2024-09-11
Pixtral 12B is probably going to repeat Mistral v0.1 history [image]
TechCrunch
Mistral releases its first multimodal model, Pixtral 12B, available on GitHub and Hugging Face, and via API-serving platforms Le Chat and Le Platforme “soon”
French AI startup Mistral has released its first model that can process images as well as text.
2024-05-22
phi-3 trained on 4.8T tokens upto cutoff Oct 2023. So can I expect it to do better with libraries and utilities that got popular in late '22 and '23? hope it is the case. [image]
VentureBeat
Microsoft announces the general availability of its Phi-3 models, including Phi-3-Silica, a 3.3B parameter model that will be embedded on all Copilot+ PCs
here's what you can use it for Pradeep Viswav / MSPoweruser : Microsoft and Khan Academy announce AI partnership Kevin Okemwa / Windows Central : Microsoft ships Azure AI Studio in...
2023-09-30
My favourite paper for today. Meta continues pretraining of llama2 with an additional 400B Tokens and closes the gap with GPT 3.5 Good news is that they used synthetic datasets and not human annotations to get quality improvement. True Tokenbenders 🫡
VentureBeat
Meta quietly unveils Llama 2 Long, which has been trained with longer sequences, outperforming GPT-3.5 Turbo and Claude 2 when responding to long user prompts
Meta Platforms showed off a bevy of new AI features for its consumer-facing services Facebook, Instagram and WhatsApp …