Pipeline Scope

This app runs the full training pipeline for the maths-conjuncture-solutions project.

An autonomous training workflow for DeepSeek-Math that runs multi-stage curriculum fine-tuning on Space GPU, executes post-training quality evaluation, and publishes only qualified adapters, checkpoints, and run reports to your Hugging Face model repository.

1. Pull released parquet splits from NorthernTribe-Research/math-conjecture-training-corpus.
2. Build runtime training configuration from configs/deepseek_math_sota.yaml.
3. Execute multi-stage DeepSeek-Math curriculum fine-tuning via scripts/train_sota.py.
4. Run post-training evaluation with pass@k-style sampling and family-level metrics.
5. Enforce autonomous quality gates before adapter promotion/push.
6. Stream live terminal telemetry and structured run summaries.

Autonomous Mode is enabled by default and applies full-stage execution parameters automatically. Continuous Auto-Restart is enabled by default so the next training cycle starts automatically.