Overview

This model, DCAgent/a1-defects4j, is an 8 billion parameter language model derived from the Qwen/Qwen3-8B base architecture. It has undergone a specific fine-tuning process to specialize its capabilities.

Training Details

The model was fine-tuned on a unique dataset: /e/scratch/jureap59/raoof1/sft_data/hf_hub/datasets--DCAgent--exp_rpt_defects4j-v3_10k_glm_4.7_traces_jupiter/snapshots/c2c5dde80f4ceb33fc781ade3b1285cddb53b59a_thinking_preprocessed. Key training hyperparameters included:

• Learning Rate: 4e-05
• Batch Size: 1 (train), 8 (eval)
• Optimizer: ADAMW_TORCH_FUSED with betas=(0.9,0.98) and epsilon=1e-08
• LR Scheduler: Cosine type with 0.1 warmup ratio
• Epochs: 7.0

Intended Use

While specific intended uses and limitations require more detailed information, the fine-tuning on a dataset related to "defects4j" suggests its application in areas such as software defect analysis, bug reporting, or code-related problem-solving. Users should consult further documentation for precise guidance on its optimal deployment and any known constraints.