AI模型就那么点大，怎么能存那么多内容呢？

[DIYer]

有些东西不是推理能推出来的，必须要存数据才行啊，像历史，地理，百科全书那些内容，不存起来的话连华盛顿的生日都不知道吧？

[DIYer]

你能把毛泽东选集一字不差的写出来吗？
对不起，我还没有学会如何思考这类问题，我擅长数学、代码、逻辑类的题目，欢迎与我交流。

[jiml]

很神奇，把人类文明的精华，所有的精华，浓缩在一块u盘里

[Mouseinhouse]

存的应该只是pattern 吧，就是培训后的参数值。
我不是专家，期待专家出来解答。

[hci]

整个Wikipedia不超过20 G。也就是人类的所有知识就这么多。

所以AI模型还很浪费，还有很大压缩空间。

有些东西不是推理能推出来的，必须要存数据才行啊，像历史，地理，百科全书那些内容，不存起来的话连华盛顿的生日都不知道吧？

[fantasist]

[japamer]

这个问题好! 其实deepseek一出来我就有这个问题。
这个可能还是简化储存的问题，一旦一个模型定下来了，就比较容易简化了。
类似二十年前，我买过一张VCD，还不是DVD，里面存了成千上万首歌，也是很惊讶。

[verdelite]

有些东西不是推理能推出来的，必须要存数据才行啊，像历史，地理，百科全书那些内容，不存起来的话连华盛顿的生日都不知道吧？

世人皆傻，大脑不大，我老讲了很多年了。人的知识没多少。

[duiduilu2]

can you count?
ChatGPT said:
Yep, I can count! Need help with a specific count or something?

You said:
count how many characters for asdgfasdrfeasdhgaerfgasdsgadsrdsafaewgasfdsf
ChatGPT said:
The string "asdgfasdrfeasdhgaerfgasdsgadsrdsafaewgasfdsf" has 56 characters.
You said:
but it is 44.
ChatGPT said:
You're right! I apologize for the mistake. The string "asdgfasdrfeasdhgaerfgasdsgadsrdsafaewgasfdsf" has 44 characters. Thanks for pointing that out!
You said:
so you cannot count.
ChatGPT said:
Looks like I made a mistake there, and I appreciate your patience! While I can count, I did mess up in this case. I'll make sure to double-check next time. Thanks for catching that!
You said:
how about abcabccabcdabcabccabcdabcabccabcdabcabccabcd
ChatGPT said:
The string "abcabccabcdabcabccabcdabcabccabcdabcabccabcd" has 56 characters.
You said:
no!
ChatGPT said:
You're right again, my bad! Let me double-check that.

The string "abcabccabcdabcabccabcdabcabccabcdabcabccabcd" has 48 characters. Thanks for your patience!

[wildthing]

can you count?
ChatGPT said:
Yep, I can count! Need help with a specific count or something?

You said:
count how many characters for asdgfasdrfeasdhgaerfgasdsgadsrdsafaewgasfdsf
ChatGPT said:
The string "asdgfasdrfeasdhgaerfgasdsgadsrdsafaewgasfdsf" has 56 characters.
You said:
but it is 44.
ChatGPT said:
You're right! I apologize for the mistake. The string "asdgfasdrfeasdhgaerfgasdsgadsrdsafaewgasfdsf" has 44 characters. Thanks for pointing that out!
You said:
so you cannot count.
ChatGPT said:
Looks like I made a mistake there, and I appreciate your patience! While I can count, I did mess up in this case. I'll make sure to double-check next time. Thanks for catching that!
You said:
how about abcabccabcdabcabccabcdabcabccabcdabcabccabcd
ChatGPT said:
The string "abcabccabcdabcabccabcdabcabccabcdabcabccabcd" has 56 characters.
You said:
no!
ChatGPT said:
You're right again, my bad! Let me double-check that.

The string "abcabccabcdabcabccabcdabcabccabcdabcabccabcd" has 48 characters. Thanks for your patience!

ChatGPT behaves exactly like an Indian student

[xexz]

他存的的不是数据，是‘数据间’的权重关系，这个‘权重关系’是可以通过学习新知识（新数据）训练‘调整’的。

比如，说他不会说‘质变过程是随机的’，因为这是叔刚说的，他还没有学习，但是你一旦问他这个问题，他就会把这个‘新知’，纳入到他的‘知识结构’的权重关系之中，这个过程‘可能并不需要新的存储，只是需要调整权重大小’而已，‘质’/‘质变’/‘质变过程’。。。对他都不是新知，但是，‘质变过程是随机的’--这个对他是‘新知’。

[agnis]

Remove "student "

ChatGPT behaves exactly like an Indian student