fix: clean up all ruff lint errors (closes #275)

optillm/__init__.py

@@ -14,7 +14,8 @@
         del _os.environ[_hf_token_var]
 
 # Import from server module
-from .server import (
+from .server import (  # noqa: E402
+
     main,
     server_config,
     app,

optillm/autothink/classifier.py

@@ -6,7 +6,7 @@
 """
 
 import logging
-from typing import Dict, Any, Tuple, Optional, List, Union
+from typing import Tuple, List
 import os
 import sys
 
@@ -40,7 +40,6 @@ def _load_model(self):
             except ImportError:
                 logger.info("Installing adaptive-classifier library...")
                 os.system(f"{sys.executable} -m pip install adaptive-classifier")
-                import adaptive_classifier
 
             # Import the AdaptiveClassifier class
             from adaptive_classifier import AdaptiveClassifier

optillm/autothink/processor.py

@@ -9,7 +9,7 @@
 import random
 import logging
 from transformers import PreTrainedModel, PreTrainedTokenizer, DynamicCache
-from typing import Dict, List, Any, Optional, Union, Tuple
+from typing import Dict, List, Any, Tuple
 
 from .classifier import ComplexityClassifier
 from .steering import SteeringVectorManager, install_steering_hooks, remove_steering_hooks

optillm/autothink/steering.py

@@ -10,8 +10,7 @@
 import random
 import json
 import datasets
-from typing import Dict, List, Any, Tuple, Optional, Union
-from collections import defaultdict
+from typing import Dict, List, Any, Tuple, Optional
 
 logger = logging.getLogger(__name__)
 
@@ -540,7 +539,6 @@ def update_token_history(self, new_tokens: List[int]):
         if random.random() < 0.01:
             logger.debug(f"STEERING: Token history updated, now has {len(self.token_history)} tokens")
 
-    def update_context(self, new_tokens: str):
         """
         Update the context buffer with new tokens.
 

optillm/batching.py

@@ -17,7 +17,7 @@
 import queue
 import time
 import logging
-from typing import Dict, List, Any, Tuple, Optional
+from typing import Dict, List, Any, Optional
 from concurrent.futures import Future
 from dataclasses import dataclass
 

optillm/bon.py

@@ -1,5 +1,4 @@
 import logging
-import optillm
 from optillm import conversation_logger
 
 logger = logging.getLogger(__name__)

optillm/cepo/cepo.py

@@ -6,12 +6,10 @@
 import time
 import math_verify
 
-from optillm import conversation_logger
 from collections import Counter
 from concurrent.futures import ThreadPoolExecutor, as_completed
 from dataclasses import dataclass
 from typing import Literal, Any, Optional
-from cerebras.cloud.sdk import BadRequestError as CerebrasBadRequestError
 from openai import BadRequestError as OpenAIBadRequestError
 from openai import InternalServerError as OpenAIInternalServerError
 
@@ -325,7 +323,7 @@ def llm_call_reason_effort_fallback(
             if len(reasoning_effort_levels) == 1 and bre.message.startswith("Error code: 400 - {'error': {'message': 'think value"):
                 logger.info(f"The think level {effort} was not supported by the model; Disabling thinking")
                 cepo_config.use_reasoning = False
-        except (OpenAIBadRequestError, OpenAIInternalServerError) as e:
+        except (OpenAIBadRequestError, OpenAIInternalServerError):
             # After 2 retries at this reasoning effort level it failed with error 400/500, lower level
             logger.debug(f"400/500 persisted after retries at reasoning effort {effort}; degrading effort")
             if logger.getEffectiveLevel() == logging.DEBUG:
@@ -491,9 +489,9 @@ def generate_single_plan(i):
         messages.append({"role": "assistant", "content": response})
 
         plans.append(response)
-        cb_log[f"messages_planning_fallback_used"] = messages
+        cb_log["messages_planning_fallback_used"] = messages
         if cepo_config.print_output:
-            print(f"\nCePO: No plans generated successfully. Taking the fallback.\n")
+            print("\nCePO: No plans generated successfully. Taking the fallback.\n")
 
     # Step 3 - Review and consolidate plans
     plans_message = ""
@@ -575,7 +573,7 @@ def generate_single_plan(i):
 
     cb_log["messages"] = messages
     if cepo_config.print_output:
-        print(f"\nCePO: Answer generated for one bestofn_n attempt.")
+        print("\nCePO: Answer generated for one bestofn_n attempt.")
 
     return final_output, completion_tokens, cb_log
 
@@ -692,7 +690,7 @@ def run_single_completion(i):
             cb_log[f"completion_{i}_completion_tokens"] = tokens_i
 
     if cepo_config.print_output or logger.getEffectiveLevel() == logging.DEBUG:
-        logger.debug(f"\nCePO: All Answers generated!")
+        logger.debug("\nCePO: All Answers generated!")
 
     completions = [c if isinstance(c, str) else "" for c in completions]
     return completions, completion_tokens, cb_log
@@ -882,7 +880,7 @@ def extract_answer_mathverify(response_str, last_n_chars=100):
     try:
         float(response_str)
         return [float(response_str)]
-    except:
+    except Exception:
         response_str = response_str.split("</think>", 1)[1] if "</think>" in response_str else response_str
         if last_n_chars is not None:
             response_str = response_str[-last_n_chars:]

optillm/cot_decoding.py

@@ -1,7 +1,6 @@
 import torch
 from transformers import PreTrainedModel, PreTrainedTokenizer
-from typing import List, Tuple, Dict, Optional
-import numpy as np
+from typing import List, Tuple, Dict
 
 def get_device():
     if torch.backends.mps.is_available():

optillm/cot_reflection.py

@@ -1,7 +1,6 @@
 import re
 import logging
 import optillm
-from optillm import conversation_logger
 
 logger = logging.getLogger(__name__)
 
