agentinputlab 0.1.0a2__tar.gz → 0.1.0a4__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/PKG-INFO +22 -6
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/README.md +21 -5
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/api.md +6 -2
- agentinputlab-0.1.0a4/docs/benchmark-setup.md +76 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/getting-started.md +24 -3
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/pyproject.toml +1 -1
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/__init__.py +1 -1
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/api.py +50 -4
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/artifacts/store.py +8 -1
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/llm/openai_compatible.py +8 -6
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/__init__.py +3 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/cache.py +9 -2
- agentinputlab-0.1.0a4/src/agentinputlab/benchmarks/discovery.py +87 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/registry.py +4 -1
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/cli/main.py +55 -1
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/registry/runtime.py +12 -2
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab.egg-info/PKG-INFO +22 -6
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab.egg-info/SOURCES.txt +2 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/.release/PYPI_PUBLISHING_DISABLED +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/CITATION.cff +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/LICENSE +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/MANIFEST.in +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/NOTICE +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/PUBLISHING.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/THIRD_PARTY_NOTICES.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/ace-bfcl.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/ace-native-statebench.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/ace-native-tau.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/ace-native-toolsandbox.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/ace-online.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/acon-live-tau.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/acon-live-tooltalk.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/adaptive-rag-fit.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/decomp-planner.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/hyde-external.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/infocqr-student.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/ircot-native-qa.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/ircot-prediction-replay.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/llmlingua2-multilingual.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/llmlingua2-training.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/longllmlingua-controlled.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/plan-and-budget-full.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/plan-and-budget-hard-cap.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/qrag-training.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/r2ag-training.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/real-execution.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/recomp-abstractive-training.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/recomp-extractive-training.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/recomp-selective.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/reformer-induction.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/releases/0.1.0a2.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/retriever_profiles.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/rrr-controlled.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/smartsearch-training.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/toolkit-acceptance.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/udcs-deterministic.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/docs/validation.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/scripts/selectors/_common.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/scripts/selectors/build_policy_manifest.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/scripts/selectors/calibrate_selector.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/scripts/selectors/export_discovery.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/scripts/selectors/freeze_selector.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/scripts/selectors/train_selector.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/setup.cfg +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/__main__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/adaptive_rag/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/adaptive_rag/replay.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/adaptive_rag/runtime.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/adaptive_rag_common_support.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/hyde.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/hyde_controlled.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/hyde_external.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/ircot.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/ircot_common_support.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/ircot_native.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/ircot_replay.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/qrag/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/qrag/bridge.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/qrag/common_support.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/qrag/native_axis.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/qrag/reader_adapter.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/r2ag/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/r2ag/bridge.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/r2ag/common_support.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/acquire/r2ag/socket_runtime.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/clarify/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/clarify/_structured.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/clarify/active_task_disambiguation.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/clarify/ask_or_assume/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/clarify/ask_or_assume/monitor.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/clarify/future_turns.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/clarify/udcs.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/clarify/udcs_deterministic.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/acon/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/acon/common.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/acon/common_support.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/acon/history_bridge.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/acon/observation_bridge.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/llmlingua/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/llmlingua/adapters.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/llmlingua/budget.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/llmlingua/controlled.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/llmlingua/multilingual.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/llmlingua/span_map.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/recomp/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/recomp/external.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/curate/recomp/selective.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/decomp/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/decomp/common_support.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/decomp/planner_component.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/decomp/worker_bridge.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/llm_compiler/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/llm_compiler/bridge.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/llm_compiler/common_support.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/plan_and_budget/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/plan_and_budget/cached_soft_budget.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/plan_and_budget/decomposer_component.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/plan_and_budget/full.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/plan_and_budget/hard_cap.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/plan_and_budget/regenerated_soft_budget.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/plan_and_solve/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/plan_and_solve/plan_component.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/plan_and_solve/ps_prompt.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/self_ask/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/self_ask/loop.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/decompose/self_ask/planner_component.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/infocqr/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/infocqr/cache_replay.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/infocqr/external.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/infocqr/student.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/query2doc/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/query2doc/controlled_live.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/query2doc/models.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/query2doc/official_cached.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/query2doc/querygym_bridge.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/query2doc/querygym_runtime.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/reformer/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/reformer/external.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/algorithms/reformulate/rrr.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/annotation/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/annotation/independent.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/annotation/server.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/annotation/static/app.js +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/annotation/static/index.html +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/annotation/static/review.css +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/annotation/static/review.html +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/annotation/static/review.js +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/annotation/static/styles.css +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/annotation/store.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/artifacts/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/artifacts/json_file.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/artifacts/manifest.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/embedding/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/embedding/base.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/embedding/callable.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/embedding/replay.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/llm/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/llm/base.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/llm/callable.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/llm/decomp_completion.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/llm/final_content_stop.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/llm/gpt_oss_final_channel.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/llm/transformers.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/retrieval/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/retrieval/base.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/retrieval/dense.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/retrieval/memory.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/retrieval/sqlite_fts.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/user/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/user/base.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/user/callable.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/user/replay.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/vector.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/acquisition.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/ale.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/appworld_ul.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/assets.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/base.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/beir_archive.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/beir_sample.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/bfcl_native.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/bfcl_v3.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/clariq.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/data/registry.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/data/smoke_assets.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/data/statebench_v0_8_1_task_ids.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/data/toolsandbox_165848b_scenario_names.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/execution.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/interactive.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/ir_query.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/longbench.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/paired_replay.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/preprocess.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/production.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/qa.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/retrieval.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/runtime.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/statebench.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/tau3.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/toolsandbox.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/benchmarks/trec_dl_sample.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/campaign.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/campaign_runner.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/cli/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/connectors/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/connectors/knowledge/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/connectors/tools/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/connectors/tools/base.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/connectors/tools/mcp.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/connectors/tools/memory.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/connectors/tools/openapi.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/connectors/tools/schema.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/availability.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/dependency.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/enums.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/errors.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/evidence.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/intent.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/lifecycle.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/patch.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/serialization.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/state.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/trace.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/contracts/worker.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/core/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/data/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/data/checksums.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/data/downloader.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/data/http.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/data/index.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/data/overlap.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/data/preprocess.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/data/storage.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/data/trec_sample.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/evaluation/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/evaluation/outcomes.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/analysis.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/artifacts.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/automatic_measurement.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/auxiliary_matrix.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/bindings.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/blueprints.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/builders/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/builders/executable_external_payload.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/builders/external_process_arena.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/builders/registry_intervention.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/cli.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/controls.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/driver_guard.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/drivers/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/drivers/external_arena.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/drivers/method_process.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/drivers/parity_process.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/drivers/static_production.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/executor.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/loader.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/measurement.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/paper_analysis.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/parity_identity.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/plan.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/production_builders.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/readiness.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/schema.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/secrets.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/stateful_analysis.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/transfer_analysis.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/experiments/unseen_analysis.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/action_space/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/ace.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/base.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/decomp_live.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/decomp_planner.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/infocqr_student.