A dataset hosted by the mlx-community on Hugging Face, last updated on 2026-01-30. It is a 400K-instance collection designed for supervised fine-tuning (SFT) of large language models, specifically excluding tool-use instructions. The description provides examples for training with the MLX-LM and MLX-LM-LoRA frameworks.
Use Cases
- Supervised fine-tuning of language models based on the dataset's 400K instruction-response pairs.
- Instruction-tuning for general-purpose chat models based on the 'No-Tools' focus mentioned in the title.
- Benchmarking SFT training pipelines based on the dataset's scale and specific application examples provided.
- Training lightweight models (e.g., 0.5B parameter) using the provided MLX-LM-LoRA example configuration.
Strengths
- Dataset scale is explicitly stated as 400K instances in the title.
- Specific use-case examples for training with MLX-LM and MLX-LM-LoRA are provided in the description.
- Platform tags indicate the dataset is optimized and stored in Parquet format.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown beyond the title's '400K' label, which may limit suitability assessment.
- Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
- Source
- mlx-community on Hugging Face
- Collection Method
- Likely compiled or generated for instruction-tuning tasks.
- Time Range
- null
- Freshness
- Last updated 2026-01-30 23:59:21; freshness should be verified.
- Geography
- null