chore: update Colab training notebook with full step-by-step cells

fix: lazy-import openai to avoid ModuleNotFoundError in HF Jobs

feat: populate episode log with last task on run all

fix: add loading state to runAllTasks

feat: visually stunning dataset evaluation UI

force rebuild

fix: resolve dataset evaluation error + add dataset builder with type specification

Final polish: judging-ready training notebook and dashboard fix

Fix dataset evaluation signature mismatch

Switch to Llama 3.1 base model for HF API compatibility

Revert to API inference due to lack of local CUDA GPU on host

Switch inference to local Unsloth to bypass HF Serverless API limits on custom adapters

Fix fallback node extraction logic

Add live HTTP inference script

Restore binary files using Git LFS

Final optimizations, RL endpoint, dataset upload UI, and Hackathon artifacts

Final hackathon submission - RecallTrace self-play

Allow empty POST body for reset endpoint

Switch reward bars to line chart and tighten episode layout

Improve dashboard visuals and episode output

Sync final Hugging Face Space changes back to GitHub

Deploy RecallTrace OpenEnv to Hugging Face Space

Sync local project updates to sham branch

Removed __pycache__ and added gitignore rules

Added env folder and core implementation

Merge branch 'sham'

Updated inference and scenario files

Updated inference and scenario files

Merge pull request #1 from MS-Shamanth/sham

Phase 1: env core setup and basic structure

Fix header formatting in README.md

Enhance README with project details and setup instructions

Initial project structure setup

Initial commit