# Tactic classification on Customer-service conversations Source: [Tactic classification on Customer-service conversations]() Collection index: [llms.txt]() Data updated: 2026-06-16T09:24:33.821Z Evaluation of Tactic classification performance on customer-service conversations, measured using Tactic F1 and Tactic Accuracy. 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 | Tactic classification | | Dataset (including any stated split) | Customer-service conversations | | Metric | Tactic F1 | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 14 of 14 | | 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 - [Tactic F1](): Higher is better; unit: PERCENT. - [Tactic Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Tactic F1 | Result date | Report paper | | --- | --- | --- | --- | | ELECTRA | 84.14 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | DeBERTa | 83.8 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | RoBERTa | 83.35 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | RoBERTa | 81.32 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | DeBERTa | 80.92 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | ELECTRA | 80.67 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | GPT-4 | 70.94 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | Qwen2-72B | 66.97 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | GPT-4 | 64.87 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | GPT-NeoX-20B | 64.51 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | Llama-3.1-8B | 62.11 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | Qwen2-72B | 53.51 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | GPT-NeoX-20B | 51.84 | Not recorded | 6a30d8fad33dfb23e783d0c4 | | Llama-3.1-8B | 38.33 | Not recorded | 6a30d8fad33dfb23e783d0c4 | ### ELECTRA Tactic F1: 84.14. Result ID: 6a31163ffbc373592c2db782. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method":"EARS"}. This record does not assert independent reproduction. ### DeBERTa Tactic F1: 83.8. Result ID: 6a31163ffbc373592c2db783. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method":"EARS"}. This record does not assert independent reproduction. ### RoBERTa Tactic F1: 83.35. Result ID: 6a31163ffbc373592c2db781. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method":"EARS"}. This record does not assert independent reproduction. ### RoBERTa Tactic F1: 81.32. Result ID: 6a31163ffbc373592c2db77e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method":"Transformer"}. This record does not assert independent reproduction. ### DeBERTa Tactic F1: 80.92. Result ID: 6a31163ffbc373592c2db780. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method":"Transformer"}. This record does not assert independent reproduction. ### ELECTRA Tactic F1: 80.67. Result ID: 6a31163ffbc373592c2db77f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method":"Transformer"}. This record does not assert independent reproduction. ### GPT-4 Tactic F1: 70.94. Result ID: 6a31163ffbc373592c2db77d. [Reporting paper]() Recorded evaluation configuration: {"Method":"LLM Few-shot"}. This record does not assert independent reproduction. ### Qwen2-72B Tactic F1: 66.97. Result ID: 6a31163ffbc373592c2db77c. [Reporting paper]() Recorded evaluation configuration: {"Method":"LLM Few-shot"}. This record does not assert independent reproduction. ### GPT-4 Tactic F1: 64.87. Result ID: 6a31163ffbc373592c2db779. [Reporting paper]() Recorded evaluation configuration: {"Method":"LLM Zero-shot"}. This record does not assert independent reproduction. ### GPT-NeoX-20B Tactic F1: 64.51. Result ID: 6a31163ffbc373592c2db77b. [Reporting paper]() Recorded evaluation configuration: {"Method":"LLM Few-shot"}. This record does not assert independent reproduction. ### Llama-3.1-8B Tactic F1: 62.11. Result ID: 6a31163ffbc373592c2db77a. [Reporting paper]() Recorded evaluation configuration: {"Method":"LLM Few-shot"}. This record does not assert independent reproduction. ### Qwen2-72B Tactic F1: 53.51. Result ID: 6a31163ffbc373592c2db778. [Reporting paper]() Recorded evaluation configuration: {"Method":"LLM Zero-shot"}. This record does not assert independent reproduction. ### GPT-NeoX-20B Tactic F1: 51.84. Result ID: 6a31163ffbc373592c2db777. [Reporting paper]() Recorded evaluation configuration: {"Method":"LLM Zero-shot"}. This record does not assert independent reproduction. ### Llama-3.1-8B Tactic F1: 38.33. Result ID: 6a31163ffbc373592c2db776. [Reporting paper]() Recorded evaluation configuration: {"Method":"LLM Zero-shot"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.