File size: 849 Bytes
0020fe4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5bddf43
0020fe4
5bddf43
0020fe4
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
---
license: apache-2.0
base_model: laion/r2egym-nl2bash-stack-bugsseq-fixthink-again
tags:
  - reinforcement-learning
  - code
  - swesmith
  - rl
  - rloo
datasets:
  - DCAgent/swesmith-oracle-verified-120s-2500
language:
  - en
pipeline_tag: text-generation
library_name: transformers
---

# SweSmith-8B-SFT-Rope-step62

RL-trained Qwen3-8B on SWEsmith tasks (65k context with YaRN rope scaling factor=4.0, 62 steps). Best evaluated at 65k context.

## Training Details

- **Base model**: [laion/r2egym-nl2bash-stack-bugsseq-fixthink-again](https://huggingface.co/laion/r2egym-nl2bash-stack-bugsseq-fixthink-again) (Qwen3-8B SFT)
- **Training method**: RLOO-N 
- **Training data**: 2,500 SWEsmith tasks (oracle-verified, 120s timeout)
- **Framework**: BenSkyRL + Harbor 
- **Sandbox**: Apptainer containers with proxychains for internet access