IBM integrates Meta’s language model, Llama2, into Watsonx. Explore how IBM’s decision is surprising yet expected.
IBM to Integrate Meta’s Llama2 into Watsonx
IBM and Meta have joined forces to incorporate Meta’s Llama 2 model into IBM’s AI platform, Watsonx. This collaboration underscores IBM’s approach of offering its proprietary AI models alongside external models, reflecting the trend of embracing open source solutions.
IBM’s Move Reflects Industry Trends
Watsonx.ai users, particularly select IBM partners, will soon gain access to the expansive capabilities of Meta’s Llama 2, a substantial leap for the AI landscape. Watsonx.ai currently empowers users with both IBM’s models and those from the Hugging Face community.
This cooperative strategy resonates with the approach adopted by AWS, which champions its in-house suite of large language models while also collaborating with vendors such as Cohere, Hugging Face, and Meta.
The partnership between Meta and IBM is monumental: It signifies IBM’s foray into integrating third-party models into its platform, setting the stage for future developments in this direction.
The collaborative move between IBM and Meta underscores the evolving landscape of generative AI. Enterprises are gradually realising the broader spectrum of AI vendors available beyond OpenAI, and are starting to gravitate towards vendors whose large
IBM’s Adaptability Ideal for Making Custom Solutions
IBM’s architecture within Watsonx allows users to employ either their preferred large language model or IBM’s pre-existing options. This adaptability is expected to resonate with enterprises seeking tailored solutions.
IBM’s affinity for open source was also highlighted. The 2019 acquisition of Red Hat exemplified IBM’s appreciation for open source, making the partnership with Meta, a leader in open source language models, a natural progression.
The recognition of open source’s potential is shared by prominent players in the industry. The vibrant open source ecosystem is increasingly appealing to big vendors. This indicates a broader shift towards open source models in the enterprise realm.
Challenges Persist
While open source models are gaining traction, a gap remains in performance compared to closed source models. Closed source models still outperform their open source counterparts.
Meta’s proactive engagement with various vendors, including Microsoft and AWS, is noteworthy, yet Chandrasekaran highlighted that Meta’s reputation as an enterprise vendor is still developing.
The imminent inclusion of Llama 2 within Watsonx.ai is just the beginning. This release will be succeeded by the introduction of AI Tuning Studio, additional models on the watsonx.ai platform, and Factsheets within watsonx.governance.
