Ling-Coder-SFT: Supervised Fine-Tuning Data for Code Generation

More than 5 million samples of supervised fine-tuning data used to train the Ling-Coder Lite model. The dataset is part of a larger collection that also includes DPO and synthetic QA subsets, created by inclusionAI and last updated on March 27, 2025.

Use Cases

• Fine-tuning code generation models based on the described SFT data.
• Training models for instruction following in a coding context based on the dataset's purpose.
• Conducting comparative training experiments using the described DPO and synthetic QA subsets.
• Researching the impact of synthetic data on model performance for code-related tasks.

Strengths

• Contains more than 5 million samples for supervised fine-tuning.
• Includes a separate subset of 250k samples for Direct Preference Optimization (DPO).
• Part of a structured collection with multiple components for different training stages.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the primary SFT subset is not precisely specified ('more than 5 million').
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

inclusionAI via Hugging Face.

Collection Method

Likely contains text data generated or curated for training the Ling-Coder Lite model.

Time Range

null

Freshness

Last updated 2025-03-27 12:40:12; freshness should be verified.

Geography

null