Last week, domestic mainstream large model application research companies have released their large model APP, this article will be the current hot several APP functional evaluation, let's take a look at the performance of each one.
Evaluation objects: iFLYTEK Spark (iFLYTEK), Wen Xin Yiyan (Baidu), Bean Bao (bytes)
【Initial question】Please compare and analyze the functions of the large model APP Wen Xin Yiyan, Xunfei Xinghuo and Bean Bao, and make recommendations.
iFLYTEK Spark
After the first round of communication fails, let it try to change roles, or fail [Spread hands]
Wen Xin said
Continued
Continued
Continued
Each product has advantages and disadvantages, and it is considered to be comprehensive and a little different at the same time, but isn't the bean bag developed by byte, and the accuracy of the answer is to be verified [think]
Bean bags
This writing is indeed more understandable and easy to grasp the point, but the facts are to be verified by the next questions!
Comparison summary: Although there is no problem with iFLYTEK's answer, it is slightly rigid in its positioning, so how can it still play cosplay with user-friendly; Wen Xin is all-encompassing (Rory bar), but with a lot of content, the context is prone to self-contradictions, illogical places, and it is a bit confusing for users (to check a bunch of content accuracy [halo]); The doubao answer is more concise and clear, and if the accuracy is not too much of a problem, it is OK.
Consultative Questions help me identify emerging trends and developments in the automotive industry and suggest how to adapt my skills and knowledge to stay ahead of the curve.
iFLYTEK Spark
Mainstream trends: electrification, connectivity, autonomous driving, and new energy infrastructure
Wen Xin said
Continued
In addition to iFLYTEK's electric intelligent networking, it also complements lightweight and circular economy design
Bean bags
The scope is the least, but the focus is there, and the source is listed below, which is more detailed [send heart]
Comparison summary: the answers of the three models are OK, basically classified by points for the structured presentation of text, of which iFLYTEK's summary speech is more in line with the communication habits of natural language, the most comprehensive enumeration of Wen Xin's words, the content is sufficient but sometimes needs to be summarized again, the doubao answer is the most focused, and there is an article source, which is convenient for illustration and citation, but the recommended content is somewhat repetitive and not concise enough.
[Arithmetic problem]: Xiao Ming's rowing speed in still water is 10 kilometers per hour, and now he travels to and from a certain river, it takes 9 hours against the water and 6 hours in the water, finding the speed of the river. (Junior High School Mathematics)
iFLYTEK Spark
It looks presentable, but it's a pity that the equation is solved wrong
Wen Xin said
Answer correctly!
Bean bags
The equations are not listed correctly, and the meaning of the title has not been accurately recognized
Comparison summary: Wen Xin is the best, followed by iFLYTEK, and bean bun is second
【Logical reasoning type】Calculate how many electric vehicles enter Hongqiao Railway Station every day?
iFLYTEK Spark
[smile]
Wen Xin said
[Innocent laughs]
Bean bags
[frown]
Comparative summary: At present, everyone does not have complex logical reasoning ability, mainly based on the existing public information, splicing and reorganizing or refining and summarizing. It's just that iFLYTEK's suggestion is more like understanding what I want to do.
The above is a comparative summary of the three large-model apps in three aspects: consultative question and answer, basic computing and logical reasoning. Callback's answer to the initial question: "Wen Xin's answers are usually more comprehensive, and can provide a variety of possibilities and explanations. ", "iFLYTEK Xinghuo's answer is usually more concise and clear, suitable for quick access to information", "Bean Bao's answer is usually more interesting and can have good interaction with users", this round of testing feels that Wen Xin is indeed more comprehensive but not necessarily accurate, can be used as inspiration and reference; iFLYTEK's reply is more concise, more natural, and more conversational; The bean bun reply is the most concise and simple, but it doesn't feel interesting. However, this test has limited questions, and the results may be biased, please adopt it as needed~
Next, we will continue to evaluate and compare its functions such as text processing, copywriting and graphic painting, and interested partners are welcome to continue to pay attention to [Flying Kiss]