style: replace em dashes with hyphens for natural-sounding text 0a6da12 Mohammed AL Sarraj commited on Apr 12
fix: add comprehensive mobile responsive styles β sidebar collapse, grid stacking, text sizing a636387 Mohammed AL Sarraj commited on Apr 12
feat: 50 test cases, deterministic metrics (BLEU/word overlap/exact match), public leaderboard tab, HuggingFace dataset published d2a3cd9 Mohammed AL Sarraj commited on Apr 12
fix: expose refreshDashboard to window so run selector dropdown works 4fbac1b Mohammed AL Sarraj commited on Apr 12
feat: add run selector dropdown to Dashboard β filter scores by specific evaluation run b3ebbeb Mohammed AL Sarraj commited on Apr 12
fix: remove paid providers (DeepSeek/Together), fix Cohere model name e407c2a Mohammed AL Sarraj commited on Apr 12
fix: correct Cohere model name to command-r-08-2024, use Arabic-specialized model for Arabic tasks 307d354 Mohammed AL Sarraj commited on Apr 12
fix: Cohere V2 API handler, handle 402 errors, fix model names c2dc1f6 Mohammed AL Sarraj commited on Apr 12
feat: add Dashboard tab to Arabic Bench β aggregate scores, category performance, score distribution chart 8b49d50 Mohammed AL Sarraj commited on Apr 12
feat: add DeepSeek, Gemini, Together AI, Cohere providers β 8 AI models now available for Arabic benchmarking 15d9104 Mohammed AL Sarraj commited on Apr 12
feat: add multi-model Arabic benchmark β compare AI providers side-by-side with leaderboard 6fb301e Mohammed AL Sarraj commited on Apr 12
feat: smart task-based model routing β Arabic tasks use 70B+ models, code tasks use code-optimized models 784a6a1 Mohammed AL Sarraj commited on Apr 12
perf: slim Arabic Bench prompt and reduce max_tokens for faster evaluation 5cdfc81 Mohammed AL Sarraj commited on Apr 12
fix: add ai_input to all 23 test cases so auto-run works in every category 47ca9fe Mohammed AL Sarraj commited on Apr 12
feat: major Arabic Bench upgrade β 23 test cases, 8 categories, expanded metrics, history, dataset browser 3998724 Mohammed AL Sarraj commited on Apr 12
fix: use relative API URLs so fetch works under blueprint prefix b4984b6 Mohammed AL Sarraj commited on Apr 12
fix: back button inline (remove position:fixed), body flex-col height 78b1d82 Mohammed AL Sarraj commited on Apr 11
fix: viewport height layout, inline back button, meeting distiller sidebar 1a8275d Mohammed AL Sarraj commited on Apr 11
restore light theme tools + light landing pages + height fix c44a0d9 Mohammed AL Sarraj commited on Apr 11