Style: Apply ruff formatting

scripts/demo_gradio.py

@@ -2,6 +2,7 @@
 Minimal Gradio demo for the LexiMind multitask model.
 Shows raw model outputs without any post-processing tricks.
 """
+
 from __future__ import annotations
 
 import json

scripts/download_data.py

@@ -1,4 +1,5 @@
 """Download datasets used by LexiMind."""
+
 from __future__ import annotations
 
 import argparse

scripts/eval_rouge.py

@@ -1,4 +1,5 @@
 """Utility script to evaluate LexiMind summaries with ROUGE."""
+
 from __future__ import annotations
 
 import argparse

scripts/evaluate.py

@@ -2,6 +2,7 @@
 Evaluate the multitask model on processed validation/test splits.
 This is used for getting definitive scores on my test set after training is complete.
 """
+
 from __future__ import annotations
 
 import argparse

scripts/export_model.py

@@ -1,4 +1,5 @@
 """Rebuild and export the trained multitask model for downstream use."""
+
 from __future__ import annotations
 
 import argparse
@@ -14,11 +15,27 @@ from src.utils.labels import load_label_metadata
 
 def parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser(description="Export LexiMind model weights")
-    parser.add_argument("--checkpoint", default="checkpoints/best.pt", help="Path to the trained checkpoint.")
-    parser.add_argument("--output", default="outputs/model.pt", help="Output path for the exported state dict.")
-    parser.add_argument("--labels", default="artifacts/labels.json", help="Label metadata JSON produced after training.")
-    parser.add_argument("--model-config", default="configs/model/base.yaml", help="Model architecture configuration.")
-    parser.add_argument("--data-config", default="configs/data/datasets.yaml", help="Data configuration (for tokenizer settings).")
+    parser.add_argument(
+        "--checkpoint", default="checkpoints/best.pt", help="Path to the trained checkpoint."
+    )
+    parser.add_argument(
+        "--output", default="outputs/model.pt", help="Output path for the exported state dict."
+    )
+    parser.add_argument(
+        "--labels",
+        default="artifacts/labels.json",
+        help="Label metadata JSON produced after training.",
+    )
+    parser.add_argument(
+        "--model-config",
+        default="configs/model/base.yaml",
+        help="Model architecture configuration.",
+    )
+    parser.add_argument(
+        "--data-config",
+        default="configs/data/datasets.yaml",
+        help="Data configuration (for tokenizer settings).",
+    )
     return parser.parse_args()
 
 

scripts/inference.py

@@ -1,4 +1,5 @@
 """Run inference with the multitask model."""
+
 from __future__ import annotations
 
 import argparse

scripts/preprocess_data.py

@@ -1,4 +1,5 @@
 """Preprocess raw datasets into JSONL splits for LexiMind training."""
+
 from __future__ import annotations
 
 import argparse
@@ -139,9 +140,10 @@ def preprocess_summarization(raw_dir: Path, processed_dir: Path) -> None:
         output_path = processed_dir / f"{split}.jsonl"
         output_path.parent.mkdir(parents=True, exist_ok=True)
         print(f"Writing summarization split '{split}' to {output_path}")
-        with source_path.open("r", encoding="utf-8", newline="") as source_handle, output_path.open(
-            "w", encoding="utf-8"
-        ) as sink:
+        with (
+            source_path.open("r", encoding="utf-8", newline="") as source_handle,
+            output_path.open("w", encoding="utf-8") as sink,
+        ):
             reader = csv.DictReader(source_handle)
             for row in reader:
                 article = row.get("article") or row.get("Article") or ""

scripts/train.py

@@ -1,4 +1,5 @@
 """End-to-end training entrypoint for the LexiMind multitask model."""
+
 from __future__ import annotations
 
 import json

src/api/app.py

@@ -1,4 +1,5 @@
 """FastAPI application entrypoint."""
+
 from fastapi import FastAPI
 
 from .routes import router

src/api/dependencies.py

@@ -1,4 +1,5 @@
 """Dependency providers for the FastAPI application."""
+
 from __future__ import annotations
 
 from functools import lru_cache

src/api/routes.py

@@ -1,4 +1,5 @@
 """API routes."""
+
 from typing import cast
 
 from fastapi import APIRouter, Depends, HTTPException, status

src/api/schemas.py

@@ -1,4 +1,5 @@
 """API schemas."""
+
 from pydantic import BaseModel
 
 

src/data/dataloader.py

@@ -1,4 +1,5 @@
 """Task-aware DataLoader builders for the LexiMind multitask suite."""
+
 from __future__ import annotations
 
 from typing import List

src/data/dataset.py

@@ -1,4 +1,5 @@
 """Dataset definitions for the LexiMind multitask training pipeline."""
+
 from __future__ import annotations
 
