# Trajectory step classification on Trajectory nine-category label set Source: [Trajectory step classification on Trajectory nine-category label set]() Collection index: [llms.txt]() Data updated: 2026-05-21T11:04:55.785Z Evaluation of trajectory step classification performance on the Trajectory nine-category label set, utilizing metrics such as Label Accuracy, Macro-F1, and step-wise error metrics. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Trajectory step classification | | Dataset (including any stated split) | Trajectory nine-category label set | | Metric | Label Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Label Accuracy](): Higher is better; unit: PERCENT. - [Macro-F1 Score](): Higher is better; unit: PERCENT. - [Step Exact Accuracy](): Higher is better; unit: PERCENT. - [Step ±1 Accuracy](): Higher is better; unit: PERCENT. - [Joint Score](): Higher is better; unit: PERCENT. - [Primary Source Score](): Higher is better; unit: PERCENT. - [Impact Score](): Higher is better; unit: PERCENT. - [Primary + Impact Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Label Accuracy | Result date | Report paper | | --- | --- | --- | --- | | Claude Haiku 4.5 | 95 | Not recorded | 6a0e919cf984739ed6738b8c | | Gemma-4-26B-A4B-it | 94.3 | Not recorded | 6a0e919cf984739ed6738b8c | | Gemini 3.1 Flash Lite | 92.7 | Not recorded | 6a0e919cf984739ed6738b8c | | Ministral-3-14B | 90.5 | Not recorded | 6a0e919cf984739ed6738b8c | | Qwen3.6-35B-A3B | 89.5 | Not recorded | 6a0e919cf984739ed6738b8c | | Claude Sonnet 4.6 | 88.5 | Not recorded | 6a0e919cf984739ed6738b8c | | gpt-5.4-mini | 81.8 | Not recorded | 6a0e919cf984739ed6738b8c | | gpt-oss-20B | 68.6 | Not recorded | 6a0e919cf984739ed6738b8c | | gpt-5.4-mini | 61.5 | Not recorded | 6a0e919cf984739ed6738b8c | | Claude Sonnet 4.6 | 61.3 | Not recorded | 6a0e919cf984739ed6738b8c | | Claude Haiku 4.5 | 56.9 | Not recorded | 6a0e919cf984739ed6738b8c | | Gemini 3.1 Flash Lite | 56.4 | Not recorded | 6a0e919cf984739ed6738b8c | | Qwen3.6-35B-A3B | 55.5 | Not recorded | 6a0e919cf984739ed6738b8c | | Gemma-4-26B-A4B-it | 54.2 | Not recorded | 6a0e919cf984739ed6738b8c | | gpt-oss-20B | 53.8 | Not recorded | 6a0e919cf984739ed6738b8c | | Ministral-3-14B | 53.7 | Not recorded | 6a0e919cf984739ed6738b8c | ### Claude Haiku 4.5 Label Accuracy: 95. Result ID: 6a0ee565fbc373592c06f704. [Reporting paper]() Recorded evaluation configuration: {"Tier":"frontier","Mode":"aware"}. This record does not assert independent reproduction. ### Gemma-4-26B-A4B-it Label Accuracy: 94.3. Result ID: 6a0ee565fbc373592c06f70c. [Reporting paper]() Recorded evaluation configuration: {"Tier":"open","Mode":"aware"}. This record does not assert independent reproduction. ### Gemini 3.1 Flash Lite Label Accuracy: 92.7. Result ID: 6a0ee565fbc373592c06f708. [Reporting paper]() Recorded evaluation configuration: {"Tier":"frontier","Mode":"aware"}. This record does not assert independent reproduction. ### Ministral-3-14B Label Accuracy: 90.5. Result ID: 6a0ee565fbc373592c06f70e. [Reporting paper]() Recorded evaluation configuration: {"Tier":"open","Mode":"aware"}. This record does not assert independent reproduction. ### Qwen3.6-35B-A3B Label Accuracy: 89.5. Result ID: 6a0ee565fbc373592c06f70a. [Reporting paper]() Recorded evaluation configuration: {"Tier":"open","Mode":"aware"}. This record does not assert independent reproduction. ### Claude Sonnet 4.6 Label Accuracy: 88.5. Result ID: 6a0ee565fbc373592c06f702. [Reporting paper]() Recorded evaluation configuration: {"Tier":"frontier","Mode":"aware"}. This record does not assert independent reproduction. ### gpt-5.4-mini Label Accuracy: 81.8. Result ID: 6a0ee565fbc373592c06f706. [Reporting paper]() Recorded evaluation configuration: {"Tier":"frontier","Mode":"aware"}. This record does not assert independent reproduction. ### gpt-oss-20B Label Accuracy: 68.6. Result ID: 6a0ee565fbc373592c06f710. [Reporting paper]() Recorded evaluation configuration: {"Tier":"open","Mode":"aware"}. This record does not assert independent reproduction. ### gpt-5.4-mini Label Accuracy: 61.5. Result ID: 6a0ee565fbc373592c06f707. [Reporting paper]() Recorded evaluation configuration: {"Tier":"frontier","Mode":"blind"}. This record does not assert independent reproduction. ### Claude Sonnet 4.6 Label Accuracy: 61.3. Result ID: 6a0ee565fbc373592c06f703. [Reporting paper]() Recorded evaluation configuration: {"Tier":"frontier","Mode":"blind"}. This record does not assert independent reproduction. ### Claude Haiku 4.5 Label Accuracy: 56.9. Result ID: 6a0ee565fbc373592c06f705. [Reporting paper]() Recorded evaluation configuration: {"Tier":"frontier","Mode":"blind"}. This record does not assert independent reproduction. ### Gemini 3.1 Flash Lite Label Accuracy: 56.4. Result ID: 6a0ee565fbc373592c06f709. [Reporting paper]() Recorded evaluation configuration: {"Tier":"frontier","Mode":"blind"}. This record does not assert independent reproduction. ### Qwen3.6-35B-A3B Label Accuracy: 55.5. Result ID: 6a0ee565fbc373592c06f70b. [Reporting paper]() Recorded evaluation configuration: {"Tier":"open","Mode":"blind"}. This record does not assert independent reproduction. ### Gemma-4-26B-A4B-it Label Accuracy: 54.2. Result ID: 6a0ee565fbc373592c06f70d. [Reporting paper]() Recorded evaluation configuration: {"Tier":"open","Mode":"blind"}. This record does not assert independent reproduction. ### gpt-oss-20B Label Accuracy: 53.8. Result ID: 6a0ee565fbc373592c06f711. [Reporting paper]() Recorded evaluation configuration: {"Tier":"open","Mode":"blind"}. This record does not assert independent reproduction. ### Ministral-3-14B Label Accuracy: 53.7. Result ID: 6a0ee565fbc373592c06f70f. [Reporting paper]() Recorded evaluation configuration: {"Tier":"open","Mode":"blind"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.