fix: import Tuple, add loopback URL on Windows, ignore heavy model files

fix: add zstd system dependency to Dockerfile for Ollama installation

docs: add Hugging Face Space metadata to README.md

feat: VidyaBot v1.0 Production with offline Ollama fallback

fix: Remove manual label assignment to let DataCollatorForLanguageModeling handle padding and labels dynamically

fix: Update Trainer instantiation to dynamically support transformers v5 processing_class

fix: Align download destination path to mistral-vidyabot-hf in modal_finetune.py

fix: Update modal_finetune.py to use Image.add_local_file for Modal v1.0.0 compatibility

feat: Add expanded synthetic student Q&A dataset of 103 pairs for fine-tuning

feat: Add Modal fine-tuning pipeline for Well-Tuned badge

Delete vidyabot_master_prompt.md

Clean up repository and update codebase for vidyabot-build-small

first commit

Update README.md

VidyaBot v2 Elite: 88% cost reduction, graceful degradation, production-ready