 import json
@@ -179,7 +180,9 @@ def _load_jsonl_generic(
         if first_non_ws == "[":
             payloads = _safe_json_load(handle, data_path)
             if not isinstance(payloads, list):
-                raise ValueError(f"Expected a JSON array in '{data_path}' but found {type(payloads).__name__}")
+                raise ValueError(
+                    f"Expected a JSON array in '{data_path}' but found {type(payloads).__name__}"
+                )
             for idx, payload in enumerate(payloads):
                 if not isinstance(payload, dict):
                     raise ValueError(

src/data/preprocessing.py

@@ -1,4 +1,5 @@
 """Text preprocessing utilities built around Hugging Face tokenizers."""
+
 from __future__ import annotations
 
 from dataclasses import dataclass, replace

src/data/tokenization.py

@@ -1,4 +1,5 @@
 """Tokenizer wrapper around HuggingFace models used across LexiMind."""
+
 from __future__ import annotations
 
 from dataclasses import dataclass
@@ -23,13 +24,19 @@ class Tokenizer:
     def __init__(self, config: TokenizerConfig | None = None) -> None:
         cfg = config or TokenizerConfig()
         self.config = cfg
-        self._tokenizer: PreTrainedTokenizerBase = AutoTokenizer.from_pretrained(cfg.pretrained_model_name)
+        self._tokenizer: PreTrainedTokenizerBase = AutoTokenizer.from_pretrained(
+            cfg.pretrained_model_name
+        )
         self._pad_token_id = self._resolve_id(self._tokenizer.pad_token_id)
         self._bos_token_id = self._resolve_id(
-            self._tokenizer.bos_token_id if self._tokenizer.bos_token_id is not None else self._tokenizer.cls_token_id
+            self._tokenizer.bos_token_id
+            if self._tokenizer.bos_token_id is not None
+            else self._tokenizer.cls_token_id
         )
         self._eos_token_id = self._resolve_id(
-            self._tokenizer.eos_token_id if self._tokenizer.eos_token_id is not None else self._tokenizer.sep_token_id
+            self._tokenizer.eos_token_id
+            if self._tokenizer.eos_token_id is not None
+            else self._tokenizer.sep_token_id
         )
 
     @property
@@ -84,7 +91,9 @@ class Tokenizer:
         )
         return cast(List[List[int]], encoded["input_ids"])
 
-    def batch_encode(self, texts: Sequence[str], *, max_length: int | None = None) -> dict[str, torch.Tensor]:
+    def batch_encode(
+        self, texts: Sequence[str], *, max_length: int | None = None
+    ) -> dict[str, torch.Tensor]:
         normalized = [text.lower() if self.config.lower else text for text in texts]
         encoded = self._tokenizer(
             normalized,

src/inference/__init__.py

@@ -4,9 +4,9 @@ from .factory import create_inference_pipeline
 from .pipeline import EmotionPrediction, InferenceConfig, InferencePipeline, TopicPrediction
 
 __all__ = [
-	"InferencePipeline",
-	"InferenceConfig",
-	"EmotionPrediction",
-	"TopicPrediction",
-	"create_inference_pipeline",
+    "InferencePipeline",
+    "InferenceConfig",
+    "EmotionPrediction",
+    "TopicPrediction",
+    "create_inference_pipeline",
 ]

src/inference/factory.py

@@ -1,4 +1,5 @@
 """Helpers to assemble an inference pipeline from saved artifacts."""
+
 from __future__ import annotations
 
 from pathlib import Path

src/inference/pipeline.py

@@ -1,4 +1,5 @@
 """Inference helpers for multitask LexiMind models."""
+
 from __future__ import annotations
 
 from dataclasses import dataclass, fields, replace

src/inference/postprocessing.py

@@ -1,4 +1,5 @@
 """Output cleaning helpers."""
+
 from typing import List
 
 

src/models/decoder.py

@@ -12,6 +12,7 @@ Conventions:
 - This decoder uses Pre-LN (RMSNorm before each sublayer).
 - RMSNorm is just simpler than LayerNorm and more computationally efficient, it's become the modern convention. These reasons are why I used it here.
 """
+
 import math
 from typing import Dict, List, Optional, Tuple, Union
 

src/models/factory.py

@@ -1,4 +1,5 @@
 """Factory helpers to assemble multitask models for inference/training."""
+
 from __future__ import annotations
 
 from dataclasses import dataclass

src/models/heads.py

@@ -9,6 +9,7 @@ Includes:
 
