docs: add measured Kaggle beam1 speed

docs: refresh Kaggle T4 x2 speed numbers

docs: update Colab T4 speed profile

bench: add cloud sweep runner

bench: add fair cloud profiling controls

feat(chunk): paragraph mode giu xuong dong - app.py

feat(chunk): paragraph mode giu xuong dong - translator.py

feat(chunk): paragraph mode giu xuong dong - line_restore.py

Add benchmark phase profile and paragraph chunk fix

fix(chunk): paragraph mode gom dong - app.py

fix(chunk): paragraph mode gom dong - token_chunker.py

Add Hirashiba CT2 model options

badge Open-in-Kaggle + tốc độ T4x2

fix link Kaggle = mở Kaggle

multi-GPU robustness: gpu_setup.py

multi-GPU robustness: hardware.py

multi-GPU robustness: translator.py

multi-GPU robustness: app.py

gpu blocker fix: hardware.py

gpu blocker fix: translator.py

multi-GPU inter_threads=1

multi-GPU: hardware.py

multi-GPU: translator.py

sync V9 teacher-student con variant from qt2 b15068a

sync V9 embedded classical and relation roles from qt2 467c02a

docs: menu Colab tiếng Việt + cảnh báo TPU

h30: cap input 160 (điểm ngọt)

h30: cap input 256 chống vỡ context

sync V9 teacher-student dialogue role fix from qt2 3597738

ui: ẩn nút reset slider Beam

ui: ẩn ô số slider Beam

ui: rút gọn note Beam 1 dòng

ui: note giải thích Beam (info slider)

ct2: anti-repetition (no_repeat_ngram + rep_penalty)

sync V9 teacher/mother object guards from qt2 5585b77

cap input 150k trên Space + sửa comment startup

sync V9 no-speech teacher-object guard from qt2 e8eee55

sync V9 no-speech teacher-object guard from qt2 e8eee55

sync V9 teacher-object guard from qt2 21ab81a

sync V9 teacher-object guard from qt2 21ab81a

sync V9 teacher-object guard from qt2 21ab81a

ui: badge thử nghiệm xưng hô + ẩn start.bat/127.0.0.1 trên Space

sync pronoun_harmonizer_v9.py (Codex harden honorific+V9 guards)

sync honorific_normalize.py (Codex harden honorific+V9 guards)

Thêm tốc độ T4 thật (~27k chữ/s, 2.4M chữ ~1ph30) vào accordion

Warmup trong load() (mọi môi trường) — lần dịch đầu nhanh + đo sạch

Warmup trong load() (mọi môi trường) — lần dịch đầu nhanh + đo sạch

Fix GPU batch không bị CPU bóp (T4: 16→96, benchmark đo)

Thêm nút Open in Colab (repo github public) vào banner + accordion

Space fast mode (Codex): beam=1 default + batch examples/window1 + bỏ auto-export file