It's strange to return after executing the command

[Chting]

./main -m ./models/7B/ggml-model-q4_0.bin -t 64 -n 256 --repeat_penalty 1.0 --color -i -r "User:" -p 'What is your name?'

Is it because I haven't installed something？
Centos 7

[gjmulder]

That's normal behaviour currently. People are working on chat interfaces and Python bindings.

[Chting]

Why can't I talk like chatgpt after executing the command according to the prompt

[gjmulder]

Because it is not ChatGPT. The command line interface was provided as a simple way to get the model up and running. Note that there is an interactive mode, but it isn't very user friendly:

./main -h
usage: ./main [options]

options:
  -h, --help            show this help message and exit
  -i, --interactive     run in interactive mode
  --interactive-start   run in interactive mode and poll user input at startup

Check out dalai llama for a quick way to get the 7B model running using Node.js and a simple web-based UI.

[G2G2G2G]

Not sure what these other responses are.
You're not actually giving it any input to know wtf you are talking about, so it is not going to type "User:" ever
make a text.txt file

copy this into it:

Bob is a helpful AI chatbot that always provides useful and detailed answers to User's requests and questions.
Transcript of a dialog, where the User interacts with an Assistant named Bob. Bob is helpful, kind, honest, good at writing, and never fails to answer the User's requests immediately and with precision.
User: Hi. Can you help me with something?
Bob: Hello this is Bob. How can I help?
User: What is the square root of 625?
Bob: 25. The square root of 625 is 25.
User: How hot is the lava in a volcano?
Bob: I can tell you that the temperature is between 700 and 1,200 degrees celsius.
User:

---

this is your command,

./main -m ./models/7B/ggml-model-q4_0.bin -t 64 -n 256 --repeat_penalty 1.0 --color -i -r "User:" -p 'What is your name?'

change it to:

./main -m ./models/7B/ggml-model-q4_0.bin -t 64 -n 1024 --repeat_penalty 1.0 --color -i -r "User:" -f 'text.txt'

(note we now use the file as input instead of your prompt) (also made your -n bigger)
feel free to change the prompt and in the initial line you can give it a different personality and give it different answers to different questions (maybe you want to make it into an asshole instead of helpful bot, for example)

keep in mind due to issue #71 these prompts will have to be short, but ideally you can input massive files so it is super fine tuned.

also wtf is with your -t, are you sure you have 64 threads and the memory bandwidth for that? it prob only needs like -t 8 lmao

as the bot progresses and you give questions with answers you really like, you can delete ones you don't like and replace it with the user/bob that you do like. Make sure the formatting is neat for optimal reading.

edit:
also added a secondary line which gives the bot more prompt of what it is doing, those top two lines can probably be selected one or the other.. or use both, I do.