Add Training Evidence: real plots, logs, updated README

Grounded description text

Upgrade landing page: inference-locked production model

Fix num_generations=2

SFT 200->350 steps for better format compliance

Fix curriculum window=50 and threshold=0.80

GRPO speed fix: 50 steps, 1 generation

Add training logs and fix gitignore

Add training logs and fix gitignore

Add training logs and fix gitignore

Remove emojis from notebook

Add WHY comments + GRPO/SFT speed caps

Speed fix: cap SFT at 200 steps

Fix URL typos

Fix URL typos

Fix HF Space URL typo in README and BLOG

Update plots\training_dashboard.png

Update plots\reward_curve.png

Update plots\baseline_comparison.png

Update .gitattributes

Update submit_job.py

Update hf_job_train.py

Update eval_data_centric.py

Update train_colab.ipynb

Update train_data_centric.py

Update BLOG.md

Update README.md

Audit fixes: remove duplicate torch import, add metadata field, fix stale strings, fix test assertions, update reward docs

Clean final training notebook: no demo, no Drive, all steps including plots + eval + download

Redesign reward for discrimination: efficiency multiplier, strict penalties, stretch bonus, start at level 1

Fix demo mode crash: use max_steps param instead of unpicklable local class

Optimize for fast iteration: 1.5B model, LoRA r=8, GRPO batch=2/gen=2, seq=512

Switch experiment tracking from W&B to TensorBoard (no API key required)

Enable W&B experiment tracking in SFT+GRPO phases (required by hackathon)

Pin torchao version to 0.6.1 for stability

refactor: extract agent_utils.py (shared prompt/commands/server utils), simplify reward to env+format, add audit.py

feat: rewrite hf_job_train.py + add submit_job.py for HF Jobs training

Data-Centric AI RL Environment — OpenEnv Hackathon Submission