# Agent Task on SciWorld (Suc@1, Suc@2, Suc@3) Source: [Agent Task on SciWorld (Suc@1, Suc@2, Suc@3)]() Collection index: [llms.txt]() Data updated: 2026-07-07T09:11:03.819Z Evaluation of agent performance on general tasks within the SciWorld environment, measured by Success Rate at k=1, k=2, and k=3. 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 | Agent Task | | Dataset (including any stated split) | SciWorld | | Metric | Success Rate @1 | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 13 of 13 | | 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 - [Success Rate @1](): Higher is better; unit: PERCENT. - [Success Rate @2](): Higher is better; unit: PERCENT. - [Success Rate @3](): Higher is better; unit: PERCENT. ## Recorded results | Method | Success Rate @1 | Result date | Report paper | | --- | --- | --- | --- | | PivoARL | 98.4 | Not recorded | 6a4c880585dba69fef8b263d | | MetaRL | 94.5 | Not recorded | 6a4c880585dba69fef8b263d | | GSPO | 89.8 | Not recorded | 6a4c880585dba69fef8b263d | | DAPO | 83.5 | Not recorded | 6a4c880585dba69fef8b263d | | SKillRL | 82.1 | Not recorded | 6a4c880585dba69fef8b263d | | SimpleMem+GRPO | 81.7 | Not recorded | 6a4c880585dba69fef8b263d | | Mem0+GRPO | 79.6 | Not recorded | 6a4c880585dba69fef8b263d | | GiGPO | 78.9 | Not recorded | 6a4c880585dba69fef8b263d | | GRPO | 75.8 | Not recorded | 6a4c880585dba69fef8b263d | | Gemini-2.5-Pro | 42.9 | Not recorded | 6a4c880585dba69fef8b263d | | GPT-4o | 32.8 | Not recorded | 6a4c880585dba69fef8b263d | | ReAct | 9.4 | Not recorded | 6a4c880585dba69fef8b263d | | Reflexion | 9.3 | Not recorded | 6a4c880585dba69fef8b263d | ### PivoARL Success Rate @1: 98.4. Result ID: 6a4cc28f60b88bb339c94710. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. ### MetaRL Success Rate @1: 94.5. Result ID: 6a4cc28f60b88bb339c9470f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. ### GSPO Success Rate @1: 89.8. Result ID: 6a4cc28f60b88bb339c9470b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. ### DAPO Success Rate @1: 83.5. Result ID: 6a4cc28f60b88bb339c94709. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. ### SKillRL Success Rate @1: 82.1. Result ID: 6a4cc28f60b88bb339c9470e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. ### SimpleMem+GRPO Success Rate @1: 81.7. Result ID: 6a4cc28f60b88bb339c9470d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. ### Mem0+GRPO Success Rate @1: 79.6. Result ID: 6a4cc28f60b88bb339c9470c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. ### GiGPO Success Rate @1: 78.9. Result ID: 6a4cc28f60b88bb339c9470a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. ### GRPO Success Rate @1: 75.8. Result ID: 6a4cc28f60b88bb339c94708. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. ### Gemini-2.5-Pro Success Rate @1: 42.9. Result ID: 6a4cc28e60b88bb339c94705. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Gemini-2.5-Pro"}. This record does not assert independent reproduction. ### GPT-4o Success Rate @1: 32.8. Result ID: 6a4cc28e60b88bb339c94704. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"GPT-4o"}. This record does not assert independent reproduction. ### ReAct Success Rate @1: 9.4. Result ID: 6a4cc28f60b88bb339c94706. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. ### Reflexion Success Rate @1: 9.3. Result ID: 6a4cc28f60b88bb339c94707. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.