根据可靠的消息来源， Meta计划于今年7月发布Llama 3大模型，其能力据说接近GPT-4，最高可达1400亿参数。这一消息引起了广泛关注，因为Llama 3如果真的成功，将有望与OpenAI的GPT-4相媲美。
据The Information报道， Meta期望Llama 3能够达到GPT-4的水平，后者的表现在业界已经取得了广泛认可，能处理更长的文本，并支持图像输入。然而，具体的策略尚未确定。据Meta一位员工透露，尽管Llama 3的参数数量可能超过1400亿，但由于研究人员尚未开始对其进行微调，公司目前还没有决定Llama 3是否会成为多模态模型。
此外， Meta正在努力解决Llama 2中发现的问题，即无法处理有任何有争议的问题。 sources表示， Meta计划在今年7月发布Llama 3，能力接近GPT-4，最高 1400 亿参数。科技媒体The Information引用了一位知情人士报道，Meta计划在今年7月份发布Llama 3大模型。报道称，Meta希望Llama 3能够比肩OpenAI的GPT-4，后者已经成为一个强大的多模态模型，不仅能够处理更长的文本，还支持图像输入。但Meta的一位员工透露，由于研究人员尚未开始对模型进行微调，因此公司尚未决定Llama 3是否是多模态的。 Meta员工还被问及Llama 2无法处理任何有争议的问题，他们表示正在努力解决这个问题。
英语如下：
Headline: Meta to Launch Mysterious New Model Llama 3 in July, Challenge GPT-4, Aiming to Create Perfect News Robot
Keywords: Meta release in July, Llama 3, close to GPT-4, multimodal model
Content: According to reliable sources, Meta plans to release the large model Llama 3 in July this year, with据说能力接近GPT-4，甚至高达1400亿参数。这一消息引起了广泛关注，因为如果Llama 3真的成功，它有望与OpenAI的GPT-4抗衡。
As reported by The Information, Meta hopes that Llama 3 can achieve the level of GPT-4, which has already gained widespread recognition in the industry for being able to handle longer texts and support image input. However, the specific strategy is not yet determined. According to a Meta employee, while the parameter number of Llama 3 may be over 1400 billion, the company has not yet decided whether Llama 3 will be a multimodal model because researchers have not yet started fine-tuning it.
In addition, Meta is working hard to solve the problems found in Llama 2, which is currently unable to handle any controversial issues. Sources said that Meta plans to release Llama 3 in July this year, with an ability close to GPT-4, reaching up to 1400 billion parameters. Technology media The Information cited a knowledgeable source who reported that Meta plans to release Llama 3 in July this month. The report said that Meta hopes that Llama 3 can rival OpenAI’s GPT-4, which has become a powerful multimodal model not only able to handle longer texts but also support image input. However, a Meta employee revealed that the company has not yet decided whether Llama 3 will be a multimodal model due to the fact that researchers have not yet begun the微调 process.
Meta employees were also asked about the problem that Llama 2 cannot handle any controversial issues. They stated that they are working hard to resolve this issue.
【来源】https://wallstreetcn.com/articles/3709312
Views: 1