"They deliberately compared llama 2 to llama 1 and other famous open source models but not with gpt4, and in these benchmarks, the trend is fairly clear: it crushes the other open source language models."
"I think there'll be a lot of focus on doing this thinking thing with deep learning with language models."
"The potential that language models have... the ceiling is really much higher."
"Microsoft released its Orca 2, which is a pair of small language models that actually outperform their larger counterparts."
"I think we need another breakthrough. I think we still can push on large language models quite a lot and we should and we will do that."
"The real story of this week is all of the news about new large language models."
"Language models are at the forefront of strong artificial intelligence."
"Prompting: Asking a large language model a question and getting a completion."
"LLMs have recently gained traction due to outperforming human reasoning in certain contexts."
"Now let's look at large language models. These use digital computation and weight sharing, which is why they can learn so much. They're actually getting knowledge from people by using distillation."
"The language models score higher in creativity than people."
"The exciting, really fantastic progress on large language models."
"Large language models that are good at doing what we train them to do are good because they've built the same kind of model of the world that we have in our minds."
"The large language models we have today, I think of as descendants of this tiny language model."
"We've built a specialized hardware and software stack to support the training inference of large language models or LLMs."
"You can do market research without the market using large language models."
"We want to align our state-of-the-art language models with how the human brain works."
"The release of open Elm models aim to empower and enrich the open research community by providing access to state-of-the-art language models trained on publicly available datasets."
"A hacker's guide to language models: a code-first approach to understanding."
"It's going to make more sense if you know the basics of deep learning."
"The basic idea comes from a paper I created back in 2017 called ULMfit."
"You need to fine-tune language models to make them useful."
"So yes, we can have small language models that perform as well as large language models."
"Attention mechanisms are really important in large language models."
"OpenAI's language models are undoubtedly the best on the market."
"The Absurd ridiculous and yet true story of how it happening since 2017 open AI has been building generative pre-trained Transformer models or gpts language AIS with the singular focus on predicting text trained across billions of writing samples."
"Anybody can try that. You can play with any of the large language models and keep feeding your information, asking for better suggestions."
"The totality of language models is a dead end. It's going to be language models plus explicit other types of models. The ensembles of these is really, I think, going to be the future of more useful AI."
"This is the next logical step of large language models and container technology."
"Large language models are essentially extremely compressed versions of the entire internet, books, articles..."
"What I'm seeing at the moment is chat GPT going down like this and Google bard coming up like this in terms of the quality of output."
"Large language models and chain of thought prompting are making very fast progress against the hardest tasks."
"That's the architecture that is behind the current Chat GPT and large language models."
"Large language models provide a path to making programming and computation as easy as asking a computer what you want in natural language."
"Language models want to complete documents, so you can trick them into performing tasks just by arranging fake documents."
"Emergence is a really important concept... today's large language models show tremendous amounts of very powerful emergence."
"Large language models are just spicy auto correct, prove me wrong."
"The language models are writing Middle School level essays and people are like 'wow it's a great essay!'"
"For question of 'what color is the cat?' it could be gray or black, it's the different LSTM flavors producing that answer."
"The benefits of using these projects is that you can run these LLMs on your laptop without needing an internet connection."
"Large language models are a tremendous breakthrough."
"Large language models with knowledge bases are becoming incredibly important for various applications."
"LLMs achieve a certain generality just by having a very, very general training data set."
"Large language models can revolutionize information access."
"Better human language understanding can translate directly into financial returns."
"Large language models brought dramatic progress in natural language understanding tasks."
"Large language models are perfect at coming up with the structures that you need."
"So, yeah, try this out for other models in your language and see if this works."
"Language models will be an important stepping stone in developing different types of systems."
"Graphs are the gold standard to reduce the element of hallucination in language models."
"Large language models in five formulas."
"...we Define a language model as a probalistic model of a document."
"People have also explored some of the attacks that you could perform on large scale language models."
"Chad GPT is able to take a pretty ridiculous question produce an amazing answer and it does this by basically just training a very large language model on a big machine for a very long time."
"Large language models are pre-trained on massive volumes of publicly available data."
"There's going to be a large language model interface for computers."
"Generation AI, they are only going to know talking to LLMs, having it write their papers, having it format data for them."
"Chat GPT and other language models like it are really good at explaining things in regular English."
"So language models are used in a large number of applications."
"Perhaps the recent evolution of language models puts a little bit of pressure on the claim that there are things that we didn't think were learnable without innate priors."
"Larger models are just getting much better and better at predicting the next word given more and more context."
"Language models can display zero-shot learning."
"Language models are few-shot learners."
"If you are really good at predicting the next word in text, you really begin to see language models sort of in a way as rudimentary world models."
"...language models are trained to predict the most likely continuation of tokens and that is not the same as what we want language models to do which is to assist people."
"With this human feedback, we end up producing a language model that actually ends up producing summaries that are judged to be better."
"What's changed in the last 75 years? What's special about the modern language models which have these amazing capabilities?"
"Connect a language model to other sources of data, so so important right because if you don't have connections to other data, it's not going to be able to give you appropriate context."
"We're going to see more of a trend for LMs with larger and larger context windows."
"Language models do this by analyzing the words that have already been used in the sentence and then using that information to calculate the probability of each possible word."
"Shows the power of curiosity and the potential of the large language models."
"It's hard to fully comprehend the number of language model powered features developers are going to be asked to implement in the next few years."
"This is all you need to generate text using a large language model, so that's pretty cool."
"The wide availability and the remarkable results of large language models is a sea change."
"Language models can automatically acquire knowledge from the web."
"You cannot yet solve all of the problems of these large language models."
"How does the A's subtree matching contribute to reducing hallucinations in language models?"
"It's cool and exciting to see some of the things that these language models are doing."
"Autoregressive LLMs can encode and generate natural language."
"Language models are at the point where on almost every document one of their top 30 words is the next word."
"Large language models are incredible snapshots of humanity's knowledge."
"Large language models are truly awesome, they are great tools for understanding language."
"The magic with language models is that you can talk to them."
"You can run a large language model locally on a laptop; it'll be a fun class."
"The only way that we have been able to make any progress in computational linguistics is using statistical methods."
"Large language models... able to auto-complete and generate content."
"A language model is a probability distribution over next words given a preceding context."
"Our product allows you to wield large language models, meaning the results you can get are reliable."
"Code Interpreter is an innovative approach to code interpretation by evolving as well as allowing language models to execute code directly off your computer."
"The entire area of sequence to sequence with pre-trained language models hasn't been touched yet."