DagsHub Reviews & Product Details

DagsHub makes it easy to manage the complex data pipelines required for training and fine-tuning large language models. We use it to version instruction datasets, evaluate prompt variations, and track model performance over time. Having a single source of truth for data, experiments, and models has been a game-changer for reproducibility. The Git-like workflow for data is intuitive, and it integrates smoothly with our existing tools. Review collected by and hosted on G2.com.

Nothing major. It's working well for our needs. Some visualizations around dataset evolution could make debugging faster, but overall it’s a solid product. Review collected by and hosted on G2.com.

DagsHub lets us keep our LLM training data, experiments, and models tightly connected. We version everything—from raw datasets to tokenizer outputs and model checkpoints. This setup makes it simple to track which data was used, how it was processed, and which experiments led to which results. It’s especially helpful when testing prompt tuning or comparing different model variants. Everything stays reproducible and easy to collaborate on across teams. Review collected by and hosted on G2.com.

No major issues so far. The platform handles version control and experiment linkage really well. A bit more UI customization would be great, especially for larger projects. Review collected by and hosted on G2.com.

DagsHub simplifies versioning large text datasets, tracking fine-tuning experiments, and managing model checkpoints—all within a single platform. In any LLM workflow, connecting a specific dataset version to a model run is essential to ensure reproducibility. DagsHub also makes it easier to manage evaluation metrics across various prompt tuning and fine-tuning experiments, all while maintaining clear data lineage. Review collected by and hosted on G2.com.

Nothing major — it works very well for my personal LLM projects. While having even more built-in support for prompt datasets would be a nice addition, I find the platform to be already great for full dataset and model versioning. Review collected by and hosted on G2.com.

DagsHub is a best friend of Data Scientists and Machine Learning Engineers since it provides not only a version control repository for the code but also for the data artifacts, such as datasets and models. MLOps tools like DVC and MLflow are available for every repository and hosted on DagsHub out of the box so it's extremely easy to start using them right away! This is such a big advantage because, for example, MLflow tracks machine learning models locally by default so you need to set up an MLflow server when working in a team which isn't obvious and DagsHub is real time saver here. As a cherry on top of the cake, DagsHub offers many GBs of free storage for your data artifacts and you will definitely appreciate it if you want to try it out for your project. Overall, DagsHub is an amazing MLOps platform with many more stuff that will make your life so much easier, such as annotation tools, GitHub integration, Jupyter notebook diffs, etc. The DagsHub documentaion is just great but if you need extra help, the DagsHub team is super responsive on their Discord channel. Feel free to check out my DagsHub project where I describe in detail how I used its features for my model cloud deployment pipeline https://dagshub.com/PavloFesenko/gif_analyzer Review collected by and hosted on G2.com.

No complaints, I only wish that I have discovered DagsHub earlier. 😄 Review collected by and hosted on G2.com.

DagsHub helps us organize unstructured data—images, text, and more—into well-managed datasets. What stands out is how tightly it connects our data with experiment runs and trained model versions. This makes it easy to reproduce results, compare model performance, and trace issues back to specific data changes. It’s a solid platform for teams that care about reproducibility, collaboration, and data quality in ML projects. Review collected by and hosted on G2.com.

Nothing major to note—overall, the experience has been smooth. Minor UI improvements could make large-scale project navigation even better. Review collected by and hosted on G2.com.

DagsHub is ideal for managing large language model (LLM) training data. We use it to version curated text corpora, track data cleaning steps, and run experiments on fine-tuned models—all with full reproducibility. The ability to connect datasets directly to experiment runs and model outputs helps us stay organized and iterate quickly. It’s also helpful for collaborating across teams working on data prep, prompt engineering, and evaluation. Review collected by and hosted on G2.com.

No major issues. The platform handles LLM workflows well. A few more features around prompt versioning would be useful, but what’s there is already saving us a ton of time. Review collected by and hosted on G2.com.

DagsHub allows me to easily manage unstructured data in the context of complex models and versioning. It allows me to save a fixed state of code and data for every experiment and saves me time dealing with various tools. I have used it in several projects so far, it makes project management a lot easier. saves me time with tool integration because the internal offering is easy to implement. Review collected by and hosted on G2.com.

No major complaints, the platform has been reliable and intuitive so far. Review collected by and hosted on G2.com.

DagsHub provides seamless integration with the data version control tool of my choice, namely DVC; it can be easily used as remote repository for storing large data files, and for storing directories with large amount of files. I also like its integration with Git repository hosting sites, not only GitHub, but also other such services, like GitLab or Bitbucket.

DagsHub repository makes it possible to browse and analyze data files, regardless of whether they are versioned using Git, or using DVC. The visualization of data processing pipeline includes both stages, and outputs / data dependencies.

I have only lightly tried the experiment tracking part of DagsHub, but I like what I have seen so far. DagsHub includes support for both DVC experiments (`dvc exp`) and MLflow experiments tracking.

I have yet to try the data streaming support, or mounting DagsHub storage as S3 filesystem - but it looks like a neat feature. Review collected by and hosted on G2.com.

I haven't notice any major issues so far. The platform is robust, and caters well to our data tracking needs.

I don't like the very strict limitation of the free plan (maximum of 2 people in a team), but I can understand it. DagsHub does offer full version for academia, but it is at request, and it is not automated (using for example using Shibboleth login, like GitLab does it). Review collected by and hosted on G2.com.

DagsHub gives me full control over my machine learning data and experiments. I can version raw datasets, preprocess them, track experiments, and manage model outputs, all in one place. This tight integration means less time syncing tools and more time improving my models. It’s especially useful for unstructured data like images and documents, where traceability and collaboration are hard to maintain without the right setup. Review collected by and hosted on G2.com.

Nothing critical so far. The platform handles most of my workflows out of the box. Would be great to see more integrations with external training environments, but the current ones cover most use cases. Review collected by and hosted on G2.com.

DagsHub provides an out of the box platform to manage data, code and experiments in one place Review collected by and hosted on G2.com.

Didn't find an easy option to extend the abilities, for example, inheriting part of a model from a different project. Review collected by and hosted on G2.com.