ChatGPT applications, opportunities, and threats.
arxiv.org/abs/2304.09103
#مقاله #مبتدی
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
arxiv.org/abs/2304.09103
#مقاله #مبتدی
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
👍2
My lab is hiring a full-time research staff position! We are looking for someone with experience in neuroimaging analysis to assist in studies of brain-behavior evolution in dogs, foxes, primates, and humans. Come join our team! bit.ly/61689BR
👍1
Evaluating Verifiability of LLM-Powered Search Engines
-Human eval of Bing Chat, NeevaAI, perplexity ai, YouChat
-Responses are fluent but frequently contain unsupported statements & inaccurate citations
-51.5% of sentences fully supported by citations
arxiv.org/abs/2304.09848
-Human eval of Bing Chat, NeevaAI, perplexity ai, YouChat
-Responses are fluent but frequently contain unsupported statements & inaccurate citations
-51.5% of sentences fully supported by citations
arxiv.org/abs/2304.09848
وبسایت Shecodes به ایرانیا بورسیه میده علاقمندان براش اقدام کنن
shecodesfoundation.org/candidates/new
#خبر #برنامه_نویسی
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
shecodesfoundation.org/candidates/new
#خبر #برنامه_نویسی
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
👍1
بنا به گزارش Checkpoint هکرهای زیادی با استفاده از حملات brute-force در تلاش هستند تا اکانتهای پرمیوم ChatGPT رو #هک کرده و در بازارهای زیرزمینی به هکرهایی که بدنبال استفاده از این فناوری جهت توسعه بدافزار و انواع سلاحهای #سایبری هستند بفروشند.
https://blog.checkpoint.com/security/new-chatgpt4-0-concerns-a-market-for-stolen-premium-accounts/
#خبر #هوش_مصنوعی
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
https://blog.checkpoint.com/security/new-chatgpt4-0-concerns-a-market-for-stolen-premium-accounts/
#خبر #هوش_مصنوعی
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
👍3
مرجعی کامل برای دسترسی به مقالات فراگیری پرامپت انجینرینگ اینجا یادبگیرید
survey papers on LLMs and prompt engineering.
https://www.promptingguide.ai/papers
Second resources :
https://chatx.ai/marketplace/
#مبتدی #مقاله
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
survey papers on LLMs and prompt engineering.
https://www.promptingguide.ai/papers
Second resources :
https://chatx.ai/marketplace/
#مبتدی #مقاله
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
👍1
جامعتر مرجع برای محبث
RL with Human Feedback
تا کنون چه مقالاتی در این حیطه منتشر شده و آینده این مباحث رو میتوانید ایده های بیشتری که منتشر شده،ایده دیگری را بدست بیارید
https://github.com/opendilab/awesome-RLHF
#مبتدی #مقاله
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
RL with Human Feedback
تا کنون چه مقالاتی در این حیطه منتشر شده و آینده این مباحث رو میتوانید ایده های بیشتری که منتشر شده،ایده دیگری را بدست بیارید
https://github.com/opendilab/awesome-RLHF
#مبتدی #مقاله
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
Reference-based Image Composition with Sketch via Structure-aware Diffusion Model
abs: arxiv.org/abs/2304.09748
#مقاله
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
abs: arxiv.org/abs/2304.09748
#مقاله
🔸 مطالب بیشتر 👇👇
✅ @AI_DeepMind
👍1
در مقاله ای که اخیرا توسط تیم DeepMind و تیم گوگل منتشر شده است
GopherCite: Teaching language models to support answers with verified quotes
https://www.deepmind.com/publications/gophercite-teaching-language-models-to-support-answers-with-verified-quotes
با استفاده از مدل RLHP برای جوابهای خودش مستندات و دلایل خودش را برای کاربر پرسش کننده ارائه میکند این مدل اگر نتواند مستنداتی برای پاسخ خود ارایه دهد با جواب پاسخ "نمیدانم" به کاربر ادامه پرسش پاسخ را خاتمه میدهد.
