Did you know if you carefully combine all the training best practices, you can get a SoTA retrieval model with less data and compute? @bclavie
, does, so he did, so here's the new SoTA for Japanese retrieval:
LLMS
-
SoTA Japanese retrieval model training best practices efficiency
By
–
-
Sliding Window Attention: Optimizing Transformer Models
By
–
Sliding Window Attention https://
bit.ly/4bHSh1k
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
Testing a new Sonar tool based on Llama 3.1 70B
By
–
I am testing a new Sonar based on Llama 3.1 70B
-
Availability of Advanced ChatGPT Voice Mode
By
–
He will probably get Advanced ChatGPT voice mode before you 👀👀👀 https://t.co/x0FNjrhfBD
— 🚨 AI News | TestingCatalog (@testingcatalog) 2 août 2024He will probably get Advanced ChatGPT voice mode before you
-
HaloQuest: Benchmark for Multimodal AI Hallucination Reduction
By
–
Introducing #HaloQuest, our latest effort towards improving hallucination in multimodal foundation models! We hope that HaloQuest, as both a challenging benchmark and an open-sourced dataset, will enable the field's progress in advanced reasoning! Also glad that HaloQuest played
-
Build Translation Agent with Yi LLMs Guide
By
–
Build your own translation agent with Yi LLMs! This guide from @Gaianet_AI covers everything from setup to advanced translation techniques. Check it out now:
-
Q2 2024 AI Recap: Claude, Llama, Apple Intelligence
By
–
pod: The Winds of AI Winter! https://
latent.space/p/q2-2024-recap The vibes have shifted… @fanahova and @swyx celebrate 1m downloads and recap the last 3 months in AI! Discussing the Frontier Labs vibe shift between Claude 3.5, Llama 3.1, Apple Intelligence, and the expansion of -
How GPT-4o Adapts Code Explanation Between Text and Speech
By
–
🔴 ¿Puede CHATGPT leer código en Python?
— Carlos Santana (@DotCSV) 2 août 2024
Uno de los descubrimientos más interesantes de ayer fue el siguiente: GPT-4o no habla y escribe exactamente lo mismo, sino que intenta acomodarse a si habla o escribe.
En texto nos escribe el código, pero hablando intenta explicarlo 👇 pic.twitter.com/3KZSgSAzwu¿Puede CHATGPT leer código en Python? Uno de los descubrimientos más interesantes de ayer fue el siguiente: GPT-4o no habla y escribe exactamente lo mismo, sino que intenta acomodarse a si habla o escribe. En texto nos escribe el código, pero hablando intenta explicarlo
-
Microsoft Introduces SpreadsheetLLMs for Data Analysis
By
–
New from Microsoft: SpreadsheetLLMs! Author @HaoyuDong9 will be answering your questions over the next few days. Leave questions directly on the preprint!
-
Chat with the Downloaded Model Step 3
By
–
Step 3: Once it is downloaded, you can finally chat with the model: