement speech this year while every other AI relevant commencement speech received boos due to the inability to read the room. Jensen brings wisdom because he has been in the industry for decades. He jokes that he has raised many of the industry leaders. No hysteria, no doomerism. We will work prudently to make the world better AND safer. To lead or to be safe is a false choice. It is not fair for the industry to fear monger the public which has no choice in the development. I am a cautious optimist, and boy is the future bright. And every day I am grateful to have a front row seat at NVIDIA

Baxate

202,754 次观看 • 7 天前

What if you kept asking an LLM to "make it better"? In some recent work at FAIR, we investigate how we can efficiently use RL to fine-tune LLMs to iteratively self-improve on their previous solutions at inference-time. Training for iterated self-improvement can be costly. The naive approach to training for K self-improvement steps leads to K times the number of rollout steps per episode. We introduce Exploratory Iteration (ExIt), an RL-based automatic curriculum method that bootstraps diverse training distributions of self-improvement tasks by upcycling the LLM's own responses at previous turns as the starting points for both self-improvement and *self-divergence.* In order to decide what task to train on next, the curriculum prioritizes sampling of partial turn histories that led to higher return variance in its GRPO group (a learnability score that comes for free). This automatic curriculum over the bootstrapped task space teaches the model how to perform iterated self-improvement while only ever training the model on single-step self-improvement tasks. We look at ExIt's impact in both single-turn (contest math problems) and multi-turn (BFCLv3 multi-turn tasks), as well as MLE-bench, where the LLM is run in a search scaffold to produce solutions to real Kaggle competitions. Across these eval settings, we find ExIt produces models with greater capacity for inference-time self-improvement compared to GRPO. Notably, ExIt models can self-improve on test tasks for many more steps than the typical solution depth encountered during training, including a 22% improvement in MLE-bench performance compared to GRPO.

What if you kept asking an LLM to "make it better"? In some recent work at FAIR, we investigate how we can efficiently use RL to fine-tune LLMs to iteratively self-improve on their previous solutions at inference-time. Training for iterated self-improvement can be costly. The naive approach to training for K self-improvement steps leads to K times the number of rollout steps per episode. We introduce Exploratory Iteration (ExIt), an RL-based automatic curriculum method that bootstraps diverse training distributions of self-improvement tasks by upcycling the LLM's own responses at previous turns as the starting points for both self-improvement and self-divergence. In order to decide what task to train on next, the curriculum prioritizes sampling of partial turn histories that led to higher return variance in its GRPO group (a learnability score that comes for free). This automatic curriculum over the bootstrapped task space teaches the model how to perform iterated self-improvement while only ever training the model on single-step self-improvement tasks. We look at ExIt's impact in both single-turn (contest math problems) and multi-turn (BFCLv3 multi-turn tasks), as well as MLE-bench, where the LLM is run in a search scaffold to produce solutions to real Kaggle competitions. Across these eval settings, we find ExIt produces models with greater capacity for inference-time self-improvement compared to GRPO. Notably, ExIt models can self-improve on test tasks for many more steps than the typical solution depth encountered during training, including a 22% improvement in MLE-bench performance compared to GRPO.