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/ircot.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/llmcompiler_live.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/longllmlingua.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/method.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/process.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/external_workers/supervisor.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/full_architecture/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/full_architecture/hyde.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/full_architecture/plan_and_budget.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/live_history.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/prompt_optimization/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/integrations/training/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/metrics/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/metrics/catalog.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/metrics/harm.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/metrics/paired.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/metrics/qa.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/metrics/text.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/metrics/third_party.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/models/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/observability/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/observability/budget.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/observability/instrumentation.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/observability/models.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/observability/monitor.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/optimizers/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/optimizers/clarify/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/optimizers/context/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/optimizers/decompose/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/optimizers/evidence/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/optimizers/rewrite/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/parity/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/parity/base.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/production_campaign.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/py.typed +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/ace.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/ace_bfcl.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/ace_common_support.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/ace_native_plan.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/ace_online.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/ace_retrieval.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/ace_statebench.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/ace_tau.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/ace_toolsandbox.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/acon_guideline.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/adaptive_rag.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/llmlingua2_inference.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/llmlingua2_training.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/mce.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/mce_final.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/mce_text_fit.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/mce_text_only.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/qrag.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/qrag_inference.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/r2ag_features.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/r2ag_inference.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/r2ag_training.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/recomp_abstractive.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/recomp_extractive.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/reformer.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/rqrag.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/rqrag_common_support.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/rqrag_completion.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/skillopt.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/skillopt_common_support.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/skillopt_environment.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/skillopt_environment_cli.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/skillopt_retrieval.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_curriculum.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_grpo_contracts.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_grpo_data.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_grpo_meter.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_grpo_prefix.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_grpo_rollout.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_grpo_tools.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_grpo_training.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_grpo_transport.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_grpo_worker.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_inference.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_native_data.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_policy.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_policy_worker.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_sampling.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_teacher.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/recipes/smartsearch_training.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/registry/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/registry/beta.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/registry/data/algorithms.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/registry/data/auxiliary_method_profiles.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/registry/loader.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/registry/result_manifest.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/registry/schema.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/runner/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/runner/balance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/runner/cache.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/runner/composite.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/runner/engine.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/runner/runtime_guard.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/runner/worker.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/runtime_bootstrap.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/advanced.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/artifacts.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/base.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/campaigns.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/catalog.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/discovery.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/evaluation.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/features.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/learned.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/no_op.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/outcome_table.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/producer.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/rules.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/selectors/runtime.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studies/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studies/design.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studies/paired.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/campaigns.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/engine.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/model_backends.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/prepared.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/server.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/static/app.js +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/static/campaigns.html +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/static/campaigns.js +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/static/index.html +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/static/result.html +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/static/result.js +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/studio/static/styles.css +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab.egg-info/dependency_links.txt +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab.egg-info/entry_points.txt +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab.egg-info/requires.txt +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab.egg-info/top_level.txt +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/acceptance_fit_transport.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/acceptance_method_bindings.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/acceptance_plan.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/check_bfcl_native_source.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/check_distribution.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/compatibility_inventory.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/installed_sdk_smoke.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/prepare_adaptive_boundary_cases.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/prepare_adaptive_trained_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/resume_native_entry_failures.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_ace_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_ace_online_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_acon_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_acon_guideline_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_acon_history_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_adaptive_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_adaptive_controlled_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_adaptive_trained_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_beir_archive_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_bfcl_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_clariq_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_curator_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_decomp_live_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_hyde_external_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_ircot_controlled_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_ircot_native_qa_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_llmcompiler_live_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_llmcompiler_planner_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_llmlingua2_bert_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_llmlingua2_trained_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_longbench_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_mce_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_native_arena_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_pab_full_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_pab_hard_cap_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_plan_budget_component_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_qa_extended_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_qa_method_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_qa_schema_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_qrag_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_qrag_trained_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_recomp_abstractive_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_recomp_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_recomp_selective_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_recomp_sentence_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_recomp_trained_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_reformer_profiles_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_retrieval_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_rqrag_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_rqrag_fullpath_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_self_ask_component_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_skillopt_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_smartsearch_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_static_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_tau_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_tooltalk_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/run_udcs_deterministic_acceptance.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/tools/summarize_acceptance_evidence.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ace/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ace/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ace/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ace/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/acon/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/acon/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/acon/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/acon/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/acon/fitted.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/acon/guideline.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/acon/guideline_source_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/adaptive_rag/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/adaptive_rag/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/adaptive_rag/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/adaptive_rag/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/adaptive_rag/fixture_runtime.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/adaptive_rag/runtime.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ask_or_assume/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ask_or_assume/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ask_or_assume/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ask_or_assume/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/decomp/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/decomp/UPSTREAM_LICENSE.Apache-2.0.txt +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/decomp/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/decomp/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/decomp/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/decomp/live.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/decomp/runtime.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/infocqr/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/infocqr/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/infocqr/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/infocqr/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/controlled.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/fixture_bindings.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/fixture_config.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/fixture_prompt.txt +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/native_qa.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/qa_compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/ircot/runtime.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/llmcompiler/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/llmcompiler/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/llmcompiler/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/llmcompiler/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/llmcompiler/export_prompts.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/llmcompiler/live.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/llmcompiler/patch_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/mce/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/mce/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/mce/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/mce/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/mce/final_eval.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/capture_records.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/protocol.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/providers/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/providers/_live_history.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/providers/_message_compat.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/providers/appworld_ul.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/providers/fixture.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/providers/statebench.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/providers/tau3.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/providers/tau_retail.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/providers/toolsandbox.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/providers/tooltalk.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/retail_capture.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/official_arenas/retail_driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/qrag/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/qrag/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/qrag/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/qrag/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/r2ag/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/r2ag/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/r2ag/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/r2ag/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/r2ag/feature_schema.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/r2ag/runtime.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/recomp/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/recomp/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/recomp/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/recomp/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/reformer/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/reformer/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/reformer/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/reformer/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/reformer/induction.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/rrr/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/rrr/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/rrr/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/rrr/driver.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/smartsearch/README.md +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/smartsearch/__init__.py +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/smartsearch/compatibility_manifest.json +0 -0
- {agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/workers/smartsearch/driver.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: agentinputlab
|
|
3
|
-
Version: 0.1.
