Fix inference tab generate input handling for HF runtime

Add final submission demo and publish artifacts

ci: harden hf-sync workflow and add secret verification

docs: finalize premium README with detailed repo structure and image flowchart

docs: restore all detailed sections and merge with new refinements for max-detail README

docs: restore all details and add Mermaid diagrams for complete README

docs: finalize world-class README with Mermaid diagrams and academic refinement

docs: enrich README with live comparison, impact metrics, and deployment robustness

fix: comparison tab (remove 4-bit quant, handle missing adapter, VRAM guard)

Fix: Remove unsupported 'type' parameter from gr.Chatbot

Phase 3: Add 4 innovation tabs + RAG comparison section

docs: restore architecture image and remove problematic Mermaid block

fix: enforce blank lines and clean syntax for Mermaid rendering

fix: simplify Mermaid syntax in README to ensure rendering across platforms

docs: comprehensive overhaul of README with in-depth architecture, data sync, and research details

Merge branch 'main' of https://huggingface.co/spaces/Chirag0123/shiftlog-gym

fix: add vibe_coding_ratio property to EpisodeMetrics to resolve AttributeError

Sync training artifact: observatory/baselines.json

Sync training artifact: plots/03_mttr_comparison.png

Sync training artifact: plots/02_recall_bonus_curve.png

Sync training artifact: plots/01_reward_curve.png

fix: resolve TypeError in step_callback by properly unpacking payload dictionaries

fix: standardize Mermaid to dark-only nodes with white text for maximum readability

fix: improve Mermaid diagram contrast for better readability on light backgrounds

fix: resolve persistent Mermaid syntax error by using ultra-compatible flat diagram syntax

fix: resolve Mermaid syntax error by simplifying labels and removing emojis from diagram DSL

fix: resolve Mermaid rendering issue by switching to built-in gr.Markdown support

feat: upgrade Causal Flow diagram to multi-shift handoff story

style: rename performance snapshot labels to Optimization Loss, Accuracy, and Convergence

fix: resolve KeyError 'vibe_ratio' by properly propagating it from pipeline to daemon

fix: resolve NameError by fixing definition order of _FAMILY_PILLS

fix: resolve IndentationError and misplaced Research Story tab

fix: resolve IndentationError in read_shift_log

fix: populate results with Stage C benchmarks and fix Plotly/Image refreshing

feat: grand finale UI overhaul — glassmorphism, causal diagrams, and vibe ratio tracking

feat: implement 'Resume from Hub' and 'Vibe Coding Ratio' tracking

feat: premium dashboard revamp with glassmorphism + GRPO stability fixes

fix: upgrade trl and transformers to support GRPOConfig

fix: handle full HF URLs in the Repository ID input

fix: unlock training metrics and URLs for manual editing as requested

feat: live metrics integration — pipeline now reports progress directly to UI state

fix: make training start button always interactive and add page-load sync

fix: syntax error in space_training_daemon.py (removed Spanish from slice)

fix: relax python version to >=3.10 to match container environment

fix: use native python3.10 and add build-essential to fix HF build error

Add GPU training daemon + Training tab in Observatory dashboard

feat: add visibility to evaluation phase with progress prints

feat: upgrade training log to a scrollable terminal UI with unbuffered output

feat: switch dashboard to live stream full training logs for better visibility

feat: add WandB key input and log tailing to Training Lab dashboard