Nq Item Id Llm Compressed Hardneg 3Shot V4

417,748 examples of tokenized conversational text intended for training large language models, as indicated by the description. The dataset was created by author Lala8383 and last updated on Hugging Face in May 2026. A sample analysis of 1,000 examples shows an average of 892.62 tokens per example, with a minimum of 539 and a maximum of 1,772 tokens.

Use Cases

• Fine-tuning language models for question answering based on the described 'hard negative' and '3-shot' training structure.
• Benchmarking tokenization strategies for long-context LLM inputs based on the provided token statistics.
• Training retrieval systems for open-domain QA using the implied 'item id' and 'hardneg' components mentioned in the title.

Strengths

• A substantial collection of 417,748 training examples.
• Detailed token statistics are provided, including average (892.62), min (539), and max (1,772) tokens per example.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Freshness should be verified as the last update date is in the future (2026-05-11).

Provenance

Source

Hugging Face user Lala8383.

Collection Method

Likely generated or processed by a large language model, as suggested by the title and description.

Freshness

Last updated 2026-05-11 10:39:08