Google has released Gemini 1.5 Pro, an updated version of its AI model. Users are praising its performance in various tasks such as reasoning and visual understanding. Gemini 1.5 Pro outperforms its predecessor in long contexts but still falls behind GPT-4 in some aspects. The model's ability to process audio input and its performance in multimodal tasks are highlighted. Community feedback and evaluations indicate improvements in vision ability and reasoning with Gemini 1.5 Pro.
.@Google Gemini Pro 1.5 is your perfect assistant for digesting hour long videos. 🗒️🤖 https://t.co/nBdPEGNWGT
🤖Gemini 1.5 Pro is a foundation model that performs well at a variety of multimodal tasks such as visual understanding, classification, summarization, and creating content from image, audio and video. 📚Checkout this notebook to get started 👇 https://t.co/wjLeF7qJxe
🤖Gemini 1.5 Pro is a foundation model that performs well at a variety of multimodal tasks such as visual understanding, classification, summarization, and creating content from image, audio and video. Use of Gemini will require an API key. 📚Checkout this notebook to get…
🤖Gemini 1.5 Pro is a foundation model that performs well at a variety of multimodal tasks such as visual understanding, classification, summarization, and creating content from image, audio and video. Use of Gemini will require an API key. Checkout this notebook to get started…
A nice and simple guide to getting started with the Gemini 1.5 Pro in @Google AI Studio 🤖 https://t.co/Y94xPWCDA5
The thing I love the most about Gemini 1.5 is that it can natively process audio input. So I can upload tracks to it from my iTunes library, and then we can figure out the lyrics together and discuss the music. So cool. ☺️
🤖 Gemini 1.5 Pro is a foundation model that performs well at a variety of multimodal tasks such as visual understanding, classification, summarization, and creating content from image, audio and video. Use of Gemini will require an API key. 📚Checkout this notebook get started…
🤖Gemini 1.5 Pro is a foundation model that performs well at a variety of multimodal tasks such as visual understanding, classification, summarization, and creating content from image, audio and video. Use of Gemini will require an API key. 📚 Checkout this notebook get…
🤖Gemini 1.5 Pro is a foundation model that performs well at a variety of multimodal tasks such as visual understanding, classification, summarization, and creating content from image, audio and video. Use of Gemini will require an API key. 📚 Checkout this notebook get started…
It's been quite interesting to study how Gemini 1.5 Pro scales its in-context learning (ICL) from few to many shots. I found our experiments that avoid using hand-labeled examples with Reinforced ICL, and Unsupervised ICL (i.e. shots are input examples only), particularly neat. https://t.co/UljoaseN51
I had a long conversation with Gemini 1.5 today. I have to say, so far it seems to me that it’s the best model right now in terms of general contextual understanding and, shall we say, emotional maturity? In comparison, GPT-4 seems overtrained, with a flat affect. And Claude is a…
Many-Shot Learning: Now that context lengths are over 1M in size, how many examples should you use in your prompts? Depending on the task, you can probably use a lot! We also introduce a couple of new ways for doing in-context learning (ICL) when you don't have enough… https://t.co/6wDCNO0RQM
Google presents Many-Shot In-Context Learning - Proposes many-shot ICL, i.e., adding up to thousands of examples in context with Gemini 1.5, which boosts the perf significantly - Using synthetic CoT is very effect in this setting. https://t.co/wHmJKK3ZHw https://t.co/PFGs2VJ84e
We studied In-Context learning with hundreds to thousands of examples. My favorite example: I sent *one million* tokens to Gemini 1.5 Pro for linear classification with 64 dimensional integer-valued vectors and many-shot learning performs similarly to k-Nearest Neighbours. https://t.co/6YaMmZkikt
What can we improve to make it easier to build with @Google AI Studio and the Gemini API? Please shared any feedback 🧵👇🏻
🍳📚 Introducing the Gemini API Cookbook, your guide to kickstart your journey with the Gemini API! Explore new notebooks featuring audio or video input prompting, system instructions, function calling config, and #JSON mode. → https://t.co/kNFKqgRc38 https://t.co/Qy2Hkrno5B
After receiving community feedback, we added @GoogleDeepMind Gemini 1.5 Pro's results. 👇 Gemini 1.5 Pro's vision ability was significantly improved compared to 1.0 Pro and matched GPT-4's performance on our VisualWebBench! 🏆 Its action prediction (e.g., predicting what would… https://t.co/kQnZzztfEh https://t.co/xMf38jskkL
Crazy how much chain-of-thought prompting enhances Gemini 1.5 pro and GPT-4 for reasoning at varying sizes of context. https://t.co/qDcBaJUppN
New results from our recent benchmark, FlenQA: Gemini 1.5 Pro shows a significant decrease in reasoning performance as context length increases, even below 3000 tokens. This continues the trend of all other models we previously evaluated. See the thread below for more details 👇🏻 https://t.co/NBRPZ59Zix
remember Gemini Pro? 1M context window? @Alon_Jacoby and @mosh_levy just finished evaluating it with their FLenQA dataset, measuring simple reasoning tasks across input lengths (took a while because rate limits). How well does it do? well, as expected, not great. https://t.co/04CQIucKTM
New findings: We just evaluated Gemini 1.5 Pro on our recent benchmark that tests the impact of context size on reasoning performance - it is much better than 1.0 in long contexts! Though still falls behind GPT4. Also, CoT prompting now improves accuracy (unlike with 1.0). (1/4) https://t.co/Yxs8CTtdyU
Google just released Gemini 1.5 Pro It's better than GPT-4: 1 million tokens Different types of prompts You can upload big files, and more… Here are the best parts of Gemini 1.5 Pro you need to know: ↓ https://t.co/hqz04nPpaC
We have updated our @Gemini version to Gemini Pro 1.5! https://t.co/9Q5JKUO0uS
Fantastic analysis by Cjz. I've been using Gemini 1.5 Pro for almost everything nowadays. Based on my current experience with these models: - Google Gemini 1.5 - Mixtral (open source) - ChatGPT-4 https://t.co/8eoqcVZdUq
The Gemini API cookbook is shaping up quite nicely (and we just crossed 1k stars ✨) Check it out if you want to quickly test out 1M context and native multi modal support! https://t.co/SVad0K8QFN
We are rolling out support for #Google’s newly released #Gemini 1.5 Pro in v4.8. The update is now live in the Chrome Web Store, so give it a spin! #SiderAI #AI #GeminiAI #ChromeExtensions #EdgeExtensions https://t.co/VrDSTThQyk