Raw evaluation metrics, execution telemetry logs, and structural syntax outputs from running the Mostly Basic Python Problems (MBPP) benchmark against the Qwen 3.5 2B base model. The dataset documents a zero-shot functional programming synthesis test under standard local execution bounds without active reasoning tokens. It was created by ShahzebKhoso and last updated on May 24, 2026.
Use Cases
- Benchmarking small language model performance on Python code synthesis based on the MBPP benchmark.
- Analyzing execution telemetry and error patterns for code generated by a 2-billion parameter model.
- Studying the structural syntax outputs of a next-generation instruction model in a zero-shot setting.
- Comparing functional programming synthesis results across different model architectures and sizes.
Strengths
- Captures raw execution telemetry logs from a standardized benchmark run.
- Documents a specific test of a 2-billion parameter model (Qwen 3.5 2B) on zero-shot Python programming.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
Provenance
- Source
- huggingface
- Collection Method
- Benchmark execution of the MBPP test suite against the Qwen 3.5 2B model.
- Freshness
- Last updated 2026-05-24 07:35:02; freshness should be verified.