Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The Shift Toward Open and Reproducible AI Research," Under Review, 2026. doi: 10.48550/arXiv.2606.16974..
FreeInv: Free Lunch for Improving DDIM Inversion
Authors: Yuxiang Bao, Huijie Liu, xun gao, Huan Fu, Guoliang Kang
NeurIPS 2025 | Venue PDF | LLM Run Details | Input Tokens: 22,434 Total number of tokens sent to the LLM as input for this paper's analysis. | Output Tokens: 4,295 Total number of tokens produced by the LLM (including reasoning/thinking tokens) for this paper's analysis.
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Comprehensive quantitative and qualitative evaluation on PIE benchmark and DAVIS dataset shows that Free Inv remarkably outperforms conventional DDIM inversion, and is competitive among previous state-of-the-art inversion methods, with superior computation efficiency. |
| Researcher Affiliation | Collaboration | 1 Beihang University, 2 HUJING Digital Media & Entertainment Group |
| Pseudocode | No | The paper describes the methodology in Section 3, using mathematical formulations and descriptive text, but does not include a distinct pseudocode block or algorithm figure. |
| Open Source Code | Yes | Code is available at https://github.com/yuxiangbao/ Free Inv. |
| Open Datasets | Yes | Comprehensive quantitative and qualitative evaluation on PIE benchmark and DAVIS dataset shows that Free Inv remarkably outperforms conventional DDIM inversion |
| Dataset Splits | No | Dataset. Following previous works [17, 50], we employ the PIE-benchmark [17] and its officially released code to quantitatively evaluate image editing results from Free Inv and the compared methods. PIE-benchmark consists of 700 images of resolution 512 512, the content of which is from nature or artificial generation. |
| Hardware Specification | No | Table 1 and Figure 6 provide memory usage (MB) and time (Seconds) for experiments, but they do not specify the exact GPU or CPU models used, only the amount of memory. |
| Software Dependencies | Yes | In our experiments, unless otherwise stated, we adopt Stable Diffusion [36] 1.5 with a 50-step DDIM schedule for U-Net based methods, and FLUX.1-dev [1] (abbr. FLUX) with a 25-step schedule for Di T based methods. |
| Experiment Setup | Yes | In our experiments, unless otherwise stated, we adopt Stable Diffusion [36] 1.5 with a 50-step DDIM schedule for U-Net based methods, and FLUX.1-dev [1] (abbr. FLUX) with a 25-step schedule for Di T based methods. |