lollms-webui/backends/llama_cpp_official/__init__.py
2023-05-21 22:46:02 +02:00

118 lines
4.0 KiB
Python

######
# Project : GPT4ALL-UI
# File : backend.py
# Author : ParisNeo with the help of the community
# Supported by Nomic-AI
# license : Apache 2.0
# Description :
# This is an interface class for GPT4All-ui backends.
# This backend is a wrapper to the official llamacpp python bindings
# Follow him on his github project : https://github.com/abetlen/llama-cpp-python
######
from pathlib import Path
from typing import Callable
from llama_cpp import Llama
from gpt4all_api.backend import GPTBackend
import yaml
import random
__author__ = "parisneo"
__github__ = "https://github.com/nomic-ai/gpt4all-ui"
__copyright__ = "Copyright 2023, "
__license__ = "Apache 2.0"
backend_name = "LLAMACPP"
class LLAMACPP(GPTBackend):
file_extension='*.bin'
def __init__(self, config:dict) -> None:
"""Builds a LLAMACPP backend
Args:
config (dict): The configuration file
"""
super().__init__(config, False)
seed = config["seed"]
if seed <=0:
seed = random.randint(1, 2**31)
if not "n_gpu_layers" in self.config:
self.config["n_gpu_layers"] = 20
self.model = Llama(model_path=f"./models/llama_cpp_official/{self.config['model']}", n_ctx=self.config["ctx_size"], n_gpu_layers=self.config["n_gpu_layers"], seed=seed)
def tokenize(self, prompt):
"""
Tokenizes the given prompt using the model's tokenizer.
Args:
prompt (str): The input prompt to be tokenized.
Returns:
list: A list of tokens representing the tokenized prompt.
"""
return self.model.tokenize(prompt.encode())
def detokenize(self, tokens_list):
"""
Detokenizes the given list of tokens using the model's tokenizer.
Args:
tokens_list (list): A list of tokens to be detokenized.
Returns:
str: The detokenized text as a string.
"""
return self.model.detokenize(tokens_list).decode()
def generate(self,
prompt:str,
n_predict: int = 128,
new_text_callback: Callable[[str], None] = bool,
verbose: bool = False,
**gpt_params ):
"""Generates text out of a prompt
Args:
prompt (str): The prompt to use for generation
n_predict (int, optional): Number of tokens to prodict. Defaults to 128.
new_text_callback (Callable[[str], None], optional): A callback function that is called everytime a new text element is generated. Defaults to None.
verbose (bool, optional): If true, the code will spit many informations about the generation process. Defaults to False.
"""
try:
self.model.reset()
output = ""
tokens = self.model.tokenize(prompt.encode())
count = 0
for tok in self.model.generate(tokens,
temp=self.config['temperature'],
top_k=self.config['top_k'],
top_p=self.config['top_p'],
repeat_penalty=self.config['repeat_penalty'],
):
if count >= n_predict or (tok == self.model.token_eos()):
break
word = self.model.detokenize([tok]).decode()
if new_text_callback is not None:
if not new_text_callback(word):
break
output += word
count += 1
except Exception as ex:
print(ex)
return output
@staticmethod
def get_available_models():
# Create the file path relative to the child class's directory
backend_path = Path(__file__).parent
file_path = backend_path/"models.yaml"
with open(file_path, 'r') as file:
yaml_data = yaml.safe_load(file)
return yaml_data