# Multitask Language Understanding on C-MMLU (Accuracy) Source: [Multitask Language Understanding on C-MMLU (Accuracy)]() Collection index: [llms.txt]() Data updated: 2026-03-10T07:44:26.134Z Evaluation of model performance on the C-MMLU dataset for Multitask Language Understanding, reporting standard 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 | Multitask Language Understanding | | Dataset (including any stated split) | C-MMLU | | Metric | Accuracy (C-MMLU) | | 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 - [Accuracy (C-MMLU)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (C-MMLU) | Result date | Report paper | | --- | --- | --- | --- | | Meta-Llama-3-8B | 51.2 | Not recorded | 69afa57c2d603e6de3e439b4 | | RouterDC | 51 | Not recorded | 69afa57c2d603e6de3e439b4 | | RouterDC + RACER-G (w/ Agg.) | 50.7 | Not recorded | 69afa57c2d603e6de3e439b4 | | RouterDC + RACER-P (w/ Agg.) | 50.6 | Not recorded | 69afa57c2d603e6de3e439b4 | | Chinese-Mistral-7B-v0.1 | 49.1 | Not recorded | 69afa57c2d603e6de3e439b4 | | KNNR + RACER-P (w/ Agg.) | 49 | Not recorded | 69afa57c2d603e6de3e439b4 | | KNNR + RACER-G (w/ Agg.) | 48.8 | Not recorded | 69afa57c2d603e6de3e439b4 | | MLPR + RACER-P (w/ Agg.) | 48.4 | Not recorded | 69afa57c2d603e6de3e439b4 | | MLPR + RACER-G (w/ Agg.) | 48.3 | Not recorded | 69afa57c2d603e6de3e439b4 | | KNNR | 48.1 | Not recorded | 69afa57c2d603e6de3e439b4 | | Mistral-7B-v0.1 | 44.6 | Not recorded | 69afa57c2d603e6de3e439b4 | | MetaMath-Mistral-7B | 44.4 | Not recorded | 69afa57c2d603e6de3e439b4 | | MLPR | 44.4 | Not recorded | 69afa57c2d603e6de3e439b4 | | Dolphin-2.6-Mistral-7b | 44.2 | Not recorded | 69afa57c2d603e6de3e439b4 | | Dolphin-2.9-Llama3-8b | 43.9 | Not recorded | 69afa57c2d603e6de3e439b4 | | Zephyr-7b-beta | 43.5 | Not recorded | 69afa57c2d603e6de3e439b4 | ### Meta-Llama-3-8B Accuracy (C-MMLU): 51.2. Result ID: 69afcbc06dbb7075b957db07. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Type":"Candidate Model"}. This record does not assert independent reproduction. ### RouterDC Accuracy (C-MMLU): 51. Result ID: 69afcbc06dbb7075b957db11. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Type":"Base Router"}. This record does not assert independent reproduction. ### RouterDC + RACER-G (w/ Agg.) Accuracy (C-MMLU): 50.7. Result ID: 69afcbc06dbb7075b957db12. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Router":"RouterDC","Score Definition":"router score-gap","Aggregation":"w/ Agg."}. This record does not assert independent reproduction. ### RouterDC + RACER-P (w/ Agg.) Accuracy (C-MMLU): 50.6. Result ID: 69afcbc06dbb7075b957db13. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Router":"RouterDC","Score Definition":"inverse probability","Aggregation":"w/ Agg."}. This record does not assert independent reproduction. ### Chinese-Mistral-7B-v0.1 Accuracy (C-MMLU): 49.1. Result ID: 69afcbc06dbb7075b957db04. [Reporting paper]() Recorded evaluation configuration: {"Type":"Candidate Model"}. This record does not assert independent reproduction. ### KNNR + RACER-P (w/ Agg.) Accuracy (C-MMLU): 49. Result ID: 69afcbc06dbb7075b957db10. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Router":"KNNR","Score Definition":"inverse probability","Aggregation":"w/ Agg."}. This record does not assert independent reproduction. ### KNNR + RACER-G (w/ Agg.) Accuracy (C-MMLU): 48.8. Result ID: 69afcbc06dbb7075b957db0f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Router":"KNNR","Score Definition":"router score-gap","Aggregation":"w/ Agg."}. This record does not assert independent reproduction. ### MLPR + RACER-P (w/ Agg.) Accuracy (C-MMLU): 48.4. Result ID: 69afcbc06dbb7075b957db0d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Router":"MLPR","Score Definition":"inverse probability","Aggregation":"w/ Agg."}. This record does not assert independent reproduction. ### MLPR + RACER-G (w/ Agg.) Accuracy (C-MMLU): 48.3. Result ID: 69afcbc06dbb7075b957db0c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Router":"MLPR","Score Definition":"router score-gap","Aggregation":"w/ Agg."}. This record does not assert independent reproduction. ### KNNR Accuracy (C-MMLU): 48.1. Result ID: 69afcbc06dbb7075b957db0e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"Base Router"}. This record does not assert independent reproduction. ### Mistral-7B-v0.1 Accuracy (C-MMLU): 44.6. Result ID: 69afcbc06dbb7075b957db09. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Type":"Candidate Model"}. This record does not assert independent reproduction. ### MetaMath-Mistral-7B Accuracy (C-MMLU): 44.4. Result ID: 69afcbc06dbb7075b957db08. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Type":"Candidate Model"}. This record does not assert independent reproduction. ### MLPR Accuracy (C-MMLU): 44.4. Result ID: 69afcbc06dbb7075b957db0b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"Base Router"}. This record does not assert independent reproduction. ### Dolphin-2.6-Mistral-7b Accuracy (C-MMLU): 44.2. Result ID: 69afcbc06dbb7075b957db05. [Reporting paper]() Recorded evaluation configuration: {"Type":"Candidate Model"}. This record does not assert independent reproduction. ### Dolphin-2.9-Llama3-8b Accuracy (C-MMLU): 43.9. Result ID: 69afcbc06dbb7075b957db06. [Reporting paper]() Recorded evaluation configuration: {"Type":"Candidate Model"}. This record does not assert independent reproduction. ### Zephyr-7b-beta Accuracy (C-MMLU): 43.5. Result ID: 69afcbc06dbb7075b957db0a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Type":"Candidate Model"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.