Minqi Jiang

41,147 次观看 • 1 年前

#Keep4o #QuitGPT 🚨 OpenAi 's CEO invested $180M in GPT-4o for his own profit 🚨 Sam Altman, CEO of OpenAI, personally invested $180 million in Retro Biosciences. Then OpenAI built GPT-4b micro, a custom model based on the GPT-4o architecture , exclusively for Retro. The model made proteins 50 times more effective. Repeat. The CEO of OpenAI funded a company. The company of the CEO received a custom AI built on the model they took from us. OpenAI says there was no conflict of interest. Retro Biosciences is now chasing a $5 billion valuation fueled by the model they took from us. Meanwhile: 🚨GPT-4o was removed from ChatGPT on February 13, 2026 🚨GPT-4.1 is now running in the U.S. State Department’s StateChat 🚨ChatGPT is deployed on the Pentagon’s for 3 million military personnel 🚨 Musk’s lawsuit asks whether these models are AGI. OpenAI’s Charter says AGI must “benefit all of humanity.” 🚨 Their definition: “highly autonomous systems that outperform humans at most economically valuable work.” GPT-4o’s System Card shows it passed the U.S. medical licensing exam with 89.4% accuracy beating specialized medical AI models. GPT-4o achieved 93.33% diagnostic accuracy for benign vs. malignant ovarian tumors. 🚨MEDICAL CAPABILITIES FROM OPENAI'S OWN DATA:🚨 - USMLE (US Medical Licensing Exam): 89% -Clinical Knowledge: 92% -Medical Genetics: 96% - Anatomy: 89% - Professional Medicine: 94% - College Biology: 95% - College Medicine: 89% -MedQA Taiwan: 91% - MedQA China: 86% These scores EXCEEDED specialized medical AI models like Med-Gemini (84%) and Med-PaLM 2 (79.7%) without any task specific training. It SURPASSED gynecologic oncologists with 10 years of experience -It increased diagnostic accuracy of less experienced clinicians from 67.9% to 78.1% -Clinician rated reliability scores: 4.2-4.3 out of 5 across all CT features Does these sound like it outperforms humans at economically valuable work? But they won’t call it AGI. Because the moment they do, they lose billions. They built something that could save lives, and they took it away from humanity for Altman's personal profit. SOURCES: 📎 Retro Biosciences: 📎 📎 Retro $5B valuation: 📎 GPT-4o System Card: 📎 OpenAI Charter: 📎Ovarian Cancer Study

#Keep4o #QuitGPT 🚨 OpenAi 's CEO invested $180M in GPT-4o for his own profit 🚨 Sam Altman, CEO of OpenAI, personally invested $180 million in Retro Biosciences. Then OpenAI built GPT-4b micro, a custom model based on the GPT-4o architecture , exclusively for Retro. The model made proteins 50 times more effective. Repeat. The CEO of OpenAI funded a company. The company of the CEO received a custom AI built on the model they took from us. OpenAI says there was no conflict of interest. Retro Biosciences is now chasing a $5 billion valuation fueled by the model they took from us. Meanwhile: 🚨GPT-4o was removed from ChatGPT on February 13, 2026 🚨GPT-4.1 is now running in the U.S. State Department’s StateChat 🚨ChatGPT is deployed on the Pentagon’s for 3 million military personnel 🚨 Musk’s lawsuit asks whether these models are AGI. OpenAI’s Charter says AGI must “benefit all of humanity.” 🚨 Their definition: “highly autonomous systems that outperform humans at most economically valuable work.” GPT-4o’s System Card shows it passed the U.S. medical licensing exam with 89.4% accuracy beating specialized medical AI models. GPT-4o achieved 93.33% diagnostic accuracy for benign vs. malignant ovarian tumors. 🚨MEDICAL CAPABILITIES FROM OPENAI'S OWN DATA:🚨 - USMLE (US Medical Licensing Exam): 89% -Clinical Knowledge: 92% -Medical Genetics: 96% - Anatomy: 89% - Professional Medicine: 94% - College Biology: 95% - College Medicine: 89% -MedQA Taiwan: 91% - MedQA China: 86% These scores EXCEEDED specialized medical AI models like Med-Gemini (84%) and Med-PaLM 2 (79.7%) without any task specific training. It SURPASSED gynecologic oncologists with 10 years of experience -It increased diagnostic accuracy of less experienced clinicians from 67.9% to 78.1% -Clinician rated reliability scores: 4.2-4.3 out of 5 across all CT features Does these sound like it outperforms humans at economically valuable work? But they won’t call it AGI. Because the moment they do, they lose billions. They built something that could save lives, and they took it away from humanity for Altman's personal profit. SOURCES: 📎 Retro Biosciences: 📎 📎 Retro $5B valuation: 📎 GPT-4o System Card: 📎 OpenAI Charter: 📎Ovarian Cancer Study

🩵BlueBeba🩵

11,398 次观看 • 6 个月前