Local Code Arena Starcoder 15B: MBPP Benchmark Telemetry

Raw evaluation metrics, execution telemetry logs, and structural syntax outputs from running the Mostly Basic Python Problems (MBPP) benchmark against the StarCoder 15B base model. This partition documents scaling limits of unaligned foundational weights in conversational benchmarking loops. The dataset was authored by ShahzebKhoso and last updated on 2026-05-28.

Use Cases

• Benchmarking code generation models based on execution telemetry logs
• Analyzing structural syntax outputs from foundational models
• Establishing baselines for unaligned model weights in conversational loops
• Studying scaling limits of heavyweight foundational models

Strengths

• Contains raw evaluation metrics from a specific benchmark (MBPP)
• Focuses on a heavyweight foundational model (StarCoder 15B)
• Captures execution telemetry logs and structural syntax outputs

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Row count is unknown, which may limit suitability assessment
• Column-level documentation is absent; field semantics must be inferred after download

Provenance

Source

huggingface

Collection Method

Benchmark execution telemetry captured from running the MBPP benchmark

Freshness

Last updated 2026-05-28 12:44:43; freshness should be verified