"""Console view: prints training progress and results.""" from __future__ import annotations from typing import TYPE_CHECKING if TYPE_CHECKING: from src.models.config import GrassmannConfig from src.services.training_service import TrainMetrics class ConsoleView: """Simple console-based display for training progress.""" def show_model_info(self, config: GrassmannConfig, param_count: int) -> None: mode = "Seq2Seq (Encoder-Decoder)" if config.use_seq2seq else "CTC (Encoder-only)" print("\n" + "=" * 60) print(f" Grassmann Transcription Model — {mode}") print("=" * 60) print(f" Encoder layers: {config.n_layers}") if config.use_seq2seq: print(f" Decoder layers: {config.n_decoder_layers}") print(f" Cross window: {config.cross_window}") print(f" d_model: {config.d_model}") print(f" r_reduced: {config.r_reduced}") print(f" Plücker dim: {config.plucker_dim}") print(f" Window offsets: {config.window_offsets}") print(f" Vocab size: {config.vocab_size}") print(f" Parameters: {param_count:,}") print(f" Attention: NONE (fully Grassmann)") print("=" * 60 + "\n") def on_epoch_end(self, metrics: TrainMetrics) -> None: val_str = f" val_loss: {metrics.val_loss:.4f}" if metrics.val_loss is not None else "" print(f" Epoch {metrics.epoch:3d} | train_loss: {metrics.train_loss:.4f}{val_str}") def show_training_complete(self, history: list[TrainMetrics]) -> None: if history: best = min(history, key=lambda m: m.val_loss if m.val_loss is not None else m.train_loss) print(f"\n Best epoch: {best.epoch} | loss: {best.val_loss or best.train_loss:.4f}") print(" Training complete.\n") def show_test_results(self, results: dict) -> None: print("\n Test Results:") for k, v in results.items(): print(f" {k}: {v:.4f}") print() def show_message(self, msg: str) -> None: print(f" {msg}")