|
|
3
|
+
Version: 0.1.0a4
|
|
4
4
|
Summary: Auditable, intent-aware input intervention for LLM agents
|
|
5
5
|
Author: Chuang Zhao, Xiaomeng Li
|
|
6
6
|
License-Expression: MIT
|
|
@@ -287,7 +287,7 @@ includes `docs/getting-started.md` with a real HTTP backend example and troubles
|
|
|
287
287
|
Python 3.10+; optional model dependencies have their own Python/platform constraints.
|
|
288
288
|
|
|
289
289
|
```sh
|
|
290
|
-
python -m pip install agentinputlab==0.1.
|
|
290
|
+
python -m pip install agentinputlab==0.1.0a4
|
|
291
291
|
python -m agentinputlab --version
|
|
292
292
|
python -m agentinputlab doctor --offline --json
|
|
293
293
|
python -m agentinputlab smoke --json
|
|
@@ -296,9 +296,9 @@ python -m agentinputlab smoke --json
|
|
|
296
296
|
To obtain the public source and included documentation from PyPI:
|
|
297
297
|
|
|
298
298
|
```sh
|
|
299
|
-
python -m pip download --no-deps --no-binary=agentinputlab agentinputlab==0.1.
|
|
300
|
-
tar -xzf agentinputlab-0.1.
|
|
301
|
-
cd agentinputlab-0.1.
|
|
299
|
+
python -m pip download --no-deps --no-binary=agentinputlab agentinputlab==0.1.0a4
|
|
300
|
+
tar -xzf agentinputlab-0.1.0a4.tar.gz
|
|
301
|
+
cd agentinputlab-0.1.0a4
|
|
302
302
|
python -m pip install .
|
|
303
303
|
```
|
|
304
304
|
|
|
@@ -318,7 +318,7 @@ The Python distribution is an initial alpha. The JavaScript SDK remains source-o
|
|
|
318
318
|
```sh
|
|
319
319
|
python -m pip install build
|
|
320
320
|
python -m build
|
|
321
|
-
python -m pip install dist/agentinputlab-0.1.
|
|
321
|
+
python -m pip install dist/agentinputlab-0.1.0a4-py3-none-any.whl
|
|
322
322
|
cd packages/sdk
|
|
323
323
|
npm ci
|
|
324
324
|
npm test
|
|
@@ -329,6 +329,22 @@ npm install /absolute/path/to/agentinputlab-sdk-0.1.0-alpha.1.tgz
|
|
|
329
329
|
|
|
330
330
|
## Python: apply a real intervention
|
|
331
331
|
|
|
332
|
+
### Discover before running (0.1.0a3+)
|
|
333
|
+
|
|
334
|
+
```sh
|
|
335
|
+
agentinputlab list --releases --search query2doc
|
|
336
|
+
agentinputlab show plan.plan_and_solve.ps_prompt --parameters
|
|
337
|
+
agentinputlab datasets list --search beir
|
|
338
|
+
agentinputlab datasets show beir.arguana
|
|
339
|
+
```
|
|
340
|
+
|
|
341
|
+
These commands inspect metadata and constructor parameters without running an
|
|
342
|
+
algorithm or downloading data. Dataset details include source sizes and native
|
|
343
|
+
runtime prerequisites. Start with the [first-run guide](docs/getting-started.md)
|
|
344
|
+
and [dataset walkthrough](docs/benchmark-setup.md).
|
|
345
|
+
|
|
346
|
+
### Apply a prompt intervention
|
|
347
|
+
|
|
332
348
|
```python
|
|
333
349
|
from agentinputlab import FrozenAgentState, InterventionContext
|
|
334
350
|
from agentinputlab.api import create_intervention, run_intervention
|
|
@@ -12,7 +12,7 @@ includes `docs/getting-started.md` with a real HTTP backend example and troubles
|
|
|
12
12
|
Python 3.10+; optional model dependencies have their own Python/platform constraints.
|
|
13
13
|
|
|
14
14
|
```sh
|
|
15
|
-
python -m pip install agentinputlab==0.1.
|
|
15
|
+
python -m pip install agentinputlab==0.1.0a4
|
|
16
16
|
python -m agentinputlab --version
|
|
17
17
|
python -m agentinputlab doctor --offline --json
|
|
18
18
|
python -m agentinputlab smoke --json
|
|
@@ -21,9 +21,9 @@ python -m agentinputlab smoke --json
|
|
|
21
21
|
To obtain the public source and included documentation from PyPI:
|
|
22
22
|
|
|
23
23
|
```sh
|
|
24
|
-
python -m pip download --no-deps --no-binary=agentinputlab agentinputlab==0.1.
|
|
25
|
-
tar -xzf agentinputlab-0.1.
|
|
26
|
-
cd agentinputlab-0.1.
|
|
24
|
+
python -m pip download --no-deps --no-binary=agentinputlab agentinputlab==0.1.0a4
|
|
25
|
+
tar -xzf agentinputlab-0.1.0a4.tar.gz
|
|
26
|
+
cd agentinputlab-0.1.0a4
|
|
27
27
|
python -m pip install .
|
|
28
28
|
```
|
|
29
29
|
|
|
@@ -43,7 +43,7 @@ The Python distribution is an initial alpha. The JavaScript SDK remains source-o
|
|
|
43
43
|
```sh
|
|
44
44
|
python -m pip install build
|
|
45
45
|
python -m build
|
|
46
|
-
python -m pip install dist/agentinputlab-0.1.
