Assignment

LLM stands for large language model

LLM is a computer advanced program which helps the user by giving the solution on the basis of the input of the user. The main advantage of this program is, this program is very good in guessing and all these guessing is come from the data in which this program (Model) is trained on.

Popular example of LLM are GPT, Claude, Deep Seek, Gemini, lovable, boat and many more

There are a very wide scope of LLM are for example you have an essay then you can improve the essay as well as if you give an image of a math problem or directly give a math problem then they can solve that problem also in IT industry if you give the some errors or any requirement related to the coding then with the help of LLM models you can solve it also, you can also create the complex UI and Frontend design with the help of them.

When you send a message to any LLM model then there some operations were happened before the output of the result.

1. So for example if you send the message "Hello, how are you ?" then first this sentence is convert into the number or Hash_Code like this "2356 256 2547 5975 2687" that stage is called token visualizer and now these token/Hash were go inside the LLM model and now LLM model will processed this data

   1. According to my info, Processing also categorized in various step like

      1. Embedding : If input is related to something or any object or any previous chat then data is embedded with them

      2. Encoding: try to understanding the flow of data

      3. Self Attention: link the input string with the past info/chat and try to remove the ambiguity from the string

      4. Neural Network: give the possible outcome of the solution

      5. Probability: pick that solution according to the temp. of the data

2. Response is not copy from the internet because they are generating the data by guessing the output according to the input of the string and this loop is continue until the output string is considered as ok for the LLM model

Computer not Under Stand human language because in human language there is a lot of ambiguity were there

every text has different meaning according to the sentence but if we convert text into a number then we perform mathematical operation on the numbers, and computer understand the mathematics because in maths the rule is defined

A transformer is the backbone of the LLM model the processing of the data and give the valid output of the data (according to the given string) were occurred inside the section and that section is called a transformer. and there is white paper were also available which tells the flow of data inside the LLM model "Attention is all you need".

Temperature is also called a random_ness. So if you set the Temp. is low then the output of the answer is more accurate but they takes more time and more energy and more power at the same time if the Temp is high then the output of the answer is less accurate but they takes less time and less energy and less power