Keep Hub SDK compatible with Trackio 0.33

Accept authorized HF OAuth tokens for live metrics

Rename dashboard display title to SmolDataEnv RL

Open a focused RL overview and add navigation for models and diagnostics

Separate SFT dashboard link and bound RL plotting payloads without deleting records

Add six SFT runs: audited runtimes, epoch metrics and article citation exports

Preserve sparse checkpoint metrics when reading comparison charts

Show complete LFM checkpoint progression, coverage, and tool/token diagnostics

Default to unified Data Agent project with readable runs and clear evaluation status

Fix dashboard links after FineEnvs rename and link continuation evaluations

Document provisional hard-task continuation checkpoint evaluations

Add latest provisional continuation evaluation scores

Complete all three training runs with audited checkpoint evaluations through 1000

Complete Harbor multi-harness 900/1000 with audited verifier-budget outcomes

Keep final-checkpoint annotation inside comparison plot

Refresh three-run comparison with Harbor OpenCode-only baseline and checkpoints 100/200

Publish verified checkpoint and difficulty tables with comparison view links

Add unified Harbor and OpenCode comparison dashboard and report

Upload app.py with huggingface_hub

Upload requirements.txt with huggingface_hub

Upload README.md with huggingface_hub

initial commit