AI has deep roots in Europe, and we are committed to supporting developer and researcher communities across Europe. As the EU enters the home stretch of #AIAct deliberations, we urge the EU to protect and promote grassroots innovation in open models
OPEN SOURCE
-
Code Llama: Open Foundation Models for Code Generation
By
–
Code Llama: Open Foundation Models for Code https://
bit.ly/3RCyvhr A version of Llama 2, fine-tuned for generating and discussing code. -
Open access to observability data: debunking myths about transparency
By
–
Another myth, busted. Swipe to see why providing open access to observability #data for everyone is crucial
-
spaCy Introduces Premium Subscription Service for Advanced Language Models
By
–
Here's the plan. We want to introduce a subscription service with premium up-to-date models with an easy PyPi server. You'd write: $ pip install spacy-models[pro_en_ner_web_trf,pro_de_core_chat_sm] -f https://$KEY@models.spacy.io …and get models trained on better data!
-
spaCy seeks sustainable funding for development and data
By
–
We need a sustainable way to fund more spaCy development, especially data. If this sounds like something your company could use, sign up & help make sure we're building what you really need. You'll get early beta access & a 20% discount for the 1st year once it goes on sale.
-
spaCy Users Survey: Improving Trained Pipelines and Deployment
By
–
spaCy users! Are you using trained @spacy_io pipelines (en_core_web_sm etc.) at work? We've heard they're doing good work but you want: better accuracy, esp. on new text PyPi server to put them in requirements more vectors & component combos
-
Thanks to compute providers and partners for model release support
By
–
Huge thanks to our compute providers @CoreWeave and the EuroHPC, @tri_dao and @d_haziza for their help with FlashAttention and xFormers, and @huggingface
, vLLM, @skypilot_org
, FastChat for their help and support with this release. -
GQA and Sliding Window Optimization for Efficient Decoding
By
–
We trained it with GQA and a sliding window of 4096 tokens, resulting in constant cache size and a linear decoding speed. Our changes to FlashAttention v2 and xFormers to support sliding window are available to the community.
-
Mistral 7B Released: Outperforms Llama 2 13B on All Benchmarks
By
–
Mistral 7B is out. It outperforms Llama 2 13B on every benchmark we tried. It is also superior to LLaMA 1 34B in code, math, and reasoning, and is released under the Apache 2.0 licence. https://
mistral.ai/news/announcin
g-mistral-7b/
… -
Mistral AI Releases First Model, Best Open Source 7B
By
–
At @MistralAI we're releasing our very first model, the best 7B in town (outperforming Llama 13B on all metrics, and good at code), Apache 2.0. We believe in open models and we'll push them to the frontier https://
mistral.ai/news/about-mis
tral-ai/
… Very proud of the team !