FineEnvs/MiMo-V2.6-RL-harbor-terminal
MiMo-V2.6-RL-harbor-terminal: MiMo-V2.6-RL Terminal (Harbor): Harbor dataset on Hugging Face with 64 tasks. Terminal-Bench style tasks. 64 Harbor tasks from the Terminal domain of Xiaomi's MiMo-V2.6-RL-oss, the RL environments MiMo-V2.6 was trained on, converted so every one runs as a standard…
Tasks
- Repair wildcard-expanded output flag bookkeeping in a vendored workflow engine and run its offline integration replay.
- Repair finite joint-limit validation across Python and compiled numerical IK restart paths.
- Repair Bandit's file-plugin metadata integration and emit a verified offline Trojan Source scan report.
- Repair a four-module Windows forensic plugin API migration in a frozen Volatility3 tree.
- Repair Bandit's cross-module Trojan Source finding normalization and produce an offline JSON scan report.
- Repair a real Fairlearn MetricFrame feature-processing path so missing sensitive and control values are rejected…
- Repair persisted MLP incremental training across the estimator and stochastic optimizer source modules.
- Repair Partitura's real MusicXML-to-MIDI pipeline for malformed chord anchors, empty scores, one-shot iterables, and…
- Repair retrieval top-k validation consistently across functional and stateful TorchMetrics evaluation APIs.
- Repair Ibex Zcmp expansion state across exception redirects and run the offline regression.
- Repair ONNX TensorScatter circular KV-cache indexing and run the offline reference inference fixture.
- Repair effective-batch normalization in a frozen Transformers label-smoothing path and produce a deterministic CPU…
- Repair coordinated lexer ownership when cloning Lark LALR interactive parsers.
- Repair a cross-module React dropzone browser-picker fallback and validate the real source integration path.
- Repair a frozen computational-chemistry orbital-analysis integration and emit its validated report.
- Repair compressed interpreter-tag validation and wheel filename error propagation in a frozen packaging source tree…
- Repair the offline Plaso Windows Registry extraction workflow so corrupt typed values are recovered with warnings and…
- Repair structural readable-stream detection across a frozen Requests request-preparation pipeline.
- Repair Vagrant VirtualBox private-network identity normalization across driver generations.
- Repair a real routing state machine's configurable repeated-slash behavior and run its offline integration workflow.
- Repair scalar and array result normalization in three vendored Evaluate classification metrics.
- Repair bundled-development client ordering across HTML generation and memory serving in a frozen frontend build-tool…
- Repair coordinated contextual diagnostics in a real C-like binary-layout parser while preserving decoding behavior.
- Repair compressed interpreter-tag validation and wheel filename error translation in a vendored packaging library…
- Repair Arcade camera state routing, batch-camera protocol behavior, and equivalent window-block binding across both…
- Repair petl's cross-module fluent Table API collision while preserving lazy ETL semantics and produce a validated…
- Repair cross-backend virtual-address semantics in a frozen capa extractor slice.
- Repair malformed wheel interpreter-tag validation across the real packaging parser pipeline while preserving valid…
- Repair Fairlearn adversarial backend feature-matrix validation and run its integration diagnostic.
- Repair uneven-row behavior across petl's existing sort-based and hash-based relational join implementations.
- Repair Peewee transaction lock-mode propagation across SQLite and APSW wrappers, then run the offline integration check.
- Repair binary campaign-outcome validation across a frozen uplift evaluation library's metrics and visualization…
- Repair nucleus-aware Python identifier generation across a frozen particle-physics library's utility and public domain…
- Repair nested text-config context-length resolution in a frozen language-model evaluation harness.
- Repair scikit-learn PoissonRegressor mixed Array API namespace handling across fit, score, and estimator-check…
- Repair Bandit's line-oriented Trojan Source issue locations while preserving generic AST-context fallback behavior.
- Repair namespace-aware executable-link removal in an offline SVG optimization pipeline.
- Repair temporary transformation copy isolation in an offline ezdxf ACIS workflow.
- Repair a cross-module Transformers adapter-loading contract and emit an offline integration report.
- Repair the offline EarlyStopping and FitLoop lifecycle so minimum-epoch deferral, ordinary stopping, and independent…
- Repair compressed wheel-tag validation across the packaging tag parser and wheel filename utility, then run the…
- Repair nonce-derived AES-CCM message-length enforcement across construction, streaming encryption/decryption, and…
- Repair Dask's two Parquet metadata engines for all-or-none overlap between Hive partitions and physical columns.
- Repair organizer-aware calendar movement authorization across a real AngularJS service and event-form controller.
- Repair experiment exposure eligibility and flush-window deduplication in a frozen production Python SDK.
- Repair Bandit’s real B506 AST and plugin integration so safe positional YAML loaders are accepted without hiding…
- Repair XI VAT jurisdiction routing and GB canonicalization in a frozen standards library, then run an offline…
- Repair lexer ownership when copying LALR interactive parser state and produce a verified branch-isolation report.
- Repair Authlib JOSE key provenance across JWS and JWE and emit a regression report.
- Repair a frozen PyJWT cross-layer verification workflow for key-family safety, detached JWS payloads, option…
- Repair OpenSCAD range semantics across AST construction, rendering, round-trip parsing, and serialization.
- Repair a real JOSE HMAC key-classification boundary and emit a verified compact-JWT behavior report.
- Repair declarative pipeline environment propagation across discovery and worker execution.
- Repair Paramiko exception boundaries across SSH-agent, channel, client, and transport lifecycles.
- Repair a real Bandit cross-module regression so Trojan Source findings retain precise source ranges through scanning…
- Repair coupled binary-classifier ROC and precision-recall plotting for preprocessing-style estimators.
- Repair Dask Parquet metadata handling for full versus partial overlap between Hive partition columns and physical file…
- Repair variance-state reconstruction across River Bayesian streaming inference and produce an offline uncertainty audit.
- Repair hierarchical ML evaluator aggregation and produce an offline integration artifact.
- Repair MLX CPU scan, sort, and argsort evaluation for zero-length axes while preserving allocation lifecycle and…
- Repair three frozen Evaluate classification metric adapters so scalar and per-class outputs preserve their public…
- Repair River's online AdaBoost weighting and probability aggregation, then run deterministic prequential training on…
- Repair Bandit Trojan Source metadata propagation so real B613 scans serialize correct SARIF locations.
- Repair scalar and vector result adaptation across three real classification metrics.
FineEnvs/MiMo-V2.6-RL-harbor-terminal on the Hugging Face Hub