Newsletter
2024: AI's Thriving Frontier - Happy New Year from Prompter
2024: AI's Thriving Frontier - Happy New Year from Prompter
Jan 7, 2024
Happy New Year! As we embark on 2024, we're excited to share the latest and most inspiring developments in AI. This year, we continue our commitment to building Prompter, enhancing your experience with every update.
——————————————————————
📰 Top AI News
OpenAI will launch the GPT Store next week after delays. You can follow GPT builder guide to build your custom GPT.
API key usage metrics are now available in OpenAI dashboard
OpenAI has introduced a feature for tracking API key usage metrics on its dashboard. To activate this feature, users need to visit the API key page and generate a tracking token. Once this token is created, new usage will be automatically tracked. This functionality is particularly useful for organizations with multiple users or for tracking usage across different products or phases, such as development and production environments.
Mobile ALOHA: Open-sourced robot to autonomously complete complex mobile manipulation tasks
If the video showcasing Stanford University's robot Mobile ALOHA is authentic and not manually curated, then this is truly remarkable. Mobile ALOHA appears capable of autonomously performing a variety of complex tasks, such as cooking and serving shrimp, operating and taking an elevator, doing laundry, self-charging, and even interacting playfully with pets, among other activities. This level of versatility in a robot is impressive.
Perplexity Raises $73.6 Million Series B Round
AI search engine startup, Perplexity AI has reportedly raised $73.6 million in a series B funding round led by Institutional Venture Partners LLC with additional investments from Databricks Ventures, former Twitter VP Elad Gil, former GitHub CEO Nat Friedman, Shopify CEO Tobi Lutke and participation from Nvidia and Jeff Bezos. The startup is now valued at $520 million.
——————————————————————
🔥 Trending on Social Networks
How has ChatGPT changed your life?
Greg Brockman, co-founder of OpenAI, recently posed a question on X, and here's an impactful response from Aaron Stormerr, a blind CS student. ChatGPT revolutionized their learning experience, especially in overcoming challenges presented by visual-centric teaching methods. It's a testament to how tech can profoundly enhance lives. Reminds me of my own experience: once, my ChatGPT Plus subscription unexpectedly didn't renew during a late-night coding session. Suddenly without GPT-4, I felt completely lost. How did we actually program before AI came along?
——————————————————————
📖 Worth Reading
📌 LLMs and Programming in the first days of 2024
Salvatore Sanfilippo(aka antirez), co-founder of Redis, reflects in his blog post on the state of programming today. He observes that most programming involves repeating similar tasks, requiring less high-level reasoning, a job at which Large Language Models (LLMs) excel, despite their context size limitations. This reality prompts a significant question for programmers: Is it worth engaging in this kind of programming if LLMs can handle a part of it? While it's a well-paying field, he suggests reconsidering its future relevance.
📌 Microsoft announces Improving Text Embeddings with Large Language Models
They've introduced a new method for high-quality text embeddings using synthetic data and less than 1k training steps. It outperforms traditional methods, showing strong results on benchmarks without labeled data.
DocLLM is a new approach to document understanding, merging textual and spatial elements in enterprise documents like forms, invoices, and contracts. This model, an extension of large language models (LLMs), focuses on bounding box information instead of image encoders, providing insights into complex layouts and heterogeneous content. It employs a novel pre-training objective for text infill and is fine-tuned with a comprehensive instruction dataset. DocLLM has shown superior performance over state-of-the-art LLMs in most tasks and demonstrates strong generalization capabilities on new datasets.
📌 32 techniques to mitigate hallucination in LLMs
This paper summarizes 32 techniques to mitigate hallucination in LLMs. Introduces a taxonomy categorizing methods like RAG, Knowledge Retrieval, CoVe, and more. Provides tips on how to apply these methods and highlights the challenges and limitations inherent in them.
