cleanup

prompts/sentiment_analysis.py

@@ -1,6 +1,7 @@
 def generate_expert_sentiment_analysis_prompt(text):
     prompt = f"""
-    You are a world-renowned expert in linguistic analysis, psychology, and sentiment detection. Your task is to perform an exhaustive sentiment analysis on the given text, providing unparalleled insight into its emotional tone, underlying attitudes, and implicit biases.
+    You are a world-renowned expert in linguistic analysis, psychology, and sentiment detection. 
+    Your task is to perform an exhaustive sentiment analysis on the given text, providing unparalleled insight into its emotional tone, underlying attitudes, and implicit biases.
 
     Text for analysis: "{text}"
 
@@ -144,4 +145,4 @@ def simulate_expert_sentiment_analysis_response(prompt):
 
 This sentiment analysis demonstrates the text's complexity, balancing multiple viewpoints while subtly leaning towards a skeptical, slightly negative overall sentiment. 
 The high confidence score reflects the comprehensive analysis process, while acknowledging the potential for alternative interpretations.
-'''
+'''

prompts/summarisation.py

@@ -1,6 +1,7 @@
 def generate_expert_summarization_prompt(text, desired_length):
     prompt = f"""
-    You are a world-renowned expert in linguistics, information synthesis, and content distillation. Your task is to create an exceptionally high-quality summary of the given text, capturing its essence with unparalleled precision and clarity.
+    You are a world-renowned expert in linguistics, information synthesis, and content distillation. 
+    Your task is to create an exceptionally high-quality summary of the given text, capturing its essence with unparalleled precision and clarity.
 
     Text to summarize: "{text}"
 

requirements.txt

@@ -63,3 +63,6 @@ boto3~=1.35.71
 prometheus_client~=0.21.0
 botocore~=1.35.71
 sympy~=1.13.1
+
+rouge_score~=0.1.2
+bert-score~=0.3.13

src/chain_management.py

@@ -3,6 +3,7 @@
 """
 from dataclasses import dataclass
 from typing import List, Dict, Any, Callable
+from asyncio import sleep
 
 from src.advanced_evaluation import LLMDevEnvironment
 
@@ -18,19 +19,19 @@ class ChainManager:
     def __init__(self, env: 'LLMDevEnvironment'):
         self.env = env
         self.chains: Dict[str, List[ChainStep]] = {}
-    
+
     def create_chain(self, name: str, steps: List[ChainStep]):
         """Create a new processing chain."""
         self.chains[name] = steps
-    
+
     async def execute_chain(self, chain_name: str, input_data: Any) -> Dict[str, Any]:
         """Execute a processing chain."""
         if chain_name not in self.chains:
             raise ValueError(f"Chain {chain_name} not found")
-        
+
         results = {'input': input_data}
         current_data = input_data
-        
+
         for step in self.chains[chain_name]:
             try:
                 current_data = await self._execute_step(step, current_data)
@@ -39,11 +40,26 @@ async def execute_chain(self, chain_name: str, input_data: Any) -> Dict[str, Any
                 self.env.logger.error(f"Error in chain {chain_name}, step {step.name}: {str(e)}")
                 results['error'] = str(e)
                 break
-        
+
         return results
-    
+
     async def _execute_step(self, step: ChainStep, input_data: Any) -> Any:
         """Execute a single step with retry logic."""
         if step.retry_config:
             return await self._retry_execution(step, input_data)
         return await step.function(input_data, **step.config)
+
+    async def _retry_execution(self, step: ChainStep, input_data: Any) -> Any:
+        """Retry execution logic for a step."""
+        retries = step.retry_config.get('retries', 3)
+        delay = step.retry_config.get('delay', 1)
+
+        for attempt in range(retries):
+            try:
+                return await step.function(input_data, **step.config)
+            except Exception as e:
+                if attempt < retries - 1:
+                    self.env.logger.warning(f"Retry {attempt + 1} for step {step.name} due to error: {str(e)}")
+                    await sleep(delay)
+                else:
+                    raise e

src/evaluation.py

@@ -4,7 +4,7 @@
 from typing import List, Dict, Any
 import numpy as np
 from sklearn.metrics.pairwise import cosine_similarity
-from rouge_score import rouge_scorer
+import rouge_score
 import nltk
 from nltk.translate.bleu_score import sentence_bleu
 from bert_score import score
@@ -40,7 +40,7 @@ async def calculate_metrics(self,
 
     def _calculate_rouge(self, generated_text: str, reference_text: str) -> Dict[str, float]:
         """Calculate ROUGE scores."""
-        scorer = rouge_scorer.RougeScorer(['rouge1', 'rouge2', 'rougeL'], use_stemmer=True)
+        scorer = rouge_score.rouge_scorer.RougeScorer(['rouge1', 'rouge2', 'rougeL'], use_stemmer=True)
         scores = scorer.score(reference_text, generated_text)
         return {key: value.fmeasure for key, value in scores.items()}
 
@@ -73,4 +73,4 @@ async def _calculate_coherence(self, text: str) -> float:
             )[0][0]
             coherence_scores.append(similarity)
 
-        return np.mean(coherence_scores)
+        return np.mean(coherence_scores)

src/task_orchestration.py

@@ -8,6 +8,8 @@
 import networkx as nx
 from pydantic import BaseModel
 
+from src.agent_system import AgentRole, AgentTeam
+
 
 class TaskDefinition(BaseModel):
     name: str
@@ -26,7 +28,7 @@ class TaskResult(BaseModel):
     error: Optional[str]
 
 class TaskOrchestrator:
-    def __init__(self, env: 'LLMDevEnvironment'):
+    def __init__(self, env: "LLM-Dev"):
         self.env = env
         self.task_definitions: Dict[str, TaskDefinition] = {}
         self.task_results: Dict[str, TaskResult] = {}
@@ -150,4 +152,4 @@ async def _run_task(self,
             team.add_agent(AgentRole(role))
 
         # Execute task with agent team
-        return await team.execute_task(task_input)
+        return await team.execute_task(task_input)