با این حال پژوهشگران پاسخ داده اند که لزوما شواهد ارائه شده توسط مدل آنها ممکن است فاقد اعتبار باشد
#مقاله
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
GopherCite: Teaching language models to support answers with verified quotes
https://www.deepmind.com/publications/gophercite-teaching-language-models-to-support-answers-with-verified-quotes
با استفاده از مدل RLHP برای جوابهای خودش مستندات و دلایل خودش را برای کاربر پرسش کننده ارائه میکند این مدل اگر نتواند مستنداتی برای پاسخ خود ارایه دهد با جواب پاسخ "نمیدانم" به کاربر ادامه پرسش پاسخ را خاتمه میدهد.
با این حال پژوهشگران پاسخ داده اند که لزوما شواهد ارائه شده توسط مدل آنها ممکن است فاقد اعتبار باشد
#مقاله
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
آیا ChatGPT می تواند برچسب های تولید شده توسط انسان را بازتولید کند؟
مطالعه خوبی در مورد کاربرد ChatGPT برای تفسیر وظایف داده ها انجام گردیده است. در این مقاله نتایج نشان میدهد که ChatGPT این پتانسیل را دارد که انواع مختلفی از وظایف تفسیر دادهها مانند تشخیص حالت ، تجزیه و تحلیل احساسات و تشخیص ربات را انجام دهد. این مقاله همچنین تعدادی از محدودیت ها را مورد بحث قرار می دهد.
Can ChatGPT Reproduce Human-Generated Labels? A Study of Social Computing Tasks
https://arxiv.org/abs/2304.10145
#مقاله
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
مطالعه خوبی در مورد کاربرد ChatGPT برای تفسیر وظایف داده ها انجام گردیده است. در این مقاله نتایج نشان میدهد که ChatGPT این پتانسیل را دارد که انواع مختلفی از وظایف تفسیر دادهها مانند تشخیص حالت ، تجزیه و تحلیل احساسات و تشخیص ربات را انجام دهد. این مقاله همچنین تعدادی از محدودیت ها را مورد بحث قرار می دهد.
Can ChatGPT Reproduce Human-Generated Labels? A Study of Social Computing Tasks
https://arxiv.org/abs/2304.10145
#مقاله
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
گوگل اپدیت جدیدی از Bard، که الان میتونه تو برنامهنویسی هم کمک کند و رقابت خودشو با chatgpt نزدیکتر کنه
https://blog.google/technology/ai/code-with-bard/
https://blog.google/technology/ai/code-with-bard/
Google
Code and debug with Bard
Bard can now help with programming and software development tasks, across more than 20 programming languages.
We are inviting applications for a PhD studentship (at University of Sussex) on addressing the problem of early diagnosis, monitoring, and management of Parkinson's disease using sensor and machine learning technologies. For details, see: Cooper Carlson Scholarship at University of Sussex on
https://www.findaphd.com/phds/project/cooper-carlson-scholarship/?p157265
#اپلای
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
https://www.findaphd.com/phds/project/cooper-carlson-scholarship/?p157265
#اپلای
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
PostDoc Position @ TU Berlin, Learning & Intelligent Systems Lab
Application deadline: May 12, 2023
https://argmin.lis.tu-berlin.de/jobs/
The Learning & Intelligent Systems Lab @ TU Berlin is seeking for PostDoc candidates with a strong track-record in AI & robotics research. The state-funded position is independent from pre-defined projects and therefore provides freedom to pursue own research ideas within our scope, as well as to raise own funding for PhD positions or an own sub-group led by the researcher (e.g. via DFG). The position also includes teaching responsibilities, and candidates should be highly motivated to offer great education in AI & Robotics to our students. While there are long-term options (esp. when own funding is raised), the current position is initially limited to 2.5 years.
The Learning & Intelligent Systems Lab
(<https://argmin.lis.tu-berlin.de/>) focuses on research in the field of AI and Robotics, in particular methods that combine AI planning, machine learning and optimization to solve fundamental problems in Robotics and AI. The department is part of the Cluster of Excellence Science of Intelligence and the ELLIS network.
*Application deadline:* May 12, 2023
In your application, please include (next to the default cover letter, CV, and study certificates) also a list of publications, google scholar link, and a list of references (name and Email of senior researchers you've worked with that we can directly contact for a reference letter).
