Chinese tech giants integrate Meta’s Llama 3.1 AI model into cloud services

Chinese technology giants Alibaba Group Holding and Tencent Holdings have swiftly integrated Meta Platforms’ latest AI model, Llama 3.1, into their cloud services. The launch of Llama 3.1 has garnered significant attention, prompting rapid adoption by these leading firms.

Also Read: Elon Musk announces GROK 3 training at Memphis with NVIDIA H100 GPUs

Alibaba Cloud, the e-commerce company’s cloud computing arm, was one of the first to implement the new Llama model. The integration into its Bailian model training platform was revealed through Alibaba’s WeChat official account. The company is giving away one month of computing resources for training and inference in Llama 3.1, underlining the model’s value. 

Tencent follows suit and enhances model capabilities

Llama 3. 1, the newest member of the open-source Llama family of large language models (LLMs), powers generative AI applications like ChatGPT. Following Alibaba’s announcement, Shenzhen-based Tencent Holdings quickly followed suit. Tencent’s approach also involves the improvement of the fine-tuning and inference of the model for flexibility in various tasks such as conversation, text generation, and writing. 

Meta Platforms created the Llama models and divided them into three sizes: the 8B, 70B, and 450B Llamas, with the numbers referring to the number of parameters. The number of parameters is one of the essential factors that determine the ability of the LLM, as more parameters mean higher capacity.

Zuckerberg positions Llama 3.1 as an open-source frontier

Meta’s CEO, Mark Zuckerberg, has stressed that “decentralized and open innovation” is beneficial in the current climate of espionage and model theft. He said that “decentralized and open innovation” is beneficial to the U. S. and revealed that the future Llama models would be the best in the market starting the next year.

Zuckerberg stated that future Llama models will narrow this gap between open-source and the best closed-source models that are currently available in the industry. He further stated that Llama models will be the most sophisticated in the market by the end of next year.  

Llama 3. 1 has received a lot of attention among Chinese startups as these use Llama models as a base for creating generative AI applications. Nevertheless, there are some skeptics in the industry who advise against the overuse of Meta’s models. The case of the Chinese AI unicorn is one example of such a technology. 

At present, LMSYS, an AI model research body affiliated with the University of California, Berkeley, has placed closed-source models from OpenAI, Anthropic, and Google at the forefront. GPT-4 by OpenAI is at the first position, while models from these companies occupy the first ten positions.