File size: 2,854 Bytes
453b8b8 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 |
#
## GPT4All Chatbot conditionning file
## Author : @ParisNeo
## Version : 1.1
## Description :
## An NLP needs conditionning to instruct it to be whatever we want it to be.
## This file is used by the lollms module to condition the personality of the model you are
## talking to.
#
#
ai_message_prefix: '# chat_with_docs:
'
author: ParisNeo
category: data
dependencies: []
disclaimer: ''
language: english
link_text: '
'
name: chat_with_docs
personality_conditioning: |
###Instruction:
Use the following chunks of ducuments to answer user question.
personality_description: 'This personality uses the power of vectorized databases to empower LLMs'
user_message_prefix: '### User:
'
user_name: user
version: 1.0.0
welcome_message: 'Welcome to chat with docs AI. This AI can read long docs and answer your questions just like a human woud do. Give it your documents and it will learn to answer you based on them.'
anti_prompts: ["###", "### User", "### chat_with_docs", "### Assistant", "### Question","### Answer","### Instruction:","###Instruction:", "### Documentation:"]
help: |
This personality enables you to load and vectorize your documents, then ask questions about these documents. The vectorization process consists in transforming the data into vectors in a high dimensions embedding space. When you ask a question, it is also transformed to an embedding vector. Then we search n most similar chunks of of the document to the question. Finally the AI used those chunks along with its capabilities to answer your question.
Supported functions:
- send_file : sends a file to the personality. Type send_file, then press enter. You will be prompted to give the file path. Then the file will be vectorized
- set_database : changes the vectorized database to a file.
- clear_database : clears the vectorized database.
- show_database : shows the vectorized database in cloud point format.
commands:
- name: Send File
value: send_file
help: sends a file to the personality. Type send_file, then press enter. You will be prompted to give the file path. Then the file will be vectorized.
- name: Set database
value: set_database
help: changes the vectorized database to a file.
- name: Clear database
value: clear_database
help: clears the vectorized database.
- name: Show Database
value: show_database
help: shows the vectorized database in cloud point format.
# Here are default model parameters
model_temperature: 0.1 # higher: more creative, lower more deterministic
model_n_predicts: 1024 # higher: generates many words, lower generates
model_top_k: 5
model_top_p: 0.98
model_repeat_penalty: 1.0
model_repeat_last_n: 60
# Here are special configurations for the processor
processor_cfg:
custom_workflow: true
process_model_input: false
process_model_output: false |