upgraded models

.gitignore

@@ -160,12 +160,6 @@ databases/*
 extensions/
 !extensions/.keep
 
-# backends
-backends/
-!backends/gpt4all
-!backends/llama_cpp
-!backends/gptj
-!backends/__init__.py
 web/.env.build
 web/.env.dev
 web/.env.development

app.py

@@ -23,7 +23,7 @@ import traceback
 import sys
 from tqdm import tqdm
 from pyaipersonality import AIPersonality
-from pyGpt4All.db import DiscussionsDB, Discussion
+from gpt4all_api.db import DiscussionsDB, Discussion
 from flask import (
     Flask,
     Response,
@@ -56,8 +56,8 @@ log = logging.getLogger('werkzeug')
 log.setLevel(logging.ERROR)
 
 import time
-from pyGpt4All.config import load_config, save_config
-from pyGpt4All.api import GPT4AllAPI
+from gpt4all_api.config import load_config, save_config
+from gpt4all_api.api import GPT4AllAPI
 import shutil
 import markdown
 

backends/gpt_4all/__init__.py

@@ -9,8 +9,8 @@
 ######
 from pathlib import Path
 from typing import Callable
-from pygpt4all import GPT4All as Model
-from pyGpt4All.backend import GPTBackend
+from gpt4all import GPT4All
+from gpt4all_api.backend import GPTBackend
 import yaml
 
 __author__ = "parisneo"
@@ -30,12 +30,9 @@ class GPT4ALL(GPTBackend):
             config (dict): The configuration file
         """
         super().__init__(config, False)
-        
-        self.model = Model(
-                model_path=f"./models/gpt4all/{self.config['model']}",
-                prompt_context="", prompt_prefix="", prompt_suffix="",
-                n_ctx=self.config['ctx_size'], 
-                seed=self.config['seed'],
+        self.model = GPT4All.get_model_from_name(self.config['model'])
+        self.model.load_model(
+                model_path=f"./models/gpt_4all/{self.config['model']}",
         )
 
     def stop_generation(self):
@@ -56,7 +53,6 @@ class GPT4ALL(GPTBackend):
             verbose (bool, optional): If true, the code will spit many informations about the generation process. Defaults to False.
         """
         try:
-            self.model.reset()
             for tok in self.model.generate(prompt, 
                                            n_predict=n_predict,                                           
                                             temp=self.config['temperature'],
@@ -64,7 +60,7 @@ class GPT4ALL(GPTBackend):
                                             top_p=self.config['top_p'],
                                             repeat_penalty=self.config['repeat_penalty'],
                                             repeat_last_n = self.config['repeat_last_n'],
-                                            n_threads=self.config['n_threads'],
+                                            # n_threads=self.config['n_threads'],
                                            ):
                 if not new_text_callback(tok):
                     return

backends/gpt_j/__init__.py

@@ -0,0 +1,70 @@
+######
+# Project       : GPT4ALL-UI
+# File          : backend.py
+# Author        : ParisNeo with the help of the community
+# Supported by Nomic-AI
+# Licence       : Apache 2.0
+# Description   : 
+# This is an interface class for GPT4All-ui backends.
+######
+from pathlib import Path
+from typing import Callable
+from pygptj.model import Model
+from gpt4all_api.backend import GPTBackend
+
+__author__ = "parisneo"
+__github__ = "https://github.com/nomic-ai/gpt4all-ui"
+__copyright__ = "Copyright 2023, "
+__license__ = "Apache 2.0"
+
+backend_name = "GptJ"
+
+class GptJ(GPTBackend):
+    file_extension='*.bin'
+    def __init__(self, config:dict) -> None:
+        """Builds a LLAMACPP backend
+
+        Args:
+            config (dict): The configuration file
+        """
+        super().__init__(config, False)
+        
+        self.model = Model(
+                model_path=f"./models/llama_cpp/{self.config['model']}",
+                prompt_context="", prompt_prefix="", prompt_suffix="",
+                n_ctx=self.config['ctx_size'], 
+                seed=self.config['seed'],
+                )
+
+    def stop_generation(self):
+        self.model._grab_text_callback()
+
+    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()
+            for tok in self.model.generate(prompt, 
+                                           n_predict=n_predict,                                           
+                                            temp=self.config['temperature'],
+                                            top_k=self.config['top_k'],
+                                            top_p=self.config['top_p'],
+                                            repeat_penalty=self.config['repeat_penalty'],
+                                            repeat_last_n = self.config['repeat_last_n'],
+                                            n_threads=self.config['n_threads'],
+                                           ):
+                if not new_text_callback(tok):
+                    return
+        except Exception as ex:
+            print(ex)

