+ 1
Which dataset i can add
I have alraedy information about 1800 things gk 400,all countries famous dish and math calculation problem. Which of what information i can add in my code for answere. Have any one suggestion or idea. https://sololearn.com/compiler-playground/W1xy6u9HeM2I/?ref=app
2 Respostas
+ 6
The design of your AI requires people to ask specific questions or mention specific items. Unfortunately, the number of questions people might ask are endless and the items people might mention are endless. Further, it doesn't support context. So if I ask, "Can I eat a fox", it tells me what is a fox. If I ask a question using only a couple words from a question, such as "largest continent", it doesn't understand the query.
It's a fun project. But with this design, you will never have enough questions listed or enough items described. Again, it's a great project. But it won't work like an LLM because you are not parsing the questions and building the context of the question.
A proper language model has to decipher all verbs, all nouns, as well as adjectives and adverbs. That's why ChatGPT is so powerful. If I asked what is the difference between a red fox and a brown fox. It should realize that:
object 1 is red fox
object 2 is brown fox
question is: what is the difference
etc.
So it's a cool app you've written. But the structure of how you have put it together makes it impossible to expand it to any level of completion.
If you want to try and expand the capabilities, try to identify small phrases and words representing the objects and verbs. Then add a matrix that combines the verbs with the objects. For each object, there can be different answers for different verbs.
can I throw a fox?
can I eat a fox?
can a fox run?
how fast is a fox?
With 20-50 possible verbs and thousands of possible objects, you can see your responses should find one of many verbs and one of many objects and then create an answer from those combinations. Keeping it simple, you would have at least two indexes to create one response.
object: fox, cow, banana, etc.
verb: "what is", "how old", "age", "size", etc
response" [ based on combined choices]
fox, eats, "A fox eats rodents"
fox, age, "A fox can live ten years"
fox, run, "A fox can run 12 mph"
fox, size, "A fox is up to 15 lbs"
And so on for every object/verb.
+ 1
I agree with you