This quote is especially important:
> "Results on AIME and GPQA are really strong, but that doesn’t necessarily translate to something that a user can feel. Even as someone working in science, it’s not easy to find the slice of prompts where GPT-4o fails, o1 does well, and I can
RESEARCH
-
Important quote: AIME and GPQA results strong but not user-visible
By
–
-
DeepPCB Pro: Reinforcement Learning for PCB Layout Design
By
–
Nice to see this launch of a Pro version of DeepPCB, an RL-based system for laying out PCB boards, which runs on @googlecloud. Nice work, @kbeguir and team! 🎉
— Jeff Dean (@JeffDean) 13 septembre 2024
The use of RL for many kinds of hardware design and layout problems is accelerating. I remember excitedly comparing… https://t.co/yhfTJHoGBcNice to see this launch of a Pro version of DeepPCB, an RL-based system for laying out PCB boards, which runs on @googlecloud
. Nice work, @kbeguir and team! The use of RL for many kinds of hardware design and layout problems is accelerating. I remember excitedly comparing -
DreamHOI: Zero-Shot 3D Human-Object Interaction Synthesis
By
–
DreamHOI Subject-Driven Generation of 3D Human-Object Interactions with Diffusion Priors paper page: https://
huggingface.co/papers/2409.08
278
… We present DreamHOI, a novel method for zero-shot synthesis of human-object interactions (HOIs), enabling a 3D human model to realistically interact -
Microsoft Windows Agent Arena: Evaluating Multi-Modal OS Agents
By
–
Microsoft presents Windows Agent Arena
— AK (@_akhaliq) 13 septembre 2024
Evaluating Multi-Modal OS Agents at Scale
discuss: https://t.co/SGWxR49gHN
Large language models (LLMs) show remarkable potential to act as computer agents, enhancing human productivity and software accessibility in multi-modal tasks that… pic.twitter.com/L85CqrJf5LMicrosoft presents Windows Agent Arena Evaluating Multi-Modal OS Agents at Scale discuss: https://
huggingface.co/papers/2409.08
264
… Large language models (LLMs) show remarkable potential to act as computer agents, enhancing human productivity and software accessibility in multi-modal tasks that -
Source2Synth: Synthetic Data Generation from Real Data Sources
By
–
Source2Synth Synthetic Data Generation and Curation Grounded in Real Data Sources discuss: https://
huggingface.co/papers/2409.08
239
… Large Language Models still struggle in challenging scenarios that leverage structured data, complex reasoning, or tool usage. In this paper, we propose -
TextBoost: One-Shot Personalization of Text-to-Image Models
By
–
TextBoost Towards One-Shot Personalization of Text-to-Image Models via Fine-tuning Text Encoder discuss: https://
huggingface.co/papers/2409.08
248
… Recent breakthroughs in text-to-image models have opened up promising research avenues in personalized image generation, enabling users to create -
Solar Pro Preview Outperforms Top Fine-Tuned LLMs and GPT-4
By
–
Across 30 tasks, @UpstageAI
's Solar Pro Preview fine-tuned outperforms every top fine-tuned open-source #LLM and #GPT4 and #GPT4o-mini base models! See more at our new Fine-Tuning Leaderboard: https://
pbase.ai/4gnsoaI
Try #SolarProPreview for free: https://
pbase.ai/3Zk1N8j -
o1-mini Achieves Surprising 60% Score on AIME Math Competition
By
–
o1-mini is the most surprising research result i've seen in the past year obviously i cannot spill the secret, but a small model getting >60% on AIME math competition is so good that it's hard to believe congrats @ren_hongyu @shengjia_zhao for the great work!
-
Gratitude toward artificial intelligence systems before accessing new capabilities
By
–
how about a couple of weeks of gratitude for magic intelligence in the sky, and then you can have more toys soon?
-
Rigorous Pre-Release Testing and Safety Evaluations Framework
By
–
We are also extremely proud of how rigorous and thorough our pre-release testing process is, including a full stack of frontier risk evaluations according to our Preparedness Framework and external red teaming. Read more safety work and evaluations in: