Update README.md

fix: include full dream text in prompt + robust JSON parsing

feat: JSON output for reliable dream content parsing

fix: structured prompt for parser-compatible output

fix: move @spaces.GPU to app.py for ZeroGPU detection

fix: correct ZeroGPU setup — import spaces directly + local fallback

fix: top-level @spaces.GPU for ZeroGPU detection

fix: remove bucket dependency for ZeroGPU compatibility

fix: ZeroGPU @spaces.GPU support for model generation

feat: ZeroGPU (RTX Pro 6000) support with 8B model

fix: 1024 token budget for thinking model + robust fallback

fix: robust thinking tag handling for MiniCPM5-1B

fix: strip thinking tags from combined generation output

fix: robust thinking tag handling for MiniCPM5-1B

perf: single LLM call for dream pipeline (4x faster on CPU)

fix: remove token_type_ids for MiniCPM5-1B compatibility

fix: handle MiniCPM5-1B thinking tags, extract content from <think>

fix: model generation reliability + performance

fix: refresh Explorer/Journal dropdowns after dream submission

feat: wire up Explorer, Journal, Player, History, Map refresh handlers

feat: HF Buckets persistent storage at /data

fix: update slot button labels after creation, add HF Buckets persistence

fix: per-user session isolation via ensure_session_id

fix: remove device_map='cpu' - not supported in HF Spaces

fix: add show_error=True, error handling in dream submission

fix: handle None session_id gracefully

fix: session_id as gr.State input, update README with correct model

Update README.md

fix: use transformers+MiniCPM5-1B, fix session_id State capture

fix: use ctransformers (pre-built wheel) instead of llama-cpp-python

fix: add llama-cpp-python to requirements (was missing)

feat: per-user data isolation, fix model name, clean gitignore

fix: default port 7860 for HF Spaces

feat: complete single-page layout with all tabs, 74 components

fix: use cpu-basic hardware for HF Spaces

fix: use GRADIO_SERVER_PORT env var for HF Spaces compatibility

chore: force rebuild

fix: switch to Gradio SDK, remove Docker, clean requirements

chore: trigger Docker rebuild

ignore screenshots

remove binary icon, use online favicon

fix: CSS via gr.HTML style tag, background as div, no CSS variables

refactor: follow lolaby pattern — embedded CSS, ATTRIBUTION_HTML footer

feat: two-screen layout with gr.render, background, favicon, editor fix

fix: remove all gr.Column/gr.Group outputs, add favicon

fix: single-page layout, no gr.Column/Group output components

fix: move name_input_box inside slot_screen column

fix: reliable slot selection with single HTML panel + Gradio buttons

fix: slot selection logic + input validation

feat: restore slot selection, Settings load save, player-only display