|
|
46
|
+
python -m pip install dist/agentinputlab-0.1.0a4-py3-none-any.whl
|
|
47
47
|
cd packages/sdk
|
|
48
48
|
npm ci
|
|
49
49
|
npm test
|
|
@@ -54,6 +54,22 @@ npm install /absolute/path/to/agentinputlab-sdk-0.1.0-alpha.1.tgz
|
|
|
54
54
|
|
|
55
55
|
## Python: apply a real intervention
|
|
56
56
|
|
|
57
|
+
### Discover before running (0.1.0a3+)
|
|
58
|
+
|
|
59
|
+
```sh
|
|
60
|
+
agentinputlab list --releases --search query2doc
|
|
61
|
+
agentinputlab show plan.plan_and_solve.ps_prompt --parameters
|
|
62
|
+
agentinputlab datasets list --search beir
|
|
63
|
+
agentinputlab datasets show beir.arguana
|
|
64
|
+
```
|
|
65
|
+
|
|
66
|
+
These commands inspect metadata and constructor parameters without running an
|
|
67
|
+
algorithm or downloading data. Dataset details include source sizes and native
|
|
68
|
+
runtime prerequisites. Start with the [first-run guide](docs/getting-started.md)
|
|
69
|
+
and [dataset walkthrough](docs/benchmark-setup.md).
|
|
70
|
+
|
|
71
|
+
### Apply a prompt intervention
|
|
72
|
+
|
|
57
73
|
```python
|
|
58
74
|
from agentinputlab import FrozenAgentState, InterventionContext
|
|
59
75
|
from agentinputlab.api import create_intervention, run_intervention
|
|
@@ -27,8 +27,12 @@ from agentinputlab import InterventionContext
|
|
|
27
27
|
from agentinputlab.algorithms.clarify import ActiveTaskDisambiguation
|
|
28
28
|
from agentinputlab.backends.llm import OpenAICompatibleBackend
|
|
29
29
|
|
|
30
|
-
backend = OpenAICompatibleBackend(
|
|
31
|
-
|
|
30
|
+
backend = OpenAICompatibleBackend(
|
|
31
|
+
model_id=os.environ['MODEL_ID'],
|
|
32
|
+
base_url=os.environ['MODEL_BASE_URL'],
|
|
33
|
+
deployment_identity={'revision': os.environ['MODEL_REVISION']},
|
|
34
|
+
api_key_env='MODEL_API_KEY',
|
|
35
|
+
)
|
|
32
36
|
service = ToolkitService(
|
|
33
37
|
factories={'clarify.active_task_disambiguation': ActiveTaskDisambiguation},
|
|
34
38
|
context_factory=lambda seed: InterventionContext(backend=backend, seed=seed),
|
|
@@ -0,0 +1,76 @@
|
|
|
1
|
+
# Find, prepare and use a dataset
|
|
2
|
+
|
|
3
|
+
These discovery commands (0.1.0a3+) read packaged metadata. They do not create
|
|
4
|
+
caches, download data or call a model:
|
|
5
|
+
|
|
6
|
+
```sh
|
|
7
|
+
agentinputlab datasets list
|
|
8
|
+
agentinputlab datasets list --search beir
|
|
9
|
+
agentinputlab datasets list --family reformulate --json
|
|
10
|
+
agentinputlab datasets show beir.arguana
|
|
11
|
+
```
|
|
12
|
+
|
|
13
|
+
`automatic_loader` tells you whether `load_benchmark()` supports preparation for
|
|
14
|
+
that entry. `source_download_bytes` is the size of **all source files for a cold
|
|
15
|
+
cache**, not the currently missing bytes, unpacked disk usage or GPU memory.
|
|
16
|
+
`exceeds_default_download_limit` flags sources larger than the default 512 MiB
|
|
17
|
+
download ceiling. Review upstream terms before using `--accept-license-review`.
|
|
18
|
+
Do not raise the ceiling just to inspect a dataset; use a documented source-side
|
|
19
|
+
sample for large datasets when appropriate.
|
|
20
|
+
|
|
21
|
+
## Cache-first data loading
|
|
22
|
+
|
|
23
|
+
Install the optional dependencies required by the selected adapter:
|
|
24
|
+
|
|
25
|
+
```sh
|
|
26
|
+
python -m pip install 'agentinputlab[benchmarks,evaluation]==0.1.0a4'
|
|
27
|
+
agentinputlab datasets prepare beir.arguana --cache-dir ./toolkit-cache
|
|
28
|
+
```
|
|
29
|
+
|
|
30
|
+
Preparation verifies pinned source hashes and reuses valid cached files. It never
|
|
31
|
+
silently overwrites a corrupt cache. To prohibit network access, pass `--offline`.
|
|
32
|
+
Set `AGENTINPUTLAB_CACHE` or `cache_dir` to choose a storage location; no changes
|
|
33
|
+
are made to your other research or model caches.
|
|
34
|
+
|
|
35
|
+
```python
|
|
36
|
+
from agentinputlab.benchmarks import describe_benchmark, list_benchmarks, load_benchmark
|
|
37
|
+
|
|
38
|
+
print([item['cell_id'] for item in list_benchmarks(search='arguana')])
|
|
39
|
+
info = describe_benchmark('beir.arguana')
|
|
40
|
+
adapter = load_benchmark('beir.arguana', cache_dir='./toolkit-cache', offline=True)
|
|
41
|
+
example = adapter.examples(info['split'])[0]
|
|
42
|
+
state = example.visible_state()
|
|
43
|
+
print(state.query)
|
|
44
|
+
```
|
|
45
|
+
|
|
46
|
+
Pass `visible_state()` to the model, **not** the entire example: the full example
|
|
47
|
+
also carries evaluator-only gold. Use the adapter's typed prediction object and
|
|
48
|
+
`score()` after real downstream execution. A successful input rewrite is not a
|
|
49
|
+
task score, and a retrieval query is not a QA answer. The
|
|
50
|
+
[real-execution guide](real-execution.md) covers readers, retrieval and native
|
|
51
|
+
scorers.
|
|
52
|
+
|
|
53
|
+
## Native and interactive runtimes
|
|
54
|
+
|
|
55
|
+
`native_adapter` means there is no automatic loader in `load_benchmark()` for
|
|
56
|
+
that entry; it does not mean the dataset has no adapter. Do not substitute a
|
|
57
|
+
different dataset to bypass its prerequisites.
