Text embedding models play an important role in RAG systems and enterprise search applications. Join @Nils_Reimers and @SerranoAcademy at 11am ET today to learn more about our recently released state-of-the-art text embedding model, Embed v3. https://
info.cohere.ai/embed-v3-webin
ar
…
LLMS
-
Cohere Launches Embed v3 State-of-the-Art Text Embedding Model
By
–
-
OpenAI Special Edition Recording Replaces AI Top Questions Post
By
–
Well, the plan was a special top questions about AI post. That changed. I’m recording an OpenAI edition today.
-
BERT Contextualized Word Embeddings: NLP Revolution Explained
By
–
🎥New Video Alert🎥https://t.co/Mccoi7ndvF
— Satya Mallick (@LearnOpenCV) 20 novembre 2023
Exploring the NLP Revolution with BERT 🚀🤖! Our new video unravels the magic of BERT's contextualized word embeddings, a milestone since 2018 in sentiment analysis & more. Discover why BERT still matters in the world of AI & NLP.… pic.twitter.com/T6InAfbdaSNew Alert https://
youtube.com/watch?v=V9eU8z
1cXqY
… Exploring the NLP Revolution with BERT ! Our new video unravels the magic of BERT's contextualized word embeddings, a milestone since 2018 in sentiment analysis & more. Discover why BERT still matters in the world of AI & NLP. -
Inflection-2 Training Complete: Second Best LLM in World
By
–
Utterly insane weekend. So sad. Wishing everyone involved the very best. In the meantime, we finished training Inflection-2 last night! ✨ It's now the 2nd best LLM in the world… & we're scaling MUCH further. Details v soon. Come run with us!
→ View original post on X — @inflectionai, 2023-11-20 13:32 UTC
-
Microsoft acquires OpenAI talent and secures AGI technology access
By
–
Satya wins it seems. > OpenAI board gets torched for their failure to play ball. > Top talent all follows Sam and Greg to Microsoft, which already contractually has access to all OpenAI’s “pre-AGI technology”. > Now Microsoft will develop and own all of the next models
-
StreamingLLM: Speed Optimization Without Long-term Memory
By
–
Unfortunately, StreamingLLM doesn't solve long-term memory or continual learning. It's just a (useful) technique for improving LLM inference speed. On their github they state: As emphasized earlier, we neither expand the LLMs' context window nor enhance their long-term memory.
-
Fine-tuning LoRA with Adafactor and activation checkpointing
By
–
Even fine tuning regular lora with adafactor would do for now though. With activation checkpointing that’ll handle a reasonable size model
-
QLoRA Fine-Tuning for Single Box Decoder Applications
By
–
I think you really need a single box decoder only fine tuning task, since that’s a really common application. E.g qlora fine tune mistral on flan v2
-
GPT-4 Token Limits: Older vs Newer API Models
By
–
Max output for older GPT-4 models is the same for input and output tokens. Only the newer API models are limited to 4096 tokens.
-
Ilya Sutskever presents GPT-2, foundation of ChatGPT
By
–
Ilya Sutskever @ilyasut presenting GPT-2 back in March 2019, the core of ChatGPT. pic.twitter.com/9I6IUbq3JA
— Reza Zadeh (@Reza_Zadeh) 19 novembre 2023Ilya Sutskever @ilyasut presenting GPT-2 back in March 2019, the core of ChatGPT.