Please submit applications and address any administrative questions to:
<i.cicchetti-nilsson@tu-berlin.de>
Please contact me (<toussaint@tu-berlin.de>) if you have any questions regarding the position. Here is the official announcement:
<https://www.jobs.tu-berlin.de/stellenausschreibungen/164571>
#اپلای
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
Application deadline: May 12, 2023
https://argmin.lis.tu-berlin.de/jobs/
The Learning & Intelligent Systems Lab @ TU Berlin is seeking for PostDoc candidates with a strong track-record in AI & robotics research. The state-funded position is independent from pre-defined projects and therefore provides freedom to pursue own research ideas within our scope, as well as to raise own funding for PhD positions or an own sub-group led by the researcher (e.g. via DFG). The position also includes teaching responsibilities, and candidates should be highly motivated to offer great education in AI & Robotics to our students. While there are long-term options (esp. when own funding is raised), the current position is initially limited to 2.5 years.
The Learning & Intelligent Systems Lab
(<https://argmin.lis.tu-berlin.de/>) focuses on research in the field of AI and Robotics, in particular methods that combine AI planning, machine learning and optimization to solve fundamental problems in Robotics and AI. The department is part of the Cluster of Excellence Science of Intelligence and the ELLIS network.
*Application deadline:* May 12, 2023
In your application, please include (next to the default cover letter, CV, and study certificates) also a list of publications, google scholar link, and a list of references (name and Email of senior researchers you've worked with that we can directly contact for a reference letter).
Please submit applications and address any administrative questions to:
<i.cicchetti-nilsson@tu-berlin.de>
Please contact me (<toussaint@tu-berlin.de>) if you have any questions regarding the position. Here is the official announcement:
<https://www.jobs.tu-berlin.de/stellenausschreibungen/164571>
#اپلای
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
👍1
شاید تا الان براتون سوال شده باشه که چرا در پرسش و پاسخ با chatGPT و Bard پاسخ به سوالات از نظر توضیحات و واقعیت کمی تفاوت باهم دارند؟!
در این مقاله به بررسی چرا ChatGPT در پاسخگویی صادقانه به سوالات کوتاهی می کند؟
چه دلایلی باعث پاسخ دادند کوتاه به سوالات میشود؟!
- دانش ناکافی
- مشکل در فهم ارتباط درونی سوالات
- توانایی های محدود در استدلال
و...
در لین مقاله به این سوالات مدنظر که تاکنون پیش آومده پاسخ داده شده است.
🔸 Why Does ChatGPT Fall Short in Answering Questions Faithfully?
#مقاله
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
در این مقاله به بررسی چرا ChatGPT در پاسخگویی صادقانه به سوالات کوتاهی می کند؟
چه دلایلی باعث پاسخ دادند کوتاه به سوالات میشود؟!
- دانش ناکافی
- مشکل در فهم ارتباط درونی سوالات
- توانایی های محدود در استدلال
و...
در لین مقاله به این سوالات مدنظر که تاکنون پیش آومده پاسخ داده شده است.
🔸 Why Does ChatGPT Fall Short in Answering Questions Faithfully?
#مقاله
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
در سایت بیش از ۳۰۰۰ ابزار هوش مصنوعی به صورت طبقهبندی شده قرار دارد.
با جستجو در این سایت میتوانید ابزار مورد نیاز خودتان را پیدا کنید.
https://topai.tools
#خبر
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
با جستجو در این سایت میتوانید ابزار مورد نیاز خودتان را پیدا کنید.
https://topai.tools
#خبر
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
👍3
ده #ایده_جذاب در هفته گذشته که مقالاتشان منتشر گردید
1) DINOv2: Learning Robust Visual Features without Supervision - a new method for training high-performance computer vision models based on self-supervised learning; enables learning rich and robust visual features without supervision which are useful for both image-level visual tasks and pixel-level tasks; tasks supported include image classification, instance retrieval, video understanding, depth estimation, and much more.
2) Learning to Compress Prompts with Gist Tokens - an approach that trains language models to compress prompts into gist tokens reused for compute efficiency; this approach enables 26x compression of prompts, resulting in up to 40% FLOPs reductions.
3) Scaling the leading accuracy of deep equivariant models to biomolecular simulations of realistic size - presents a framework for large-scale biomolecular simulation; this is achieved through the high accuracy of equivalent deep learning and the ability to scale to large and long simulations; the system is able to “perform nanoseconds-long stable simulations of protein dynamics and scale up to a 44-million atom structure of a complete, all-atom, explicitly solved HIV capsid on the Perl-mutter supercomputer.”
