# LLM Pretraining on C4 (Final val) Source: [LLM Pretraining on C4 (Final val)]() Collection index: [llms.txt]() Data updated: 2026-06-01T10:05:04.487Z Evaluation of Large Language Model performance during the pre-training phase using the final validation split of the C4 dataset, measured primarily by perplexity. 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 | LLM pretraining | | Dataset (including any stated split) | C4 (Final val) | | Metric | Perplexity | | Unit | SCALAR | | Direction | Lower is better | | Results loaded | 5 of 5 | | 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 - [Perplexity](): Lower is better; unit: SCALAR. ## Recorded results | Method | Perplexity | Result date | Report paper | | --- | --- | --- | --- | | SoftMuon | 18.288 | Not recorded | 6a1d1234dbd7150506f84151 | | SoftSignum | 18.519 | Not recorded | 6a1d1234dbd7150506f84151 | | Muon | 18.591 | Not recorded | 6a1d1234dbd7150506f84151 | | Signum | 18.688 | Not recorded | 6a1d1234dbd7150506f84151 | | AdamW | 18.709 | Not recorded | 6a1d1234dbd7150506f84151 | ### SoftMuon Perplexity: 18.288. Result ID: 6a1d590afbc373592c171a9c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaMa 130M","Batch size":"512","Sequence length":"256","Learning rate scheduling":"cosine-annealing","Number of runs":"3"}. This record does not assert independent reproduction. ### SoftSignum Perplexity: 18.519. Result ID: 6a1d590afbc373592c171a9e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaMa 130M","Batch size":"512","Sequence length":"256","Learning rate scheduling":"cosine-annealing","Number of runs":"3"}. This record does not assert independent reproduction. ### Muon Perplexity: 18.591. Result ID: 6a1d590afbc373592c171a9d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaMa 130M","Batch size":"512","Sequence length":"256","Learning rate scheduling":"cosine-annealing","Number of runs":"3"}. This record does not assert independent reproduction. ### Signum Perplexity: 18.688. Result ID: 6a1d590afbc373592c171a9f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaMa 130M","Batch size":"512","Sequence length":"256","Learning rate scheduling":"cosine-annealing","Number of runs":"3"}. This record does not assert independent reproduction. ### AdamW Perplexity: 18.709. Result ID: 6a1d590afbc373592c171aa0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaMa 130M","Batch size":"512","Sequence length":"256","Learning rate scheduling":"cosine-annealing","Number of runs":"3"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.