5 Commits

Author	SHA1	Message	Date
Matt Van Horn	c31c50be51	fix(model): forward Qwen timeout and only set enable_thinking when true ... Two bugs made local vLLM targets score acc=0.000: the router did not forward 'timeout' to the Qwen backend (so runs used the 300s default), and qwen_backend always injected chat_template_kwargs.enable_thinking, which non-Qwen vLLM servers reject or answer with <think> output and no <answer> tag. Forward timeout and only set the field when enabled. Closes #28 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>	2026-06-07 07:41:35 -07:00
kaikai-macbook	41012e2d5e	Support Qwen chat as optimizer backend	2026-06-01 16:44:49 +08:00
Declan Murphy	309ea64ff4	feat: integrate MiniMax into model router, backend config, and common ... common.py: - Add minimax_chat → MiniMax/MiniMax-Text-01 to _BACKEND_DEFAULT_MODELS - Add minimax/minimax_chat aliases to _BACKEND_ALIASES backend_config.py: - Add minimax_chat to set_optimizer_backend() valid set - Add minimax_chat to set_target_backend() valid set - Add minimax_chat to is_optimizer_chat_backend() - Add minimax_chat to is_target_chat_backend() __init__.py: - Import minimax_backend as _minimax - Add minimax_chat to set_backend() legacy handler - Add minimax_chat to get_backend_name() reporting - Route chat_target() and chat_target_messages() to _minimax - Update NotImplementedError messages to list minimax_chat - Aggregate _minimax into get_token_summary() - Add _minimax.reset_token_tracker() - Add configure_minimax_chat() delegator - Add _minimax to set_reasoning_effort() and set_target_deployment()	2026-05-31 05:22:33 +08:00
Cuzyoung	4a1b984d87	refactor: rename teacher/student to optimizer/target, remove best skills, fix slow update ... - Rename teacher -> optimizer, student -> target across all code, configs, docs, prompts - CLI: --teacher_model -> --optimizer_model, --student_model -> --target_model - Remove best_skill files, keep only initial skills - Fix slow update gate (force write into skill) - Fix SLOW_UPDATE marker stripping - Remove deep_reflect and meta_reflect mechanisms - Update .env.example with export prefix and azure_cli docs - Add endpoint empty validation in azure_openai.py Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>	2026-05-24 19:15:10 +00:00
CharlesYang030	244e346b83	SkillOpt v0.1.0: initial release ... - Skill optimization framework with training loop analogy - 11 benchmarks, 4 model backends (Azure OpenAI, Claude, Codex, Qwen) - WebUI for browser-based training control - Pluggable architecture for extending benchmarks and backends	2026-05-21 17:22:04 +00:00