@@ -68,7 +67,7 @@ def cot_reflection(system_prompt, initial_query, client, model: str, return_full
     thinking_match = re.search(r'<thinking>(.*?)</thinking>', full_response, re.DOTALL)
     output_match = re.search(r'<output>(.*?)(?:</output>|$)', full_response, re.DOTALL)
 
-    thinking = thinking_match.group(1).strip() if thinking_match else "No thinking process provided."
+    thinking_match.group(1).strip() if thinking_match else "No thinking process provided."
     output = output_match.group(1).strip() if output_match else full_response
 
     logger.info(f"Final output :\n{output}")

optillm/deepconf/confidence.py

@@ -10,7 +10,7 @@
 import torch
 import torch.nn.functional as F
 import numpy as np
-from typing import List, Dict, Tuple, Optional
+from typing import Dict, Optional
 import logging
 
 logger = logging.getLogger(__name__)

optillm/deepconf/processor.py

@@ -10,11 +10,9 @@
 
 import torch
 import logging
-import random
-from typing import List, Dict, Any, Optional, Tuple
+from typing import List, Dict, Any, Tuple
 from transformers import PreTrainedModel, PreTrainedTokenizer, DynamicCache
 from collections import Counter, defaultdict
-import numpy as np
 
 from .confidence import ConfidenceCalculator, ConfidenceThresholdCalibrator
 
@@ -125,7 +123,7 @@ def generate_single_trace(self, messages: List[Dict[str, str]],
                 kv_cache = outputs.past_key_values
 
             # Calculate confidence for current token
-            token_confidence = self.confidence_calculator.add_token_confidence(logits)
+            self.confidence_calculator.add_token_confidence(logits)
 
             # Check for early termination (only after minimum trace length)
             if (use_early_termination and 

optillm/entropy_decoding.py

@@ -1,7 +1,7 @@
 import torch
 import torch.nn.functional as F
 from transformers import PreTrainedModel, PreTrainedTokenizer
-from typing import List, Tuple, Dict, Optional
+from typing import List, Tuple, Dict
 import logging
 
 # Set up logging

optillm/inference.py

@@ -7,7 +7,6 @@
 from collections import OrderedDict, defaultdict
 import torch.nn.functional as F
 import torch.nn as nn
-import math
 from transformers import AutoModelForCausalLM, AutoTokenizer, PreTrainedModel
 from peft import PeftModel, PeftConfig
 import bitsandbytes as bnb
@@ -17,7 +16,6 @@
 import threading
 import traceback
 import platform
-import sys
 import re
 
 from optillm.cot_decoding import cot_decode
@@ -1069,11 +1067,9 @@ def _load_model():
                 # Check for flash attention availability
                 try:
                     import flash_attn
-                    has_flash_attn = True
                     logger.info("Flash Attention 2 is available")
                     model_kwargs["attn_implementation"] = "flash_attention_2"
                 except ImportError:
-                    has_flash_attn = False
                     logger.info("Flash Attention 2 is not installed - falling back to default attention")
 
             elif 'mps' in device:
@@ -1155,7 +1151,7 @@ def _get_adapter_name(self, adapter_id: str) -> str:
     def validate_adapter(self, adapter_id: str) -> bool:
         """Validate if adapter exists and is compatible"""
         try:
-            config = PeftConfig.from_pretrained(
+            PeftConfig.from_pretrained(
                 adapter_id,
                 trust_remote_code=True,
                 token=os.getenv("HF_TOKEN")
@@ -1591,8 +1587,8 @@ def process_batch(
 
         for i in range(0, len(formatted_prompts), self.optimal_batch_size):
             batch_prompts = formatted_prompts[i:i + self.optimal_batch_size]
-            batch_system = system_prompts[i:i + self.optimal_batch_size]
-            batch_user = user_prompts[i:i + self.optimal_batch_size]
+            system_prompts[i:i + self.optimal_batch_size]
+            user_prompts[i:i + self.optimal_batch_size]
 
             # Check cache first if enabled
             if self.model_config.enable_prompt_caching:

optillm/leap.py

@@ -3,7 +3,6 @@
 from typing import List, Tuple
 import json
 import optillm
-from optillm import conversation_logger
 
 # Setup logging
 logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')

optillm/litellm_wrapper.py

@@ -1,9 +1,8 @@
-import os
 import time
 import litellm
 from litellm import completion
 from litellm.utils import get_valid_models
-from typing import List, Dict, Any, Optional
+from typing import List, Dict, Optional
 
 # Configure litellm to drop unsupported parameters
 litellm.drop_params = True

optillm/mars/agent.py

@@ -5,7 +5,6 @@
 import logging
 from typing import Dict, Any, Tuple
 from datetime import datetime
-import random
 from .prompts import (
     MATHEMATICAL_SYSTEM_PROMPT,
     AGENT_EXPLORATION_PROMPT,