File size: 16,853 Bytes
3d48e06 0fcfe4b e2e1d2d 0fcfe4b c2bb300 e2e1d2d c2bb300 3d48e06 c2bb300 3d48e06 e2e1d2d 3d48e06 e2e1d2d 3d48e06 e2e1d2d 3d48e06 e2e1d2d 3d48e06 c2bb300 e2e1d2d 3d48e06 c2bb300 e2e1d2d 3d48e06 e2e1d2d 3d48e06 e2e1d2d 3d48e06 e2e1d2d 3d48e06 e2e1d2d 3d48e06 c2bb300 e2e1d2d c2bb300 3d48e06 c2bb300 e2e1d2d 3d48e06 e2e1d2d 3d48e06 c2bb300 e2e1d2d 3d48e06 e2e1d2d 3d48e06 c2bb300 e2e1d2d 3d48e06 c2bb300 e2e1d2d c2bb300 3d48e06 c2bb300 e2e1d2d c2bb300 e2e1d2d c2bb300 e2e1d2d c2bb300 e2e1d2d c2bb300 e2e1d2d c2bb300 e2e1d2d 3d48e06 |
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 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 |
from scripts.parsing_utils import load_yaml_file, get_roadmap_phases, get_project_rules
import os
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
import yaml
import logging
import torch # ADD THIS LINE - Import torch
logging.basicConfig(level=logging.ERROR,
format='%(asctime)s - %(levelname)s - %(message)s')
class ProjectGuidanceChatbot:
def __init__(self, roadmap_file, rules_file, config_file, code_templates_dir):
self.roadmap_file = roadmap_file
self.rules_file = rules_file
self.config_file = config_file
self.code_templates_dir = code_templates_dir
self.roadmap_data = load_yaml_file(self.roadmap_file)
self.rules_data = load_yaml_file(self.rules_file)
self.config_data = load_yaml_file(self.config_file)
self.phases = get_roadmap_phases(self.roadmap_data)
self.rules = get_project_rules(self.rules_data)
self.chatbot_config = self.config_data.get('chatbot', {}) if self.config_data else {}
self.model_config = self.config_data.get('model_selection', {}) if self.config_data else {}
self.response_config = self.config_data.get('response_generation', {}) if self.config_data else {}
self.available_models_config = self.config_data.get('available_models', {}) if self.config_data else {}
self.max_response_tokens = self.chatbot_config.get('max_response_tokens', 200)
self.current_phase = None
self.active_model_key = self.chatbot_config.get('default_llm_model_id')
self.active_model_info = self.available_models_config.get(self.active_model_key)
self.llm_model = None
self.llm_tokenizer = None
self.load_llm_model(self.active_model_info)
self.update_mode_active = False
def load_llm_model(self, model_info):
"""Loads the LLM model and tokenizer based on model_info with 4-bit quantization."""
if not model_info:
error_message = "Error: Model information not provided."
logging.error(error_message)
self.llm_model = None
self.llm_tokenizer = None
return
model_id = model_info.get('model_id')
model_name = model_info.get('name')
if not model_id:
error_message = f"Error: 'model_id' not found for model: {model_name}"
logging.error(error_message)
self.llm_model = None
self.llm_tokenizer = None
return
print(f"Loading model: {model_name} ({model_id}) with 4-bit quantization...") # Indicate quantization
try:
bnb_config = BitsAndBytesConfig( # Configure 4-bit quantization
load_in_4bit=True,
bnb_4bit_quant_type="nf4", # "nf4" is recommended for Llama models
bnb_4bit_compute_dtype=torch.bfloat16, # Or torch.float16 if bfloat16 not supported
)
self.llm_tokenizer = AutoTokenizer.from_pretrained(model_id)
self.llm_model = AutoModelForCausalLM.from_pretrained(
model_id,
device_map="auto",
quantization_config=bnb_config # Apply quantization config
)
print(f"Model {model_name} loaded successfully with 4-bit quantization.") # Indicate quantization success
except Exception as e:
error_message = f"Error loading model {model_name} ({model_id}) with 4-bit quantization: {e}"
logging.exception(error_message)
self.llm_model = None
self.llm_tokenizer = None
self.active_model_info = model_info
def switch_llm_model(self, model_key):
"""Switches the active LLM model based on the provided model key."""
if model_key in self.available_models_config:
model_info = self.available_models_config[model_key]
print(f"Switching LLM model to: {model_info.get('name')}")
self.load_llm_model(model_info)
self.active_model_key = model_key
return f"Switched to model: {model_info.get('name')}"
else:
error_message = f"Error: Model key '{model_key}' not found in available models."
logging.error(error_message)
return error_message
def enter_update_mode(self):
"""Enters the chatbot's update mode."""
self.update_mode_active = True
return "Entering update mode. Please enter configuration commands (or 'sagor is python/help' for commands)."
