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; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1]
DTZO: Distributed Trilevel Zeroth Order Learning with Provable Non-Asymptotic Convergence
Authors: Yang Jiao, Kai Yang, Chengtao Jian
ICML 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Furthermore, we theoretically carry out the non-asymptotic convergence rate analysis for the proposed DTZO in achieving the ϵstationary point. Extensive experiments have been conducted to demonstrate and validate the superior performance of the proposed DTZO. |
| Researcher Affiliation | Academia | 1Department of Computer Science and Technology, Tongji University. Correspondence to: Kai Yang <EMAIL>. |
| Pseudocode | Yes | Algorithm 1 DTZO: Distributed Trilevel Zeroth Order Learning |
| Open Source Code | No | The paper does not provide an explicit statement about open-sourcing the code for the methodology described, nor does it include a link to a code repository. |
| Open Datasets | Yes | The General Language Understanding Evaluation (GLUE) benchmark (Wang et al., 2018a) is used to evaluate the proposed DTZO. Specifically, the experiments are carried out on: 1) SST-2 for sentiment analysis; 2) COLA for linguistic acceptability; and 3) MRPC for semantic equivalence of sentences. ... The digits recognition tasks in Qian et al. (2019); Wang et al. (2021) with several benchmark datasets, i.e., MNIST (Le Cun et al., 1998), USPS, Fashion MNIST (Xiao et al., 2017), and QMNIST (Yadav & Bottou, 2019), are utilized to assess the performance of the proposed DTZO. In addition, DTZO is also assessed on time series datasets, including Melbourne Pedestrian, Crop, and UWave Gesture Library All, sourced from the UCR Archive (Dau et al., 2018). |
| Dataset Splits | No | The paper mentions using training and validation datasets (e.g., 'Dtr j and Dval j', 'Xtr j and ytr j', 'Xvar j and yvar j') but does not specify explicit percentages, sample counts, or reference standard splits with specific details for reproducibility. |
| Hardware Specification | Yes | In the experiment, all the models are implemented using Py Torch, and the experiments are conducted on a server equipped with two NVIDIA RTX 4090 GPUs. |
| Software Dependencies | No | The paper states that 'all the models are implemented using Py Torch' but does not specify the version number of PyTorch or any other software dependencies with their versions. |
| Experiment Setup | Yes | Table 4. Experimental details. lists specific values for hyperparameters such as ηx1, ηx2, ηx3, µ, λl, and ϕj for various datasets (e.g., SST-2, COLA, MRPC, MNIST). For example, for SST-2: ηx1=0.01, ηx2=0.001, ηx3=0.001, µ=0.001, λl=1, ϕj=0.5. |