vlsi-autograder/antigravity-lms-grader

Support dynamic status updates (Queued, Processing, Reviewed) in callback system

Convert autograder endpoint to a background task queue using FastAPI BackgroundTasks

Increase retries to 10 and limit feedback length to 100 words in human tone

Add automatic exponential backoff retry logic for LLM calls

Optimize feedback format sanitization parser

Remove provider restriction from InferenceClient for serverless routing

Pass assignment title and late submission context to Qwen evaluator

Update Qwen system prompt to gracefully handle snippet evaluations

Update grading evaluation and scoring bounds to 100

feat: add support for GITHUB_TOKEN to clone private student repositories

feat: evaluate verilog code against assignment question details

feat: add support for student notes and restrict AI feedback length

Deploy autograder with Verilator and Qwen Coder

initial commit