Fineweb Edu Llama3 Annotations: Educational Quality Scores for Web Pages

Annotations from the Llama-3-70B-Instruct model scoring web pages from the FineWeb corpus for their educational value. The dataset contains the text sample, the prompt using the first 1000 characters, and the assigned scores. HuggingFaceFW created this dataset, last updated on 2024-06-03.

Use Cases

• Train an educational content classifier based on the provided quality scores.
• Benchmark the performance of other LLMs on educational text scoring tasks.
• Analyze correlations between web page text features and perceived educational value.
• Filter large web corpora for high-quality educational content using the provided scores.

Strengths

• Annotations generated by a large, capable model (Llama-3-70B-Instruct).
• Scores are based on a specific prompt using the first 1000 characters of each text sample.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset does not contain the full Llama 3 generation, only the scores.

Provenance

Source

HuggingFaceFW

Collection Method

Scores generated by prompting Llama-3-70B-Instruct on text samples from the FineWeb corpus.

Freshness

Last updated 2024-06-03 18:29:31; freshness should be verified.