 Keep these heads minimal, well-tested, and easy to compose on top of encoder/decoder outputs.
 """
+
 from typing import Literal, Optional
 
 import torch

src/models/multitask.py

@@ -14,6 +14,7 @@ Design goals:
   seq2seq tasks (encoder -> decoder -> LMHead)
 - Minimal dependencies on training loop; return logits and (optionally) loss
 """
+
 from typing import Any, Dict, Optional
 
 import torch

src/training/metrics.py

@@ -1,4 +1,5 @@
 """Metric helpers used during training and evaluation."""
+
 from __future__ import annotations
 
 from typing import Any, Dict, List, Sequence

src/training/trainer.py

@@ -1,4 +1,5 @@
 """Multi-task trainer coordinating summarization, emotion, and topic heads."""
+
 from __future__ import annotations
 
 import shutil
@@ -330,9 +331,9 @@ class Trainer:
         """Generate and print sample summaries to monitor quality during training."""
         self.model.eval()
         samples_generated = 0
-        print(f"\n{'='*80}")
+        print(f"\n{'=' * 80}")
         print(f"[Validation Generation - Epoch {epoch}]")
-        print(f"{'='*80}")
+        print(f"{'=' * 80}")
 
         with torch.no_grad():
             for batch in val_loader:
@@ -400,7 +401,7 @@ class Trainer:
 
                 samples_generated += 1
 
-        print(f"{'='*80}\n")
+        print(f"{'=' * 80}\n")
         self.model.train()
 
     def _print_epoch_progress(

src/utils/config.py

@@ -1,4 +1,5 @@
 """YAML config loader."""
+
 from dataclasses import dataclass
 from pathlib import Path
 from typing import Any, Dict

src/utils/io.py

@@ -1,4 +1,5 @@
 """Checkpoint IO helpers."""
+
 from pathlib import Path
 
 import torch
@@ -12,4 +13,4 @@ def save_state(model: torch.nn.Module, path: str) -> None:
 
 def load_state(model: torch.nn.Module, path: str) -> None:
     state = torch.load(path, map_location="cpu", weights_only=True)
-    model.load_state_dict(state)
+    model.load_state_dict(state)

src/utils/labels.py

@@ -1,4 +1,5 @@
 """Label metadata helpers for multitask inference."""
+
 from __future__ import annotations
 
 import json

src/utils/logging.py

@@ -1,4 +1,5 @@
 """Logging setup."""
+
 import logging
 
 

src/utils/random.py

@@ -1,4 +1,5 @@
 """Randomness helpers."""
+
 import random
 
 import numpy as np

src/visualization/attention.py

@@ -1,4 +1,5 @@
 """Attention plotting utilities."""
+
 from typing import Sequence
 
 import matplotlib.pyplot as plt

src/visualization/metrics.py

@@ -1,4 +1,5 @@
 """Metric plotting helpers."""
+
 import matplotlib.pyplot as plt
 
 

tests/test_api/test_routes.py

@@ -1,4 +1,5 @@
 """API integration tests for the inference endpoint."""
+
 from __future__ import annotations
 
 from fastapi.testclient import TestClient
@@ -31,4 +32,4 @@ def test_summarize_route_returns_pipeline_outputs() -> None:
         assert payload["topic"] == "news"
         assert payload["topic_confidence"] == 0.8
     finally:
-        app.dependency_overrides.clear()
+        app.dependency_overrides.clear()

tests/test_data/test_download_records.py

@@ -1,4 +1,5 @@
 """Unit tests for dataset record helpers in scripts.download_data."""
+
 from __future__ import annotations
 
 import importlib.util
@@ -26,11 +27,13 @@ class DummyDataset:
 
 class DownloadDataRecordTests(unittest.TestCase):
     def test_emotion_records_handles_out_of_range_labels(self) -> None:
-        dataset_split = DummyDataset([
-            {"text": "sample", "label": 1},
-            {"text": "multi", "label": [0, 5]},
-            {"text": "string", "label": "2"},
-        ])
+        dataset_split = DummyDataset(
+            [
+                {"text": "sample", "label": 1},
+                {"text": "multi", "label": [0, 5]},
+                {"text": "string", "label": "2"},
+            ]
+        )
         label_names = ["sadness", "joy", "love"]
         records = list(
             download_data._emotion_records(
@@ -45,12 +48,14 @@ class DownloadDataRecordTests(unittest.TestCase):
         self.assertEqual(records[2]["emotions"], ["2"])
 
     def test_topic_records_handles_varied_label_inputs(self) -> None:
-        dataset_split = DummyDataset([
-            {"text": "news", "label": 3},
-            {"text": "list", "label": [1]},
-            {"text": "unknown", "label": "5"},
-            {"text": "missing", "label": []},
-        ])
+        dataset_split = DummyDataset(
+            [
+                {"text": "news", "label": 3},
+                {"text": "list", "label": [1]},
+                {"text": "unknown", "label": "5"},
+                {"text": "missing", "label": []},
+            ]
+        )
         label_names = ["World", "Sports", "Business", "Sci/Tech"]
         records = list(
             download_data._topic_records(

tests/test_inference/test_pipeline.py

@@ -1,4 +1,5 @@
 """Integration tests for the inference pipeline."""
+
 from __future__ import annotations
 
 from pathlib import Path

tests/test_models/test_positional_encoding.py

@@ -4,7 +4,6 @@
 Tests for positional encoding.
 """
 
-
 import matplotlib
 import torch