# Multi-turn Dialogue on MT-Bench (Acceptance Length τ) Source: [Multi-turn Dialogue on MT-Bench (Acceptance Length τ)]() Collection index: [llms.txt]() Data updated: 2026-06-11T06:15:29.610Z Evaluation of multi-turn dialogue performance on the MT-Bench dataset, focusing specifically on the Acceptance Length (τ) metric. 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 | Multi-turn dialogue | | Dataset (including any stated split) | MT-Bench | | Metric | Acceptance Length (τ) | | Unit | SCALAR | | 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 - [Acceptance Length (τ)](): Higher is better; unit: SCALAR. ## Recorded results | Method | Acceptance Length (τ) | Result date | Report paper | | --- | --- | --- | --- | | Llama 3.2-3B | 2.717 | Not recorded | 6a2a412ed33dfb23e78393bb | | Llama 3.2-3B | 2.542 | Not recorded | 6a2a412ed33dfb23e78393bb | | Llama 3.2-3B | 2.454 | Not recorded | 6a2a412ed33dfb23e78393bb | | Llama 3.2-3B | 2.351 | Not recorded | 6a2a412ed33dfb23e78393bb | | Llama-3-8B | 2.341 | Not recorded | 6a2a412ed33dfb23e78393bb | | Llama-3-8B | 2.222 | Not recorded | 6a2a412ed33dfb23e78393bb | | Qwen3-4B | 2.123 | Not recorded | 6a2a412ed33dfb23e78393bb | | Llama-3-8B | 2.105 | Not recorded | 6a2a412ed33dfb23e78393bb | | Qwen3-8B | 2.055 | Not recorded | 6a2a412ed33dfb23e78393bb | | Qwen3-4B | 2 | Not recorded | 6a2a412ed33dfb23e78393bb | | Qwen3-8B | 1.961 | Not recorded | 6a2a412ed33dfb23e78393bb | | Qwen3-4B | 1.959 | Not recorded | 6a2a412ed33dfb23e78393bb | | Llama-3-8B | 1.931 | Not recorded | 6a2a412ed33dfb23e78393bb | | Qwen3-8B | 1.901 | Not recorded | 6a2a412ed33dfb23e78393bb | | Qwen3-4B | 1.875 | Not recorded | 6a2a412ed33dfb23e78393bb | | Qwen3-8B | 1.836 | Not recorded | 6a2a412ed33dfb23e78393bb | ### Llama 3.2-3B Acceptance Length (τ): 2.717. Result ID: 6a2a527cfbc373592c269e53. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0.3","αc (chain-loss coefficient)":"40"}. This record does not assert independent reproduction. ### Llama 3.2-3B Acceptance Length (τ): 2.542. Result ID: 6a2a527cfbc373592c269e52. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0.3","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Llama 3.2-3B Acceptance Length (τ): 2.454. Result ID: 6a2a527cfbc373592c269e51. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Llama 3.2-3B Acceptance Length (τ): 2.351. Result ID: 6a2a527cfbc373592c269e50. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"None","αf (first-error focal coefficient)":"0","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Llama-3-8B Acceptance Length (τ): 2.341. Result ID: 6a2a527cfbc373592c269e57. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0.3","αc (chain-loss coefficient)":"40"}. This record does not assert independent reproduction. ### Llama-3-8B Acceptance Length (τ): 2.222. Result ID: 6a2a527cfbc373592c269e56. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0.3","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Qwen3-4B Acceptance Length (τ): 2.123. Result ID: 6a2a527cfbc373592c269e5b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0.3","αc (chain-loss coefficient)":"40"}. This record does not assert independent reproduction. ### Llama-3-8B Acceptance Length (τ): 2.105. Result ID: 6a2a527cfbc373592c269e55. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Qwen3-8B Acceptance Length (τ): 2.055. Result ID: 6a2a527cfbc373592c269e5f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0.3","αc (chain-loss coefficient)":"40"}. This record does not assert independent reproduction. ### Qwen3-4B Acceptance Length (τ): 2. Result ID: 6a2a527cfbc373592c269e5a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0.3","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Qwen3-8B Acceptance Length (τ): 1.961. Result ID: 6a2a527cfbc373592c269e5e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0.3","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Qwen3-4B Acceptance Length (τ): 1.959. Result ID: 6a2a527cfbc373592c269e59. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Llama-3-8B Acceptance Length (τ): 1.931. Result ID: 6a2a527cfbc373592c269e54. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"None","αf (first-error focal coefficient)":"0","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Qwen3-8B Acceptance Length (τ): 1.901. Result ID: 6a2a527cfbc373592c269e5d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"10","αf (first-error focal coefficient)":"0","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Qwen3-4B Acceptance Length (τ): 1.875. Result ID: 6a2a527cfbc373592c269e58. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"None","αf (first-error focal coefficient)":"0","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. ### Qwen3-8B Acceptance Length (τ): 1.836. Result ID: 6a2a527cfbc373592c269e5c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"γ (loss-decay constant)":"None","αf (first-error focal coefficient)":"0","αc (chain-loss coefficient)":"0"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.