——————————————————————
Before we wrap up, let's remember that the advancements we've discussed today aren't just technological feats; they're stepping stones towards a future where AI seamlessly integrates into every aspect of our lives, reshaping how we interact with the world around us.
Happy New Year! As we embark on 2024, we're excited to share the latest and most inspiring developments in AI. This year, we continue our commitment to building Prompter, enhancing your experience with every update.
——————————————————————
📰 Top AI News
OpenAI will launch the GPT Store next week after delays. You can follow GPT builder guide to build your custom GPT.
API key usage metrics are now available in OpenAI dashboard
OpenAI has introduced a feature for tracking API key usage metrics on its dashboard. To activate this feature, users need to visit the API key page and generate a tracking token. Once this token is created, new usage will be automatically tracked. This functionality is particularly useful for organizations with multiple users or for tracking usage across different products or phases, such as development and production environments.
Mobile ALOHA: Open-sourced robot to autonomously complete complex mobile manipulation tasks
If the video showcasing Stanford University's robot Mobile ALOHA is authentic and not manually curated, then this is truly remarkable. Mobile ALOHA appears capable of autonomously performing a variety of complex tasks, such as cooking and serving shrimp, operating and taking an elevator, doing laundry, self-charging, and even interacting playfully with pets, among other activities. This level of versatility in a robot is impressive.
Perplexity Raises $73.6 Million Series B Round
AI search engine startup, Perplexity AI has reportedly raised $73.6 million in a series B funding round led by Institutional Venture Partners LLC with additional investments from Databricks Ventures, former Twitter VP Elad Gil, former GitHub CEO Nat Friedman, Shopify CEO Tobi Lutke and participation from Nvidia and Jeff Bezos. The startup is now valued at $520 million.
——————————————————————
🔥 Trending on Social Networks
How has ChatGPT changed your life?
Greg Brockman, co-founder of OpenAI, recently posed a question on X, and here's an impactful response from Aaron Stormerr, a blind CS student. ChatGPT revolutionized their learning experience, especially in overcoming challenges presented by visual-centric teaching methods. It's a testament to how tech can profoundly enhance lives. Reminds me of my own experience: once, my ChatGPT Plus subscription unexpectedly didn't renew during a late-night coding session. Suddenly without GPT-4, I felt completely lost. How did we actually program before AI came along?
——————————————————————
📖 Worth Reading
📌 LLMs and Programming in the first days of 2024
Salvatore Sanfilippo(aka antirez), co-founder of Redis, reflects in his blog post on the state of programming today. He observes that most programming involves repeating similar tasks, requiring less high-level reasoning, a job at which Large Language Models (LLMs) excel, despite their context size limitations. This reality prompts a significant question for programmers: Is it worth engaging in this kind of programming if LLMs can handle a part of it? While it's a well-paying field, he suggests reconsidering its future relevance.
📌 Microsoft announces Improving Text Embeddings with Large Language Models
They've introduced a new method for high-quality text embeddings using synthetic data and less than 1k training steps. It outperforms traditional methods, showing strong results on benchmarks without labeled data.
DocLLM is a new approach to document understanding, merging textual and spatial elements in enterprise documents like forms, invoices, and contracts. This model, an extension of large language models (LLMs), focuses on bounding box information instead of image encoders, providing insights into complex layouts and heterogeneous content. It employs a novel pre-training objective for text infill and is fine-tuned with a comprehensive instruction dataset. DocLLM has shown superior performance over state-of-the-art LLMs in most tasks and demonstrates strong generalization capabilities on new datasets.
📌 32 techniques to mitigate hallucination in LLMs
This paper summarizes 32 techniques to mitigate hallucination in LLMs. Introduces a taxonomy categorizing methods like RAG, Knowledge Retrieval, CoVe, and more. Provides tips on how to apply these methods and highlights the challenges and limitations inherent in them.
——————————————————————
Before we wrap up, let's remember that the advancements we've discussed today aren't just technological feats; they're stepping stones towards a future where AI seamlessly integrates into every aspect of our lives, reshaping how we interact with the world around us.