def exit_update_mode(self):
"""Exits the chatbot's update mode and reloads configuration."""
self.update_mode_active = False
self.reload_config()
return "Exiting update mode. Configuration reloaded."
def reload_config(self):
"""Reloads configuration files."""
print("Reloading configuration...")
try:
self.config_data = load_yaml_file(self.config_file)
self.roadmap_data = load_yaml_file(self.roadmap_file)
self.rules_data = load_yaml_file(self.rules_file)
self.chatbot_config = self.config_data.get('chatbot', {}) if self.config_data else {}
self.model_config = self.config_data.get('model_selection', {}) if self.config_data else {}
self.response_config = self.config_data.get('response_generation', {}) if self.config_data else {}
self.available_models_config = self.config_data.get('available_models', {}) if self.config_data else {}
self.max_response_tokens = self.chatbot_config.get('max_response_tokens', 200)
self.phases = get_roadmap_phases(self.roadmap_data)
self.rules = get_project_rules(self.rules_data)
print("Configuration reloaded.")
except Exception as e:
error_message = f"Error reloading configuration files: {e}"
logging.exception(error_message)
print(error_message)
def get_chatbot_greeting(self):
current_model_name = self.active_model_info.get('name', 'Unknown Model') if self.active_model_info else 'Unknown Model'
return f"Hello! I am the {self.chatbot_config.get('name', 'Project Guidance Chatbot')}. Currently using **{current_model_name}** (4-bit quantized). Max response tokens: {self.max_response_tokens}. {self.chatbot_config.get('description', 'How can I help you with your project?')}" # Indicate quantization in greeting
def generate_llm_response(self, user_query):
"""Generates a response using the currently active LLM."""
if not self.llm_model or not self.llm_tokenizer:
error_message = "LLM model not loaded. Please select a model."
logging.error(error_message)
return error_message
try:
inputs = self.llm_tokenizer(user_query, return_tensors="pt").to(self.llm_model.device)
outputs = self.llm_model.generate(**inputs, max_length=self.max_response_tokens, num_beams=5, no_repeat_ngram_size=2, early_stopping=True)
response = self.llm_tokenizer.decode(outputs[0], skip_special_tokens=True)
return response
except Exception as e:
error_message = f"Error generating LLM response: {e}"
logging.exception(error_message)
return self.response_config.get('error_message', "Sorry, something went wrong.")
def process_query(self, user_query):
secret_command_prefix = "sagor is python/"
if user_query.startswith(secret_command_prefix):
command = user_query[len(secret_command_prefix):].strip()
if command == "start":
return self.enter_update_mode()
elif command == "exit":
return self.exit_update_mode()
elif command == "help":
return self.get_update_mode_help()
elif self.update_mode_active:
parts = command.split()
if parts and parts[0] == "modify_rule":
if len(parts) >= 4:
category = parts[1]
rule_key = parts[2]
new_rule_text_parts = parts[3:]
new_rule_text = " ".join(new_rule_text_parts)
return self.modify_rule_in_config(category, rule_key, new_rule_text)
else:
return "Invalid 'modify_rule' command format. Usage: modify_rule <category> <rule_key> <new rule text>"
elif parts and parts[0] == "add_phase":
return "Command 'add_phase' is not yet implemented in this example."
else:
return f"Unknown update command: {command}. Type 'sagor is python/help' for commands."
else:
return f"Unknown command (not in update mode): {command}. To enter update mode, type 'sagor is python/start'."
if self.update_mode_active:
return "In update mode. Please enter a configuration command (or 'sagor is python/help' for commands)."
if not self.phases:
return "Error: Roadmap data not loaded correctly."
if not self.rules:
return "Error: Rules data not loaded correctly."
query_lower = user_query.lower()
if "roadmap" in query_lower:
return self.get_roadmap_summary()
if "rules" in query_lower:
return self.get_rules_summary()
if "switch model to" in query_lower:
parts = user_query.lower().split("switch model to")
if len(parts) > 1:
model_name_or_key = parts[1].strip()
for key, model_data in self.available_models_config.items():
if model_data['name'].lower() == model_name_or_key.lower() or key.lower() == model_name_or_key.lower():
switch_result = self.switch_llm_model(key)
return switch_result + "\n" + self.get_chatbot_greeting()
return f"Model '{model_name_or_key}' not found in available models."
if self.current_phase:
current_phase_data = self.phases.get(self.current_phase)
if current_phase_data:
phase_response = self.get_phase_guidance(current_phase_data, user_query)
if phase_response:
return phase_response
for phase_key, phase_data in self.phases.items():
if phase_data['name'].lower() in query_lower:
self.current_phase = phase_key
return self.get_phase_introduction(phase_data)
llm_response = self.generate_llm_response(user_query)
if llm_response:
return llm_response
return self.response_config.get('default_instruction', "How can I help you with your project phases or general questions?")
def get_update_mode_help(self):
"""Provides help message for update mode commands."""
