license: mit
task_categories:
- text-generation
- question-answering
- image-text-to-text
language:
- en
tags:
- DeepSWE
- code
- coding
- programming
- SWE
- SWE-bench
- py
- js
- ts
- java
- cpp
- rust
- rs
- go
- reasoning
- reason
- SWE-smith
- gym
- agentic
- agent
- english
- software-engineering
- long-context
- code-generation
- repository-level
- multi-file
- fine-tuning
- sft
- benchmark
size_categories:
- 1K<n<10K
pretty_name: The best of SWE
Dataset Description
This dataset is an EXTREMELY filtered and deduplicated version of DeepSWE-Gym2-Edu, it aims to improve benchmark results on DeepSWE-style problems, benchmarks, and general coding skills.
It is specifically filtered for rows with complex/long code problems in the original datasets, having an average row size of 8.27MB, a total uncompressed size of 8.27GB, and a total of 1000 examples.
Dataset Details
- Curated by: MoreThought
- Funded by: MoreThought
- Shared by: MoreThought
- License: MIT
Dataset Sources
Repositorys:
- https://huggingface.co/datasets/MoreThought/DeepSWE-Gym
- https://huggingface.co/datasets/SWE-Gym/SWE-Gym
- https://huggingface.co/datasets/PrimeIntellect/SWE-rebench-V2-Filtered-Verified
- https://huggingface.co/datasets/nebius/SWE-bench-extra
- https://huggingface.co/datasets/TIGER-Lab/SWE-Next
- https://huggingface.co/datasets/PrimeIntellect/Multi-SWE-RL-Verified
- https://huggingface.co/datasets/swesynth/SWE-Synth
Papers:
- https://huggingface.co/papers/2504.21798
- https://huggingface.co/papers/2504.02605
- https://huggingface.co/papers/2603.20691
- https://huggingface.co/papers/2602.23866
Uses
Improving benchmark results
Improving general coding capabilities
Training software engineering/coding agents
Improving long-context coding
Important
Do NOT try to use this dataset along with other variants or even versions of it if you don't want overlapping examples.
Almost all LLMs cannot handle examples reaching up to 28MB, use specialized scripts to train properly.