ArrowLabs/DCLM-Ultra-1.1
DCLM Ultra DCLM Ultra 1.1 is an improved version of DCLM Ultra (Heavily filtered and deduplicated DCLM Pro). It achieves superior performance across benchmarks including hellaswag and piqa, also achieving better loss compared to the original ultra. 3M model trained on only DCLM Ultra vs DCLM Ultra 1.1, same architecture, same training budget. Metric DCLM Ultra DCLM Ultra 1.1 Training loss 3.27 3.25 PIQA 54.95% 55.55% HellaSwag 25.54% 26.34%… See the full description on the dataset page: https://huggingface.co/datasets/ArrowLabs/DCLM-Ultra-1.1.
| Author / organization | ArrowLabs |
|---|---|
| License (as declared) | apache-2.0 |
| Downloads | 333 (as reported by Hugging Face) |
| Likes | 3 |
| Created | 27 October 2026 |
| Last updated at source | 29 October 2026 |
Popularity, last 9 days
333 now (+164 since 2026-11-02). Counts as reported by the source, snapshotted daily by Silytics.
Connected in the knowledge graph
- mradermacher/tinybrainbot-360m-a75m-v5.1-base-GGUFModel · used to train
- nkthebass/tinybrainbot-10m-v2-baseModel · used to train
- mradermacher/tinybrainbot-100m-v5.1-base-GGUFModel · used to train
- onnx-community/tinybrainbot-10m-v2-base-ONNXModel · used to train
- onnx-community/tinybrainbot-100m-v5-base-ONNXModel · used to train
- nkthebass/tinybrainbot-100m-v5-baseModel · used to train
- nkthebass/tinybrainbot-100m-v5.1-baseModel · used to train
- nkthebass/tinybrainbot-360m-a75m-v5.1-baseModel · used to train
More from ArrowLabs
- ArrowLabs/DCLM-UltraDataset
- ArrowLabs/PerfectPdfsDataset
- ArrowLabs/ArrowBench-InstructDataset
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 10 November 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.