lllllllama

Generates candidate deep learning research approaches with auditable code understanding and fair experimental comparison.

Reproduce deep learning research with documented rigor by reading repositories first and capturing evidence of setup, execution, and deviations.

Inspect deep learning repositories to understand model structure, training setup, and implementation patterns without modifying code.

Audits and improves deep learning implementations by exploring isolated branches with tracked modifications and rollback capability.

Prepares conda environments, dataset paths, and cache locations for reproducing deep learning repositories from README specifications.

Execute bounded exploratory runs for deep learning research with quick validation and parameter sweeps.

Capture and normalize outputs from deep learning smoke tests or inference commands, writing standardized reproducibility files with change tracking.

Fill reproduction gaps by extracting specific paper details like dataset splits, preprocessing, and evaluation protocols from primary sources.

Scans repositories and READMEs to extract commands and generate minimal reproduction plans for deep learning projects.

Executes deep learning training commands with standardized output logging, checkpoints, and metrics for research verification.

Diagnoses deep learning errors and training failures with rigorous debugging before suggesting any fixes.