Dataset · Hugging Face
MC7ever/super-lfm-blend-v4
super-lfm-blend-v4 Blended SFT dataset for the super-lfm project. 279K rows from 87 sources. Splits train: 279080 rows validation: 14643 rows Think-fix Every assistant turn is prefixed with <think></think> to fix the empty-content bug.
| Author / organization | MC7ever |
|---|---|
| License (as declared) | other |
| Downloads | 0 (as reported by Hugging Face) |
| Likes | 0 |
| Created | 22 October 2026 |
| Last updated at source | 22 October 2026 |
Popularity, last 2 days
0 now. Counts as reported by the source, snapshotted daily by Silytics.
Recent changes
- 22 Oct 2026 License: none → other
- 22 Oct 2026 Description: none → super-lfm-blend-v4 Blended SFT dataset for the super-lfm project. 279K rows from 87 sources. Splits train: 279080 rows validation: 14643 rows Think-fix Every assistant turn is prefixed with <think></think> to fix the empty-content bug.
- 22 Oct 2026 Topics added task_categories:text-generation, persona, sft, blend, think-fix
Changes detected by Silytics between refreshes of the source.
More from MC7ever
- MC7ever/hf-training-corpusDataset
- MC7ever/OHB-20kDataset
- MC7ever/mega-corpus-english-v2Dataset
- MC7ever/all-in-one-llm-corpusDataset
- MC7ever/web-scraper-datasetDataset
- MC7ever/super-lfm-blend-v3Dataset
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 23 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.