Supports Chinese dialogue! New NVIDIA ChatRTX updated

As early as February, NVIDIA launched the LLM-based chatbot ChatwithRTX. In May, the chatbot was updated, adding new models and new functions, the packaging package was also reduced from 35G to 11G, and the software was officially renamed ChatRTX.

In the previous article and video about Chatwith RTX, we mentioned that ChatwithRTX does not have its own Chinese reply. If you want to implement Chinese answers, you need to install your own environment, large language models, etc. But this step has a relatively high threshold for users, and they have to go through many complicated steps to achieve Chinese question and answer.

Before the introduction, let’s briefly talk about what ChatRTX is. ChatRTX uses RAG technology and is accelerated by NVIDIATensorRT-LLM and NVIDIARTX to bring chatbot functions to RTXWindowsPC and workstations. . Therefore, the prerequisite for using ChatRTX is that you must have an RTX30 or RTX40 series graphics card with 8GB of video memory or above.

The main feature of ChatRTX is that it runs locally. Unlike various AI chatbots uploaded to the cloud, local computing is more secure in data storage and benefits from With the support of NVIDIATensorRT-LLM, the processing is also faster, avoiding the situation where the input question results are loaded for a long time and it takes a long time to hold back the answer.

In addition to the original Gemma model, this ChatRTX update also expands a ChatGLM3 model that supports Chinese and English bilinguals, so it is more convenient to use. There is no need to install the environment like the previous version. This update also makes it easier for more users to get started. The Chinese dialogue is also fluent. After importing the document, you can quickly retrieve the content information you need in the folder and present it to user.

In addition to the new large language model, this time ChatRTX also added the function of image prompt word retrieval. After importing a folder containing images, ChatRTX can use OpenAICLIP Extract the keywords of the picture, so that if you enter the keyword of the picture, such as mountain climbing, then ChatRTX will provide you with pictures about mountain climbing in the folder, which can be said to greatly facilitate the search for local pictures, even if you forget it. Can be found through keyword search.

In addition, ChatRTX has also added a speech recognition function, which can recognize speech within 30 seconds and input it into the dialog box, including recognition of Chinese and English, freeing your hands whether at work , or subsequent expanded functions, speech recognition are all beneficial to users, but the problem of the previous version still exists, that is, ChatRTX cannot connect with the context, that is, every time a question is asked, it will clear the memory, and every time it asks a question, it will only Could be a "one-off" issue.

However, I believe that ChatRTX will continue to be updated in the future and become a localized chat and file retrieval robot with fast feedback. However, the premise of all this is that you must have a An NVIDIARTX 30 series graphics card. As the saying goes, buy new rather than old. GALAXY RTX4070 SUPER Metal Master OC, the newly released RTX40 SUPER series graphics card, is your best choice.

GALAXY RTX4070 SUPER Metal Master OC adopts NVIDIAAda Lovelace architecture, has powerful performance, and 12GB of video memory capacity, which can easily meet the hardware needs of LLM, whether it is games or In terms of AI, both GALAXY RTX4070 SUPER and Metal Master OC are capable, so friends who like it are welcome to go to the GALAX official store to select and purchase~

The above is the detailed content of Supports Chinese dialogue! New NVIDIA ChatRTX updated. For more information, please follow other related articles on the PHP Chinese website!