4) Evaluating Verifiability in Generative Search Engines - performs human evaluation to audit popular generative search engines such as Bing Chat, Perplexity AI, and NeevaAI; finds that, on average, only 52% of generated sentences are supported by citations and 75% of citations support their associated sentence.
5. Generative Disco: Text-to-Video Generation for Music Visualization - an AI system based on LLMs and text-to-image models that generates music visualizations.
6) Architectures of Topological Deep Learning: A Survey on Topological Neural Networks
7) Visual Instruction Tuning - presents an approach that uses language-only GPT-4 to generate multimodal language-image instruction-following data; applies instruction tuning with the data and introduces LLaVA, an end-to-end trained large multimodal model for general-purpose visual and language understanding.
8) ChatGPT: Applications, Opportunities, and Threats
9) Chameleon: Plug-and-Play Compositional Reasoning with Large Language Models - a plug-and-play compositional reasoning framework that augments LLMs and can infer the appropriate sequence of tools to compose and execute in order to generate final responses; achieves 87% accuracy on ScienceQA and 99% on TabMWP.
10) Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Models - applies latent diffusion models to high-resolution video generation; validates the model on creative content creation and real driving videos of 512 x 1024 and achieves state-of-the-art performance.
#مقاله
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
1) DINOv2: Learning Robust Visual Features without Supervision - a new method for training high-performance computer vision models based on self-supervised learning; enables learning rich and robust visual features without supervision which are useful for both image-level visual tasks and pixel-level tasks; tasks supported include image classification, instance retrieval, video understanding, depth estimation, and much more.
2) Learning to Compress Prompts with Gist Tokens - an approach that trains language models to compress prompts into gist tokens reused for compute efficiency; this approach enables 26x compression of prompts, resulting in up to 40% FLOPs reductions.
3) Scaling the leading accuracy of deep equivariant models to biomolecular simulations of realistic size - presents a framework for large-scale biomolecular simulation; this is achieved through the high accuracy of equivalent deep learning and the ability to scale to large and long simulations; the system is able to “perform nanoseconds-long stable simulations of protein dynamics and scale up to a 44-million atom structure of a complete, all-atom, explicitly solved HIV capsid on the Perl-mutter supercomputer.”
4) Evaluating Verifiability in Generative Search Engines - performs human evaluation to audit popular generative search engines such as Bing Chat, Perplexity AI, and NeevaAI; finds that, on average, only 52% of generated sentences are supported by citations and 75% of citations support their associated sentence.
5. Generative Disco: Text-to-Video Generation for Music Visualization - an AI system based on LLMs and text-to-image models that generates music visualizations.
6) Architectures of Topological Deep Learning: A Survey on Topological Neural Networks
7) Visual Instruction Tuning - presents an approach that uses language-only GPT-4 to generate multimodal language-image instruction-following data; applies instruction tuning with the data and introduces LLaVA, an end-to-end trained large multimodal model for general-purpose visual and language understanding.
8) ChatGPT: Applications, Opportunities, and Threats
9) Chameleon: Plug-and-Play Compositional Reasoning with Large Language Models - a plug-and-play compositional reasoning framework that augments LLMs and can infer the appropriate sequence of tools to compose and execute in order to generate final responses; achieves 87% accuracy on ScienceQA and 99% on TabMWP.
10) Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Models - applies latent diffusion models to high-resolution video generation; validates the model on creative content creation and real driving videos of 512 x 1024 and achieves state-of-the-art performance.
#مقاله
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
آموزش کلاسی دانشگاه استنفورد منتشر گردید.
🔸 Stanford CS330: Deep Multi-Task and Meta Learning I Autumn 2022
فیلم آموزش کلاسی #منابع #پیشرفته #پردازش_زبان_طبیعی #یادگیری_عمیق
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
🔸 Stanford CS330: Deep Multi-Task and Meta Learning I Autumn 2022
فیلم آموزش کلاسی #منابع #پیشرفته #پردازش_زبان_طبیعی #یادگیری_عمیق
🔸 مطالب بیشتر 👇
✅ @AI_DeepMind
❤1