|
|
58
|
+
|
|
59
|
+
| Dataset type | Preparation route |
|
|
60
|
+
| --- | --- |
|
|
61
|
+
| Prepared QA / retrieval | `datasets show` → `datasets prepare` → `load_benchmark`. |
|
|
62
|
+
| BFCL simple | `load_bfcl_v3_simple_adapter` with its prepared bundle and official AST scorer bridge. |
|
|
63
|
+
| STATE, ToolSandbox, ToolTalk and interactive arenas | Native provider/environment, checkpoint or task selection, and official continuation/scoring. Inspect `python -m workers.official_arenas.driver --help` and [real execution](real-execution.md). |
|
|
64
|
+
| τ³ banking checkpoints | `load_benchmark` additionally requires `checkpoints` and `evaluation_bridge` before downloading. |
|
|
65
|
+
| Large BEIR / TREC collections | Use the documented `BEIRSampleAdapter` or `datasets prepare-trec-sample` route when full sources exceed your storage budget. |
|
|
66
|
+
|
|
67
|
+
Algorithm eligibility and runtime compatibility remain separate from dataset
|
|
68
|
+
availability. Check `eligibility_gate`, `family_support` and the chosen algorithm
|
|
69
|
+
profile; a catalog entry is not a universal compatibility guarantee.
|
|
70
|
+
|
|
71
|
+
## Debug without repeating model requests
|
|
72
|
+
|
|
73
|
+
Use `agentinputlab --debug COMMAND ...` for toolkit-error tracebacks locally.
|
|
74
|
+
Normal mode prints a short error and exits with status 2. Keep debug traces
|
|
75
|
+
private until reviewed for paths or application details. Fix configuration before
|
|
76
|
+
resubmitting: the CLI does not automatically retry consumed model calls.
|
|
@@ -8,7 +8,7 @@ can require a particular Python version, platform, upstream checkout or weights.
|
|
|
8
8
|
python -m venv .venv
|
|
9
9
|
# macOS/Linux; Windows PowerShell: .venv\Scripts\Activate.ps1
|
|
10
10
|
source .venv/bin/activate
|
|
11
|
-
python -m pip install agentinputlab==0.1.
|
|
11
|
+
python -m pip install agentinputlab==0.1.0a4
|
|
12
12
|
python -m agentinputlab --version
|
|
13
13
|
python -m agentinputlab doctor --offline --json
|
|
14
14
|
python -m agentinputlab smoke --json
|
|
@@ -25,16 +25,29 @@ the selected Python interpreter even when its scripts directory is not on PATH.
|
|
|
25
25
|
## Choose a profile before installing heavy dependencies
|
|
26
26
|
|
|
27
27
|
```sh
|
|
28
|
-
agentinputlab list --releases --json
|
|
28
|
+
agentinputlab list --releases --search plan_and_solve --json
|
|
29
29
|
agentinputlab show plan.plan_and_solve.ps_prompt
|
|
30
|
+
agentinputlab show plan.plan_and_solve.ps_prompt --parameters
|
|
30
31
|
```
|
|
31
32
|
|
|
32
33
|
Use the exact release ID and its prerequisite documentation. An available
|
|
33
34
|
entrypoint is not a configured runtime. Install only the extras you need, for
|
|
34
|
-
example `python -m pip install 'agentinputlab[benchmarks,evaluation]==0.1.
|
|
35
|
+
example `python -m pip install 'agentinputlab[benchmarks,evaluation]==0.1.0a4'`.
|
|
35
36
|
Do not install all extras into one environment: some native methods deliberately
|
|
36
37
|
pin incompatible upstream versions. Keep those runtimes in separate environments.
|
|
37
38
|
|
|
39
|
+
The search and parameter-inspection commands require 0.1.0a3+. They expose the
|
|
40
|
+
actual constructor signature without creating the algorithm or making a model
|
|
41
|
+
call. In Python use `describe_intervention(release_id)` from `agentinputlab.api`.
|
|
42
|
+
Required constructor arguments are listed separately from runtime requirements
|
|
43
|
+
in the profile's help and prerequisites; an optional `client=None` may still
|
|
44
|
+
require an external worker at execution time.
|
|
45
|
+
|
|
46
|
+
For datasets, start with `agentinputlab datasets list --search beir` and
|
|
47
|
+
`agentinputlab datasets show beir.arguana`. See the
|
|
48
|
+
[dataset preparation walkthrough](benchmark-setup.md) for cache configuration,
|
|
49
|
+
source sizes, offline loading and native-runtime prerequisites.
|
|
50
|
+
|
|
38
51
|
## Make one real model call
|
|
39
52
|
|
|
40
53
|
No OpenAI SDK or local GPU package is required for the HTTP backend. Configure
|
|
@@ -78,6 +91,10 @@ see [real execution and profile-specific dependencies](real-execution.md).
|
|
|
78
91
|
| HTTP 429 | Provider quota and concurrent request limits. |
|
|
79
92
|
| HTTP 5xx / transport failure | Local server/provider health and network reachability. |
|
|
80
93
|
| Profile unavailable | Check the profile's required worker, source revision and artifacts; installing the core does not provision them. |
|
|
94
|
+
| Unknown profile / wrong constructor parameters | `list --releases --search NAME`, then `show EXACT_RELEASE --parameters`. Suggestions never silently change the selected profile. |
|
|
95
|
+
| Dataset cannot be prepared | `datasets show ID` distinguishes automatic loaders from native adapters and lists source sizes and runtime requirements. |
|
|
96
|
+
| Cannot create artifact cache directory | Choose a writable `--cache-dir` / `cache_dir` or set `AGENTINPUTLAB_CACHE`; do not change shared research directories' permissions. |
|
|
97
|
+
| Response schema is invalid / missing usage | The endpoint must return explicit integer `usage.prompt_tokens` and `usage.completion_tokens`. Missing counts are unknown, not zero; fix the provider or gateway before another request. |
|
|
81
98
|
|
|
82
99
|
Version 0.1.0a2 exposes a sanitized `OpenAICompatibleTransportError`
|
|
83
100
|
with `http_status` for the built-in HTTP transport. Status hints contain no
|
|
@@ -87,3 +104,7 @@ remain generic unless they use the internal sanitized status path.
