...
When it comes to the popular subject of LLMs, there is a lot of debating going on currently about what an "Open AI model" really means. While it is out of the scope of this paper to try and settle any of those debates, it is obvious that there is a clear need to create the definition of "open LLM". The sooner such definition is created and blessed by the industry, the faster innovation can happen.
In the area of open source LLMs, both with respect to generative models [11] as well as more specialised, discriminative models such as text classifiers, QA, summarisation, and text embedding models [12], has been particularly vibrant and rapidly evolving over the past 5 years. A number of widely used, global platforms used for sharing open models, code, datasets and accompanying research papers, have been particularly instrumental in democratising access to cutting edge technologies and fostering an environment of global collaboration. Among these platforms, Huggingface [30] has played a particularly pivotal role. At the time of this writing, HuggingFace hosts [31] over 350K models, 75K datasets and 150K demo apps (Spaces), in more than 100 languages. It also maintains Transformers, a popular open source library that facilitates integrating, modifying and performing downstream task adaptation for thousands of foundation models from this vast repository. It also provides the Datasets library, as well as several widely used benchmarks and leaderboards [11][12] that are very instrumental for researchers and developers implementing LLM solutions. Other important platforms used by the AI/ML open source community in general (not necessarily LLM-focused) are Kaggle [32] (used for public datasets and high profile ML competitions in all areas) and Paperswithcode [33] (this platform links academic research papers to their respective code and implementation, as well as providing benchmarks and leaderboards comparing different competing solutions for a wide area of ML tasks).
- Integration blueprints
In recent years The Linux Foundation Networking launched a set of Open Source Networking "Super Blueprints" that outline architectures for common networking use cases and are built using OSS technology. Those blueprints consists of collections of OSS projects and commercial products, integrated together by the open source community and documented for free use by any interested party. Several of these blueprints started incorporating AI technologies and we expect that trend to accelerate, as existing blueprints inspire additional AI-driven solutions. One promising area is intent-based network automation. There is current work on blueprints that use NLP and LLM to translate user intent, expressed in natural spoken language to network requirements, and generate full network configuration based on those requirements. Such approach can significantly simplify existing network operation processes, and enable new services that can be directly consumed by the end users in an automated manner (known as Network-as-a-Service - NaaS)
...
[29] The Pile Dataset: https://pile.eleuther.ai/
[30] Huggingface platform: https://huggingface.co/
[31] Huggingface platform statistics: https://originality.ai/blog/huggingface-statistics
[32] Kaggle platform: https://www.kaggle.com/
[33] Paperswithcode platform: https://paperswithcode.com/