Happy New Year! As we embark on 2024, we're excited to share the latest and most inspiring developments in AI. This year, we continue our commitment to building Prompter, enhancing your experience with every update.
——————————————————————
📰 Top AI News
OpenAI will launch the GPT Store next week after delays. You can follow GPT builder guide to build your custom GPT.
API key usage metrics are now available in OpenAI dashboard
OpenAI has introduced a feature for tracking API key usage metrics on its dashboard. To activate this feature, users need to visit the API key page and generate a tracking token. Once this token is created, new usage will be automatically tracked. This functionality is particularly useful for organizations with multiple users or for tracking usage across different products or phases, such as development and production environments.
Mobile ALOHA: Open-sourced robot to autonomously complete complex mobile manipulation tasks
If the video showcasing Stanford University's robot Mobile ALOHA is authentic and not manually curated, then this is truly remarkable. Mobile ALOHA appears capable of autonomously performing a variety of complex tasks, such as cooking and serving shrimp, operating and taking an elevator, doing laundry, self-charging, and even interacting playfully with pets, among other activities. This level of versatility in a robot is impressive.
Perplexity Raises $73.6 Million Series B Round
AI search engine startup, Perplexity AI has reportedly raised $73.6 million in a series B funding round led by Institutional Venture Partners LLC with additional investments from Databricks Ventures, former Twitter VP Elad Gil, former GitHub CEO Nat Friedman, Shopify CEO Tobi Lutke and participation from Nvidia and Jeff Bezos. The startup is now valued at $520 million.
——————————————————————
🔥 Trending on Social Networks
How has ChatGPT changed your life?
Greg Brockman, co-founder of OpenAI, recently posed a question on X, and here's an impactful response from Aaron Stormerr, a blind CS student. ChatGPT revolutionized their learning experience, especially in overcoming challenges presented by visual-centric teaching methods. It's a testament to how tech can profoundly enhance lives. Reminds me of my own experience: once, my ChatGPT Plus subscription unexpectedly didn't renew during a late-night coding session. Suddenly without GPT-4, I felt completely lost. How did we actually program before AI came along?
——————————————————————
📖 Worth Reading
📌 LLMs and Programming in the first days of 2024
Salvatore Sanfilippo(aka antirez), co-founder of Redis, reflects in his blog post on the state of programming today. He observes that most programming involves repeating similar tasks, requiring less high-level reasoning, a job at which Large Language Models (LLMs) excel, despite their context size limitations. This reality prompts a significant question for programmers: Is it worth engaging in this kind of programming if LLMs can handle a part of it? While it's a well-paying field, he suggests reconsidering its future relevance.
📌 Microsoft announces Improving Text Embeddings with Large Language Models
They've introduced a new method for high-quality text embeddings using synthetic data and less than 1k training steps. It outperforms traditional methods, showing strong results on benchmarks without labeled data.
DocLLM is a new approach to document understanding, merging textual and spatial elements in enterprise documents like forms, invoices, and contracts. This model, an extension of large language models (LLMs), focuses on bounding box information instead of image encoders, providing insights into complex layouts and heterogeneous content. It employs a novel pre-training objective for text infill and is fine-tuned with a comprehensive instruction dataset. DocLLM has shown superior performance over state-of-the-art LLMs in most tasks and demonstrates strong generalization capabilities on new datasets.
📌 32 techniques to mitigate hallucination in LLMs
This paper summarizes 32 techniques to mitigate hallucination in LLMs. Introduces a taxonomy categorizing methods like RAG, Knowledge Retrieval, CoVe, and more. Provides tips on how to apply these methods and highlights the challenges and limitations inherent in them.
——————————————————————
Before we wrap up, let's remember that the advancements we've discussed today aren't just technological feats; they're stepping stones towards a future where AI seamlessly integrates into every aspect of our lives, reshaping how we interact with the world around us.