|
|
|
87
104
|
There are **no automatic model retries**. A failed request may already have
|
|
88
105
|
consumed tokens: retain the failure and unknown cost rather than treating it as
|
|
89
106
|
zero usage. Decide any retry explicitly within your own declared budget.
|
|
107
|
+
|
|
108
|
+
Starting in 0.1.0a4, incomplete token-usage fields are rejected even when the
|
|
109
|
+
response contains an answer. Explicit zero counts remain valid. This prevents
|
|
110
|
+
an endpoint that omits accounting fields from silently bypassing token budgets.
|
|
@@ -34,19 +34,65 @@ from agentinputlab.runner import CompositeInterventionRunner, InterventionRunner
|
|
|
34
34
|
from agentinputlab.selectors import NoOpIntervention
|
|
35
35
|
|
|
36
36
|
|
|
37
|
+
def describe_intervention(
|
|
38
|
+
release_id: str, *, registry: AlgorithmRegistry | None = None
|
|
39
|
+
) -> dict[str, Any]:
|
|
40
|
+
"""Inspect a constructor without creating a runtime or making model calls."""
|
|
41
|
+
registry = registry or AlgorithmRegistry.default()
|
|
42
|
+
if release_id == "baseline.no_op":
|
|
43
|
+
factory = NoOpIntervention
|
|
44
|
+
metadata = {"id": release_id, "execution_kind": "input_patch"}
|
|
45
|
+
else:
|
|
46
|
+
_, release = registry.get_release(release_id)
|
|
47
|
+
metadata = release.to_dict()
|
|
48
|
+
if not release.entrypoint_available:
|
|
49
|
+
return {**metadata, "parameters": [], "blocker": release.blocker}
|
|
50
|
+
factory = registry.resolve_entrypoint(release_id)
|
|
51
|
+
parameters = []
|
|
52
|
+
for name, param in inspect.signature(factory).parameters.items():
|
|
53
|
+
if name == "release_id":
|
|
54
|
+
continue # Bound by the registry, not chosen by the caller.
|
|
55
|
+
item = {
|
|
56
|
+
"name": name,
|
|
57
|
+
"kind": param.kind.name.lower(),
|
|
58
|
+
"required": param.default is inspect.Parameter.empty
|
|
59
|
+
and param.kind not in (param.VAR_POSITIONAL, param.VAR_KEYWORD),
|
|
60
|
+
"annotation": (
|
|
61
|
+
None if param.annotation is inspect.Parameter.empty
|
|
62
|
+
else inspect.formatannotation(param.annotation)
|
|
63
|
+
),
|
|
64
|
+
}
|
|
65
|
+
if param.default is not inspect.Parameter.empty:
|
|
66
|
+
default = param.default
|
|
67
|
+
item["default"] = (
|
|
68
|
+
default if default is None or isinstance(default, (str, int, float, bool))
|
|
69
|
+
else repr(default)
|
|
70
|
+
)
|
|
71
|
+
parameters.append(item)
|
|
72
|
+
return {**metadata, "parameters": parameters, "help": inspect.getdoc(factory)}
|
|
73
|
+
|
|
74
|
+
|
|
37
75
|
def create_intervention(release_id: str, **parameters: Any) -> Any:
|
|
38
76
|
"""Instantiate a registered profile; optional runtime assets remain explicit.
|
|
39
77
|
|
|
40
78
|
Parameters use the adapter's original Python signature. This does not turn
|
|
41
79
|
composite agents into input-patch methods or assert benchmark eligibility.
|
|
42
80
|
"""
|
|
43
|
-
|
|
44
|
-
|
|
45
|
-
|
|
81
|
+
factory = (
|
|
82
|
+
NoOpIntervention if release_id == "baseline.no_op"
|
|
83
|
+
else AlgorithmRegistry.default().resolve_entrypoint(release_id)
|
|
84
|
+
)
|
|
46
85
|
signature = inspect.signature(factory)
|
|
47
86
|
if "release_id" in signature.parameters:
|
|
48
87
|
parameters["release_id"] = release_id
|
|
49
|
-
|
|
88
|
+
try:
|
|
89
|
+
signature.bind(**parameters)
|
|
90
|
+
except TypeError as exc:
|
|
91
|
+
raise ValidationError(
|
|
92
|
+
f"invalid parameters for {release_id}: {exc}; "
|
|
93
|
+
f"inspect with describe_intervention({release_id!r}) or "
|
|
94
|
+
f"agentinputlab show {release_id} --parameters"
|
|
95
|
+
) from None
|
|
50
96
|
result = factory(**parameters)
|
|
51
97
|
actual = getattr(result, "release_id", release_id)
|
|
52
98
|
if actual != release_id:
|
|
@@ -28,7 +28,14 @@ class ArtifactStore:
|
|
|
28
28
|
|
|
29
29
|
def __init__(self, root: str | Path) -> None:
|
|
30
30
|
self.root = Path(root).expanduser().resolve()
|
|
31
|
-
|
|
31
|
+
try:
|
|
32
|
+
self.root.mkdir(parents=True, exist_ok=True)
|
|
33
|
+
except OSError as exc:
|
|
34
|
+
raise ArtifactError(
|
|
35
|
+
f"cannot create artifact cache directory {self.root}: {exc.strerror}; "
|
|
36
|
+
"choose a writable directory with cache_dir / --cache-dir or "
|
|
37
|
+
"AGENTINPUTLAB_CACHE; do not change permissions on shared research data"
|
|
38
|
+
) from exc
|
|
32
39
|
|
|
33
40
|
def path_for(self, manifest: ArtifactManifest) -> Path:
|
|
34
41
|
candidate = (self.root / manifest.artifact_id / manifest.revision / "payload").resolve()
|
{agentinputlab-0.1.0a2 → agentinputlab-0.1.0a4}/src/agentinputlab/backends/llm/openai_compatible.py
RENAMED
|
@@ -281,7 +281,11 @@ class OpenAICompatibleBackend(LLMBackend):
