Facebook’s parent company, Meta, has unveiled its new AI model called Llama 2 as a direct competitor to OpenAI’s popular ChatGPT. Llama 2, developed in partnership with Microsoft, aims to provide an alternative to ChatGPT with its freely available and open-source nature.
Originally released in February 2023, Llama was initially intended as a language model for research purposes and not as a general-purpose chatbot. However, shortly after its introduction, a version of the model was leaked online. Moreover, it faced legal challenges from authors who claimed that both Llama and ChatGPT utilized their copyrighted books without permission for training data.
Meta has now introduced Llama 2, which is available for free and open-source use, unlike OpenAI’s GPT-3.5 and GPT-4 models. Llama 2 is designed to be trained on custom datasets, such as research databases and software documentation. To further streamline its deployment, Meta is collaborating with Microsoft.
According to Meta, Llama 2 has been pretrained on 2 trillion tokens, offering double the context length compared to its predecessor. Additionally, the fine-tuned models have been built upon over 1 million human annotations. The pretrained models were trained on publicly available online data sources, while the fine-tuned model, Llama-2-chat, leverages publicly available instruction datasets and human annotations.
Although there is no official Llama 2 chatbot available at present, it is highly likely that we will see AI applications and tools incorporating Llama in the near future.
In summary, Meta, along with Microsoft, has introduced Llama 2 as a freely available and open-source alternative to ChatGPT. Llama 2 offers improved features, such as longer context length and fine-tuned models, making it a promising addition to the AI landscape. With further developments and applications on the horizon, Llama 2 has the potential to make its mark and cater to a wide range of research and commercial needs.
