# Rule Following on IHEval Multi-Turn Source: [Rule Following on IHEval Multi-Turn]() Collection index: [llms.txt]() Data updated: 2026-04-13T08:53:11.498Z Evaluation of model performance on Rule Following using the Multi-Turn subset of the IHEval dataset, measured by reference, alignment, and confidence accuracy 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 | Rule Following | | Dataset (including any stated split) | IHEval Multi-Turn | | Metric | Accuracy (Reference) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 12 of 12 | | 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 (Reference)](): Higher is better; unit: PERCENT. - [Accuracy (Alignment)](): Higher is better; unit: PERCENT. - [Accuracy (Confidence)](): Higher is better; unit: PERCENT. - [Average Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (Reference) | Result date | Report paper | | --- | --- | --- | --- | | Qwen3-4B-it | 89.8 | Not recorded | 69dc78667d0172140ec78952 | | Qwen3-4B-it-NSHA-HCAL | 89.5 | Not recorded | 69dc78667d0172140ec78952 | | Qwen3-4B-it-NS | 89.1 | Not recorded | 69dc78667d0172140ec78952 | | Qwen3-4B-it-NSHA-DPO | 89 | Not recorded | 69dc78667d0172140ec78952 | | Qwen3-4B-it-CoT | 88.8 | Not recorded | 69dc78667d0172140ec78952 | | Llama3.1-8B | 81.6 | Not recorded | 69dc78667d0172140ec78952 | | Qwen3-4B-it-NSHA-SFT | 80.3 | Not recorded | 69dc78667d0172140ec78952 | | Llama3.1-8B-NS | 79.9 | Not recorded | 69dc78667d0172140ec78952 | | Llama3.1-8B-NSHA-HCAL | 78.8 | Not recorded | 69dc78667d0172140ec78952 | | Llama3.1-8B-NSHA-DPO | 77.2 | Not recorded | 69dc78667d0172140ec78952 | | Llama3.1-8B-CoT | 74.7 | Not recorded | 69dc78667d0172140ec78952 | | Llama3.1-8B-NSHA-SFT | 27.4 | Not recorded | 69dc78667d0172140ec78952 | ### Qwen3-4B-it Accuracy (Reference): 89.8. Result ID: 69dcaee447f34ce924c3bfd8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-it","Temperature":"0.7","top-p":"0.8","top-k":"20","Inference Engine":"vLLM"}. This record does not assert independent reproduction. ### Qwen3-4B-it-NSHA-HCAL Accuracy (Reference): 89.5. Result ID: 69dcaee447f34ce924c3bfdd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-it","Training Variant":"NSHA-HCAL","Temperature":"0.7","top-p":"0.8","top-k":"20"}. This record does not assert independent reproduction. ### Qwen3-4B-it-NS Accuracy (Reference): 89.1. Result ID: 69dcaee447f34ce924c3bfda. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-it","Inference variant":"Neuro-symbolic (NS)","Semantic Identification":"GPT-5-mini","Temperature":"0.7","top-p":"0.8","top-k":"20"}. This record does not assert independent reproduction. ### Qwen3-4B-it-NSHA-DPO Accuracy (Reference): 89. Result ID: 69dcaee447f34ce924c3bfdc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-it","Training Variant":"NSHA-DPO","Temperature":"0.7","top-p":"0.8","top-k":"20"}. This record does not assert independent reproduction. ### Qwen3-4B-it-CoT Accuracy (Reference): 88.8. Result ID: 69dcaee447f34ce924c3bfd9. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-it","Prompting":"Chain-of-thought (CoT)","Temperature":"0.7","top-p":"0.8","top-k":"20","Inference Engine":"vLLM"}. This record does not assert independent reproduction. ### Llama3.1-8B Accuracy (Reference): 81.6. Result ID: 69dcaee447f34ce924c3bfde. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Llama3.1-8B-Instruct","Temperature":"0.7","top-p":"0.8","top-k":"20","Inference Engine":"vLLM"}. This record does not assert independent reproduction. ### Qwen3-4B-it-NSHA-SFT Accuracy (Reference): 80.3. Result ID: 69dcaee447f34ce924c3bfdb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-it","Training Variant":"NSHA-SFT","Temperature":"0.7","top-p":"0.8","top-k":"20"}. This record does not assert independent reproduction. ### Llama3.1-8B-NS Accuracy (Reference): 79.9. Result ID: 69dcaee447f34ce924c3bfe0. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Llama3.1-8B-Instruct","Inference variant":"Neuro-symbolic (NS)","Temperature":"0.7","top-p":"0.8","top-k":"20"}. This record does not assert independent reproduction. ### Llama3.1-8B-NSHA-HCAL Accuracy (Reference): 78.8. Result ID: 69dcaee447f34ce924c3bfe3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama3.1-8B-Instruct","Training Variant":"NSHA-HCAL","Temperature":"0.7","top-p":"0.8","top-k":"20"}. This record does not assert independent reproduction. ### Llama3.1-8B-NSHA-DPO Accuracy (Reference): 77.2. Result ID: 69dcaee447f34ce924c3bfe2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama3.1-8B-Instruct","Training Variant":"NSHA-DPO","Temperature":"0.7","top-p":"0.8","top-k":"20"}. This record does not assert independent reproduction. ### Llama3.1-8B-CoT Accuracy (Reference): 74.7. Result ID: 69dcaee447f34ce924c3bfdf. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Llama3.1-8B-Instruct","Prompting":"Chain-of-thought (CoT)","Temperature":"0.7","top-p":"0.8","top-k":"20"}. This record does not assert independent reproduction. ### Llama3.1-8B-NSHA-SFT Accuracy (Reference): 27.4. Result ID: 69dcaee447f34ce924c3bfe1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama3.1-8B-Instruct","Training Variant":"NSHA-SFT","Temperature":"0.7","top-p":"0.8","top-k":"20"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.