|
|
|
281
281
|
# Leave the handler before raising so remote-controlled values cannot
|
|
282
282
|
# survive in an exception context rendered by a traceback logger.
|
|
283
283
|
if schema_error or response is None:
|
|
284
|
-
raise RuntimeError(
|
|
284
|
+
raise RuntimeError(
|
|
285
|
+
"OpenAI-compatible response schema is invalid; require choices and explicit "
|
|
286
|
+
"usage.prompt_tokens / usage.completion_tokens; missing usage is unknown, "
|
|
287
|
+
"not zero; no automatic retry"
|
|
288
|
+
)
|
|
285
289
|
return response
|
|
286
290
|
|
|
287
291
|
def fingerprint(self) -> Mapping[str, Any]:
|
|
@@ -337,9 +341,7 @@ def _parse_generation_response(
|
|
|
337
341
|
finish_reason = ""
|
|
338
342
|
if not isinstance(finish_reason, str):
|
|
339
343
|
raise TypeError("response finish_reason must be a string")
|
|
340
|
-
usage = data.get("usage"
|
|
341
|
-
if usage is None:
|
|
342
|
-
usage = {}
|
|
344
|
+
usage = data.get("usage")
|
|
343
345
|
if not isinstance(usage, Mapping):
|
|
344
346
|
raise TypeError("response usage must be a mapping")
|
|
345
347
|
return GenerationResponse(
|
|
@@ -347,8 +349,8 @@ def _parse_generation_response(
|
|
|
347
349
|
model=model,
|
|
348
350
|
finish_reason=finish_reason,
|
|
349
351
|
usage=Usage(
|
|
350
|
-
input_tokens=_response_token_count(usage
|
|
351
|
-
output_tokens=_response_token_count(usage
|
|
352
|
+
input_tokens=_response_token_count(usage["prompt_tokens"]),
|
|
353
|
+
output_tokens=_response_token_count(usage["completion_tokens"]),
|
|
352
354
|
),
|
|
353
355
|
raw=data,
|
|
354
356
|
)
|
|
@@ -46,6 +46,7 @@ from .bfcl_v3 import (
|
|
|
46
46
|
load_bfcl_v3_simple_adapter,
|
|
47
47
|
)
|
|
48
48
|
from .cache import benchmark_cache_dir, load_benchmark, prepare_benchmark
|
|
49
|
+
from .discovery import describe_benchmark, list_benchmarks
|
|
49
50
|
from .execution import (
|
|
50
51
|
BenchmarkArmExecutionRecord,
|
|
51
52
|
BenchmarkBlockArm,
|
|
@@ -231,7 +232,9 @@ __all__ = [
|
|
|
231
232
|
"benchmark_cache_dir",
|
|
232
233
|
"build_ale_official_command",
|
|
233
234
|
"create_interactive_benchmark_adapter",
|
|
235
|
+
"describe_benchmark",
|
|
234
236
|
"interactive_benchmark_specs",
|
|
237
|
+
"list_benchmarks",
|
|
235
238
|
"load_benchmark",
|
|
236
239
|
"load_bfcl_v3_simple_adapter",
|
|
237
240
|
"load_prepared_benchmark",
|
|
@@ -29,6 +29,7 @@ from .assets import BenchmarkSmokeAssetRegistry
|
|
|
29
29
|
from .beir_archive import BEIR_ARCHIVE_LOCKS, BEIRArchiveAdapter
|
|
30
30
|
from .preprocess import _prepared_path, load_prepared_benchmark
|
|
31
31
|
from .production import load_production_benchmark_adapter, production_benchmark_specs
|
|
32
|
+
from .registry import BenchmarkRegistry
|
|
32
33
|
|
|
33
34
|
DEFAULT_DOWNLOAD_LIMIT = 512 * 1024 * 1024
|
|
34
35
|
|
|
@@ -83,6 +84,7 @@ def prepare_benchmark(
|
|
|
83
84
|
License-review sources require explicit acknowledgement, including cache use.
|
|
84
85
|
Corrupt caches raise an integrity error and are never silently overwritten.
|
|
85
86
|
"""
|
|
87
|
+
BenchmarkRegistry.default().get(cell_id)
|
|
86
88
|
root = benchmark_cache_dir(cache_dir)
|
|
87
89
|
positive_byte_limit(max_download_bytes, name="max_download_bytes")
|
|
88
90
|
positive_byte_limit(max_prepared_bytes, name="max_prepared_bytes")
|
|
@@ -192,6 +194,13 @@ def load_benchmark(
|
|
|
192
194
|
):
|
|
193
195
|
"""Prepare missing data and construct the corresponding official adapter."""
|
|
194
196
|
# Check dynamic-runtime requirements before spending bandwidth.
|
|
197
|
+
allowed = {"checkpoints", "evaluation_bridge"} if cell_id == "tau3.banking_knowledge" else set()
|
|
198
|
+
unexpected = set(runtime) - allowed
|
|
199
|
+
if unexpected:
|
|
200
|
+
raise ValidationError(
|
|
201
|
+
f"unsupported runtime arguments for {cell_id}: {', '.join(sorted(unexpected))}; "
|
|
202
|
+
"check agentinputlab datasets show before preparing data"
|
|
203
|
+
)
|
|
195
204
|
if cell_id == "tau3.banking_knowledge" and (
|
|
196
205
|
runtime.get("checkpoints") is None or runtime.get("evaluation_bridge") is None
|
|
197
206
|
):
|
|
@@ -207,8 +216,6 @@ def load_benchmark(
|
|
|
207
216
|
max_prepared_bytes=max_prepared_bytes,
|
|
208
217
|
)
|
|
209
218
|
if report["kind"] == "beir_archive":
|
|
210
|
-
if runtime:
|
|
211
|
-
raise ValidationError("BEIR archive loader accepts no dynamic-runtime arguments")
|
|
212
219
|
return BEIRArchiveAdapter(
|
|
213
220
|
benchmark_id=cell_id, source_archive=report["source_archive"], license_acknowledged=True
|
|
214
221
|
)
|
|
@@ -0,0 +1,87 @@
|
|
|
1
|
+
"""Metadata-only discovery: no downloads, cache creation or runtime execution."""