help_message = "Update Mode Commands:\n"
help_message += "- `sagor is python/exit`: Exit update mode and reload configuration.\n"
help_message += "- `sagor is python/modify_rule <category> <rule_key> <new rule text>`: Modify a rule in rules.yaml.\n"
help_message += " Example: `sagor is python/modify_rule general rule_1 Prioritize open and responsible AI.`\n"
help_message += "- `sagor is python/add_phase ...`: (Not yet implemented) Add a new phase to roadmap.yaml.\n"
help_message += "- `sagor is python/help`: Show this help message.\n"
help_message += "\nMake sure to use the correct syntax for commands. After exiting update mode, the chatbot will reload the configuration."
return help_message
def modify_rule_in_config(self, category, rule_key, new_rule_text):
"""Modifies a rule in the rules.yaml configuration."""
if not self.rules_data or 'project_rules' not in self.rules_data:
error_message = "Error: Rules data not loaded or invalid format."
logging.error(error_message)
return error_message
if category not in self.rules_data['project_rules']:
error_message = f"Error: Rule category '{category}' not found."
logging.error(error_message)
return error_message
if rule_key not in self.rules_data['project_rules'][category]:
error_message = f"Error: Rule key '{rule_key}' not found in category '{category}'."
logging.error(error_message)
return error_message
self.rules_data['project_rules'][category][rule_key] = new_rule_text
try:
with open(self.rules_file, 'w') as f:
yaml.dump(self.rules_data, f, indent=2)
self.reload_config()
return f"Rule '{rule_key}' in category '{category}' updated to: '{new_rule_text}'. Configuration reloaded."
except Exception as e:
error_message = f"Error saving changes to {self.rules_file}: {e}"
logging.exception(error_message)
return error_message
def get_roadmap_summary(self):
summary = "Project Roadmap:\n"
for phase_key, phase_data in self.phases.items():
summary += f"- **Phase: {phase_data['name']}**\n"
summary += f" Description: {phase_data['description']}\n"
summary += f" Milestones: {', '.join(phase_data['milestones'])}\n"
return summary
def get_rules_summary(self):
summary = "Project Rules:\n"
for rule_category, rules_list in self.rules.items():
summary += f"**{rule_category.capitalize()} Rules:**\n"
for rule_key, rule_text in rules_list.items():
summary += f"- {rule_text}\n"
return summary
def get_phase_introduction(self, phase_data):
return f"Okay, let's focus on **Phase: {phase_data['name']}**. \nDescription: {phase_data['description']}. \nKey milestones are: {', '.join(phase_data['milestones'])}. \nWhat would you like to know or do in this phase?"
def get_phase_guidance(self, phase_data, user_query):
query_lower = user_query.lower()
if "milestones" in query_lower:
return "The milestones for this phase are: " + ", ".join(phase_data['milestones'])
if "actions" in query_lower or "how to" in query_lower:
if 'actions' in phase_data:
return "Recommended actions for this phase: " + ", ".join(phase_data['actions'])
else:
return "No specific actions are listed for this phase in the roadmap."
if "code" in query_lower or "script" in query_lower:
if 'code_generation_hint' in phase_data:
template_filename_prefix = phase_data['name'].lower().replace(" ", "_")
template_filepath = os.path.join(self.code_templates_dir, f"{template_filename_prefix}_template.py.txt")
if os.path.exists(template_filepath):
code_snippet = self.generate_code_snippet(template_filepath, phase_data)
return "Here's a starting code snippet for this phase:\n\n```python\n" + code_snippet + "\n```\n\nRemember to adapt it to your specific needs."
else:
return f"A code template for this phase ({phase_data['name']}) is not yet available. However, the hint is: {phase_data['code_generation_hint']}"
else:
return "No code generation hint is available for this phase."
return f"For phase '{phase_data['name']}', remember the description: {phase_data['description']}. Consider the milestones and actions. What specific aspect are you interested in?"
def generate_code_snippet(self, template_filepath, phase_data):
"""Generates code snippet from a template file. (Simple template filling example)"""
try:
with open(template_filepath, 'r') as f:
template_content = f.read()
code_snippet = template_content.replace("{{phase_name}}", phase_data['name'])
return code_snippet
except FileNotFoundError:
return f"Error: Code template file not found at {template_filepath}"
except Exception as e:
return f"Error generating code snippet: {e}"
# Example usage (for testing - remove or adjust for app.py)
if __name__ == '__main__':
chatbot = ProjectGuidanceChatbot(
roadmap_file="roadmap.yaml",
rules_file="rules.yaml",
config_file="configs/chatbot_config.yaml",
code_templates_dir="scripts/code_templates"
)
print(chatbot.get_chatbot_greeting())
while True:
user_input = input("You: ")
if user_input.lower() == "exit":
break
response = chatbot.process_query(user_input)
print("Chatbot:", response) |