HKUST Xunfei Spark Model
Apply for iFlytek Xinghuo Cognitive Model Account
Registration URL: HKUST Xunfei Spark Cognitive Model
Fill in the information truthfully, and you will be prompted to fill in your real name, company information and position later, just fill in truthfully. If you are a student, you can fill in the school name as the unit name, and fill in the position as a student. After a while, there will be a The robot calls your registered mobile phone number and asks you for the name of your unit, what you need to use it for, and just explain it truthfully, such as development and programming learning, artificial intelligence learning, etc.
Afterwards, a text message will be sent to you, which contains a URL, and you can log in to the
login interface after clicking it.
HKUST Xunfei Xinghuo Cognitive Big Model Use
1. Interface introduction
2. Compared with chatgpt 3.5 in terms of programming ability
The first and most simple and crude method I thought of was Likou
I picked a more difficult question to decide the winner?
Both sides here are written in C++.
I will send them all the topics on the left (including hints and examples)
HKUST Xunfei Spark Model
send question
first answer
class Solution {
public:
bool isNumber(string s) {
bool hasE = false;
if (s[0] == 'e' || s[0] == 'E') {
hasE = true;
s = s.substr(1);
}
int i = 0;
while (i < s.length() && (isdigit(s[i]) || s[i] == '.')) {
i++;
}
if (i == 0) {
return false;
}
if (hasE) {
i++;
}
while (i < s.length() && (isdigit(s[i]) || s[i] == '.')) {
i++;
}
return i == s.length();
}
};
For the first time,
1106 use cases were passed. I sent him the prompt information. He replied
that he didn’t know what I was talking about. I wanted to ask again.
The bug started to
get stuck at this position, and I created a new dialogue (Maybe because the question is repeated?)
The answer is the same as last time
class Solution {
public:
bool isNumber(string s) {
bool hasE = false;
if (s[0] == 'e' || s[0] == 'E') {
hasE = true;
s = s.substr(1);
}
int i = 0;
while (i < s.length() && (isdigit(s[i]) || s[i] == '.')) {
i++;
}
if (i == 0) {
return false;
}
if (hasE) {
i++;
}
while (i < s.length() && (isdigit(s[i]) || s[i] == '.')) {
i++;
}
return i == s.length();
}
};
So I sent the reminder message again, this time I will be more specific
His answer is still the same, but there is one more test case
I can only count it passing a large number of use cases.
chatgpt 3.5 model
Enter the same question and
the first answer
code is as follows:
class Solution {
public:
bool isNumber(string s) {
int n = s.length();
int i = 0;
// Skip leading spaces
while (i < n && isspace(s[i])) {
i++;
}
// Check for an optional sign character
if (i < n && (s[i] == '+' || s[i] == '-')) {
i++;
}
// Check for digits before decimal point
bool hasDigitsBeforeDecimal = false;
while (i < n && isdigit(s[i])) {
i++;
hasDigitsBeforeDecimal = true;
}
// Check for optional decimal point
if (i < n && s[i] == '.') {
i++;
// Check for digits after decimal point
while (i < n && isdigit(s[i])) {
i++;
hasDigitsBeforeDecimal = true;
}
}
// Check for optional exponent
if (hasDigitsBeforeDecimal && i < n && (s[i] == 'e' || s[i] == 'E')) {
i++;
// Check for optional sign character in exponent
if (i < n && (s[i] == '+' || s[i] == '-')) {
i++;
}
bool hasDigitsInExponent = false;
// Check for digits in exponent
while (i < n && isdigit(s[i])) {
i++;
hasDigitsInExponent = true;
}
// If there are no digits in exponent, it's not a valid number
if (!hasDigitsInExponent) {
return false;
}
}
// Skip trailing spaces
while (i < n && isspace(s[i])) {
i++;
}
// If we have reached the end of the string, then it's a valid number
return i == n;
}
};
1485 passes for the first time in the result of Lituo , and the difference is a few. We will directly send the error message
code as follows
class Solution {
public:
bool isNumber(string s) {
int n = s.length();
int i = 0;
// Skip leading spaces
while (i < n && isspace(s[i])) {
i++;
}
// Check for an optional sign character
if (i < n && (s[i] == '+' || s[i] == '-')) {
i++;
}
bool hasDigitsBeforeDecimal = false;
bool hasDigitsAfterDecimal = false;
// Check for digits before decimal point
while (i < n && isdigit(s[i])) {
i++;
hasDigitsBeforeDecimal = true;
}
// Check for optional decimal point
if (i < n && s[i] == '.') {
i++;
// Check for digits after decimal point
while (i < n && isdigit(s[i])) {
i++;
hasDigitsAfterDecimal = true;
}
}
// If there are no digits before or after the decimal point, it's not a valid number
if (!hasDigitsBeforeDecimal && !hasDigitsAfterDecimal) {
return false;
}
// Check for optional exponent
if (i < n && (s[i] == 'e' || s[i] == 'E')) {
i++;
// Check for optional sign character in exponent
if (i < n && (s[i] == '+' || s[i] == '-')) {
i++;
}
bool hasDigitsInExponent = false;
// Check for digits in exponent
while (i < n && isdigit(s[i])) {
i++;
hasDigitsInExponent = true;
}
// If there are no digits in exponent, it's not a valid number
if (!hasDigitsInExponent) {
return false;
}
}
// Skip trailing spaces
while (i < n && isspace(s[i])) {
i++;
}
// If we have reached the end of the string, then it's a valid number
return i == n;
}
};
The high score passed after the force button result was corrected according to the wrong information for the second time
3. Compared with new bing in image generation ability
If you ask me why I don’t compare with gpt, then I can only tell you that the author is a student party, without money, can’t afford 4.0
HKUST Xunfei Xinghuo Model
Q: Generate a picture with mountains and rivers as the background, and an old man wearing a hat is fishing by the river, and it is drizzling. newbing Q: Generate a
picture
with mountains and rivers as the background, and there is an old man wearing a hat The old man was fishing by the river, and it was drizzling
emm~, why did he feel like a complete failure
and give him another chance?
Well, it can only be like this.
The next question
is that the Xunfei Xunhuo model of HKUST
started talking nonsense. . . Then it crashed again , although
newbing
understood it, but the size still remained the same. . .
Summarize
Compared with gpt3.5 products in terms of language model, it is indeed lacking, and the image processing ability is also average, but this also shows that there is still room for improvement.
Although this product has some flaws when it first came out, I think it is understandable. After all, this thing is trained slowly. I believe that our domestic AI can slowly move towards a bright future! ! !
(Here is a point, there is no meaning of sarcasm, it is just an objective comparison, it is a new dialogue used, no adjustment is made)