CanItEdit

CanItEdit: Measures model capability on programming, code generation, code repair, or repository-level software tasks.

12rows
canitedit_accuracyprimary metric
2026-05-28sampled

Metadata

ID
canitedit
Category
Coding
Release
Unknown

Metrics

CanItEdit Accuracy, HumanEvalFix Accuracy, Aider Accuracy, Aider Polyglot Accuracy

Latest Results

Rows are imported from a public ICLR 2025 paper comparison table that directly reports CanItEdit accuracy for QwenCoder and NextCoder variants. This is not an official CanItEdit leaderboard.

Rank Subject CanItEdit Accuracy Model Match Provenance Sampled
1 NextCoder-32B 62.4% — Imported 2026-05-28
2 QwenCoder-2.5-32B 61% — Imported 2026-05-28
3 NextCoder-14B 60.2% — Imported 2026-05-28
4 QwenCoder-2.5-14B 58.1% — Imported 2026-05-28
5 QwenCoder-2.5-32B-LoRA 52.4% — Imported 2026-05-28
6 QwenCoder-2.5-14B-LoRA 50.9% — Imported 2026-05-28
7 QwenCoder-2.5-32B-SFT 49.5% — Imported 2026-05-28
8 NextCoder-3B 42.4% — Imported 2026-05-28
9 QwenCoder-2.5-14B-SFT 42.4% — Imported 2026-05-28
10 QwenCoder-2.5-3B 37.1% — Imported 2026-05-28
11 QwenCoder-2.5-3B-LoRA 36.2% — Imported 2026-05-28
12 QwenCoder-2.5-3B-SFT 32.4% — Imported 2026-05-28