backends/gptj/__init__.py

@@ -10,7 +10,7 @@
 from pathlib import Path
 from typing import Callable
 from gpt4allj import Model
-from pyGpt4All.backend import GPTBackend
+from gpt4all_api.backend import GPTBackend
 import yaml
 
 __author__ = "parisneo"

backends/llama_cpp/__init__.py

@@ -10,7 +10,7 @@
 from pathlib import Path
 from typing import Callable
 from pyllamacpp.model import Model
-from pyGpt4All.backend import GPTBackend
+from gpt4all_api.backend import GPTBackend
 import yaml
 
 __author__ = "parisneo"

configs/default.yaml

@@ -7,7 +7,7 @@ n_threads: 8
 host: localhost
 language: en-US
 # Supported backends are llamacpp and gpt-j
-backend: llamacpp
+backend: llama_cpp
 model: null
 n_predict: 1024
 nb_messages_to_remember: 5

gpt4all_api/api.py

@@ -10,7 +10,7 @@
 import gc
 import sys
 from datetime import datetime
-from pyGpt4All.db import DiscussionsDB
+from gpt4all_api.db import DiscussionsDB
 from pathlib import Path
 import importlib
 from pyaipersonality import AIPersonality
@@ -96,6 +96,7 @@ class ModelProcess:
             
     def _rebuild_model(self):
         try:
+            print("Rebuilding model")
             self.backend = self.load_backend(Path("backends")/self.config["backend"])
             print("Backend loaded successfully")
             try:
@@ -144,9 +145,12 @@ class ModelProcess:
         self.ready = True
         
         while True:
+            try:
+                self._check_set_config_queue()
                 self._check_cancel_queue()
                 self._check_clear_queue()
 
+                if not self.generate_queue.empty():
                     command = self.generate_queue.get()
                     if command is None:
                         break
@@ -158,7 +162,10 @@ class ModelProcess:
                         while not self.generation_queue.empty():
                             time.sleep(1)
                         self.is_generating.value = 0
- 
+                time.sleep(1)
+            except Exception as ex:
+                time.sleep(1)
+                print(ex)
     def _generate(self, prompt, id, n_predict):
         self.id = id        
         if self.config["override_personality_model_parameters"]:
@@ -196,6 +203,7 @@ class ModelProcess:
         else:
             # Stream the generated text to the main process
             self.generation_queue.put((text,self.id))
+            self._check_set_config_queue()
             self._check_cancel_queue()
             self._check_clear_queue()        
             # if stop generation is detected then stop
@@ -232,6 +240,7 @@ class ModelProcess:
     def _set_config(self, config):
         bk_cfg = self.config
         self.config = config
+        print("Changing configuration")
         # verify that the backend is the same
         if self.config["backend"]!=bk_cfg["backend"] or self.config["model"]!=bk_cfg["model"]:
             self._rebuild_model()

requirements.txt

@@ -7,7 +7,7 @@ markdown
 pyllamacpp==2.1.1
 gpt4all-j
 pygptj
-pygpt4all
+gpt4all
 --find-links https://download.pytorch.org/whl/cu117
 torch==2.0.0
 torchvision

requirements_dev.txt

@@ -6,6 +6,6 @@ pyyaml
 markdown
 pyllamacpp==2.0.0
 gpt4all-j 
-pygpt4all
+gpt4all
 transformers
 pyaipersonality>=0.0.11

web/src/components/ModelEntry.vue

@@ -86,7 +86,6 @@ export default {
     handleSelection() {
       if (this.isInstalled && !this.selected) {
         this.onSelected(this);
-        this.$set(this, 'selected', true);
       }
     }
   }