|
|
2
|
+
|
|
3
|
+
from __future__ import annotations
|
|
4
|
+
|
|
5
|
+
from agentinputlab.contracts.serialization import to_jsonable
|
|
6
|
+
|
|
7
|
+
from .assets import BenchmarkSmokeAssetRegistry
|
|
8
|
+
from .beir_archive import BEIR_ARCHIVE_LOCKS
|
|
9
|
+
from .cache import DEFAULT_DOWNLOAD_LIMIT
|
|
10
|
+
from .production import production_benchmark_specs
|
|
11
|
+
from .registry import BenchmarkRegistry
|
|
12
|
+
|
|
13
|
+
|
|
14
|
+
def describe_benchmark(cell_id: str) -> dict:
|
|
15
|
+
"""Return source sizes, prerequisites and an explicit preparation route.
|
|
16
|
+
|
|
17
|
+
Source size is for a cold cache, not an estimate of normalized storage or
|
|
18
|
+
GPU memory. An automatic loader is not proof of algorithm compatibility.
|
|
19
|
+
"""
|
|
20
|
+
cell = BenchmarkRegistry.default().get(cell_id)
|
|
21
|
+
result = {
|
|
22
|
+
**to_jsonable(cell),
|
|
23
|
+
"automatic_loader": False,
|
|
24
|
+
"loader_kind": "native_adapter",
|
|
25
|
+
"split": None,
|
|
26
|
+
"sources": [],
|
|
27
|
+
"source_download_bytes": None,
|
|
28
|
+
"exceeds_default_download_limit": None,
|
|
29
|
+
"runtime_parameters": [],
|
|
30
|
+
"prepare_command": None,
|
|
31
|
+
"next_step": "Use the documented native adapter and its runtime prerequisites; "
|
|
32
|
+
"this cell is not supported by load_benchmark(). See docs/benchmark-setup.md.",
|
|
33
|
+
}
|
|
34
|
+
specs = {spec.cell_id: spec for spec in production_benchmark_specs()}
|
|
35
|
+
if cell_id in specs:
|
|
36
|
+
spec = specs[cell_id]
|
|
37
|
+
assets = BenchmarkSmokeAssetRegistry.default()
|
|
38
|
+
sources = [assets.get(key) for key in (spec.query_asset_id, spec.corpus_asset_id) if key]
|
|
39
|
+
result.update(
|
|
40
|
+
loader_kind=spec.adapter_kind,
|
|
41
|
+
split=spec.split,
|
|
42
|
+
sources=[{
|
|
43
|
+
"asset_id": asset.asset_id, "url": asset.source_url,
|
|
44
|
+
"size_bytes": asset.size_bytes, "sha256": asset.sha256,
|
|
45
|
+
} for asset in sources],
|
|
46
|
+
license_review_required=any(a.status == "license_review_required" for a in sources),
|
|
47
|
+
source_download_bytes=sum(a.size_bytes for a in sources),
|
|
48
|
+
)
|
|
49
|
+
if spec.adapter_kind == "checkpoint":
|
|
50
|
+
result["runtime_parameters"] = ["checkpoints", "evaluation_bridge"]
|
|
51
|
+
elif cell_id in BEIR_ARCHIVE_LOCKS:
|
|
52
|
+
lock = BEIR_ARCHIVE_LOCKS[cell_id]
|
|
53
|
+
result.update(
|
|
54
|
+
loader_kind="beir_archive", split="test", license_review_required=True,
|
|
55
|
+
source_download_bytes=lock["size_bytes"],
|
|
56
|
+
sources=[{
|
|
57
|
+
"url": "https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/"
|
|
58
|
+
+ lock["dataset"] + ".zip",
|
|
59
|
+
"size_bytes": lock["size_bytes"], "sha256": lock["sha256"],
|
|
60
|
+
}],
|
|
61
|
+
)
|
|
62
|
+
else:
|
|
63
|
+
return result
|
|
64
|
+
result["automatic_loader"] = True
|
|
65
|
+
size = result["source_download_bytes"]
|
|
66
|
+
result["exceeds_default_download_limit"] = size > DEFAULT_DOWNLOAD_LIMIT
|
|
67
|
+
command = f"agentinputlab datasets prepare {cell_id}"
|
|
68
|
+
if result["license_review_required"]:
|
|
69
|
+
command += " --accept-license-review"
|
|
70
|
+
result["prepare_command"] = command
|
|
71
|
+
result["next_step"] = (
|
|
72
|
+
"Review upstream terms before acknowledging them. Prepare a verified cache, then "
|
|
73
|
+
"use load_benchmark() with the listed runtime parameters. Large sources require "
|
|
74
|
+
"an explicit download limit; preparation can need additional disk space. "
|
|
75
|
+
"See docs/benchmark-setup.md."
|
|
76
|
+
)
|
|
77
|
+
return result
|
|
78
|
+
|
|
79
|
+
|
|
80
|
+
def list_benchmarks(*, search: str = "", family: str | None = None) -> list[dict]:
|
|
81
|
+
"""Search the packaged inventory by ID, suite or task (case insensitive)."""
|
|
82
|
+
query = search.casefold()
|
|
83
|
+
return [
|
|
84
|
+
describe_benchmark(cell.cell_id)
|
|
85
|
+
for cell in BenchmarkRegistry.default().list(family=family)
|
|
86
|
+
if query in f"{cell.cell_id} {cell.suite} {cell.task}".casefold()
|
|
87
|
+
]
|
|
@@ -5,6 +5,7 @@ from __future__ import annotations
|
|
|
5
5
|
import json
|
|
6
6
|
from collections.abc import Mapping
|
|
7
7
|
from dataclasses import dataclass
|
|
8
|
+
from difflib import get_close_matches
|
|
8
9
|
from importlib import resources
|
|
9
10
|
from typing import Any
|
|
10
11
|
|
|
@@ -94,7 +95,9 @@ class BenchmarkRegistry:
|
|
|
94
95
|
try:
|
|
95
96
|
return self._by_id[cell_id]
|
|
96
97
|
except KeyError as exc:
|
|
97
|
-
|
|
98
|
+
matches = get_close_matches(cell_id, self._by_id, n=3)
|
|
99
|
+
hint = f"; did you mean: {', '.join(matches)}?" if matches else "; use agentinputlab datasets list"
|
|
100
|
+
raise RegistryError(f"unknown benchmark cell: {cell_id}{hint}") from exc
|
|
98
101
|
|
|
99
102
|
def list(
|
|
100
103
|
self,
|