selection-coder-100m

A 212.4M-parameter Python code model. Context length 2048.

Benchmarks (pass@1, greedy)

Benchmark	standard	reasoning
HumanEval	0.61%	0.0%
MBPP	3.5%	0.5%

reasoning = chain-of-thought prompt + 2 self-refine iterations.

### Instruction:
<your task>

### Response:

Safetensors

Model size

0.2B params

Tensor type

F16