Nirav-Madhani/Meditation

Update README.md

Update project config

Add checkpoint monitor script and update training docs

Tier-dependent reward design + paper/memory updates

Add training optimizations: n-gram penalty, DAPO loss, Flash Attn, resume

Update paper with live GRPO training metrics and corrected hyperparameters

Add Flash Attention 2, increase GRPO K to 8 for better training

Switch to Gemini APIs, LFM2.5-1.2B, Lightning AI cloud training

Remove GPU grant reference from status box

Address feedback: replace placeholders with real data, add benchmarks

Initial commit