FSL-Mate is two directories, a paper list with a dated changelog, and one sentence about a library
FSL-Mate: A collection of resources for few-shot learning (FSL).
At a glance
- What is it?
- A collection resource for few-shot learning that has grown into a Python library and a running reading list. Everything interesting about it is in the asymmetry between those two halves: one is dated and versioned by conference batch, the other is described in a single line with no install path at all.
- Who is it for?
- Treat FSL-Mate as a pointer rather than a toolkit. FewShotPapers is worth bookmarking if you want the field's reading list in one place and can tolerate the gaps, which are visible in the changelog itself, and PaddleFSL is worth opening only if PaddlePaddle is already your stack, since the repository says nothing about installing it.
- Can I use it commercially?
- Not without permission. GitHub finds no licence file in the repository, and without a licence all rights are reserved by default: you may read the code but not reuse it. Check the README, or ask the authors, before using it.
- Is it still maintained?
- Yes. The repository last received commits 175 days ago.
- What is it written in?
- Mainly Python, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on October 4, 2026, and from our analysis. They are not legal advice.
Editorial analysis
Two directories, a README and a logo file
The whole repository is five top-level entries: a .gitignore, FewShotPapers/, PaddleFSL/, README.md and logo-fsl-mate.png.
The two directories are the project. FewShotPapers is described as a paper list tracking research advances in few-shot learning, and PaddleFSL as a PaddlePaddle-based Python library for the same field. There is no packaging file at the root, no requirements file, no notebook, and no .github directory, which means the tree contains no workflow files, no issue templates and nothing that could add a paper automatically.
The default branch is master rather than main, and every internal link in the README is written against tree/master. For a repository that mostly links to its own subdirectories, that is a small thing to notice only because the branch name is the kind of detail that quietly breaks older tooling.
So this is a curated collection with a library tucked beside it, not a library project with a reading list attached. Which of the two is the centre is decided by which directory a reader opens.
The list is versioned by conference batch, and the newest batch is narrower than the one before it
The changelog is three dated entries, each naming the venues covered.
The entry dated 2026-04-18 adds papers from EMNLP 2025, ICCV 2025, ICML 2025, WWW 2026 and AAAI 2026. The entry dated 2025-10-14 adds CVPR 2025, ACL 2025 and IJCAI 2025. The entry dated 2025-07-17 adds a much longer run: CVPR 2024, ICML 2024, IJCAI 2024, ACL 2024, NeurIPS 2024, EMNLP 2024, ICCV 2024, ICLR 2025, plus the double-year venues WWW, KDD, AAAI, NAACL and SIGIR for 2024 and 2025.
Two things fall out of that comparison. The 2026 batch includes ICML 2025 while leaving out NeurIPS 2025, even though the 2025 batch included NeurIPS 2024. And the cadence is irregular, roughly three months between the first two entries and roughly six between the second and the third.
The list presents itself as tracking research advances, and a venue that was in the last batch and is missing from the next one is a hole a reader can see without any tooling. The last recorded change to the repository is dated 2026-04-18, the same day as the newest changelog line.
PaddleFSL gets one sentence and no install path
The library half of the repository is introduced in a single line and then never mentioned again. The README says it is a PaddlePaddle-based Python library for few-shot learning, and the directory is linked, and that is the entirety of the description.
There is no install command, no package name on an index, no version, no list of what is inside the directory, no example of importing anything, and no statement of which PaddlePaddle release it expects. A reader who wants to try it has to open the directory and read the source, which is a reasonable thing to want and not something the repository offers.
The repository's primary language is Python, which is consistent with the library and tells you nothing about the list. The list itself is presumably plain text or Markdown, and the README contains no Python at all: no snippet, no call, no configuration. That is normal for a reading list and it is the reason the library half is so much harder to evaluate from the front page than the list half is.
Cite Us points at a 2020 survey, and the contact address belongs to one of its authors
There is a citation request, and what it asks you to cite is not work from this repository. It is a survey: Generalizing from a few examples, a survey on few-shot learning, by Wang, Yao, Kwok and Ni, published in ACM Computing Surveys, volume 53, number 3, pages 1 to 34, in 2020. The repository supplies a BibTeX entry for it, with the key naming the first author and the year, and links to the ACM Digital Library record.
That is a normal convention for a community reading list. It is also the reason this resource is coherent: the person asking you to cite the canonical few-shot learning survey is one of its authors.
The contact section names Yaqing Wang with a mail address at an institutional domain ending in bimsa.cn, alongside the invitation to open an issue on the repository. One person, one address, one place to send feedback, and a paper list that they update themselves.
There is no licence file and no declared licence
Nothing in the tree grants permission to do anything with it. There is no LICENSE file among the five top-level entries, and the repository carries no declared licence.
That is worth pausing on for this particular project. One directory is a curated list of other people's research papers, and the other is a library. Those two things usually get different treatment: a reading list of citations raises no expectations, while a library raises an expectation about who may use it, modify it and redistribute it. This repository does not say which way it lands.
The same gap shows up in how the library would be obtained. There are no releases, so there is no packaged artefact and no version history to point at. Anyone using PaddleFSL is building from the default branch at whatever commit they happened to clone, with nothing recorded about which commit that was.
None of this makes the resource less useful. It does mean the questions a reader usually settles by reading a licence file have to be settled by asking, which for a list of papers and a research library is an unfortunate amount of friction.
The promise to update constantly sits above three dates
Between the description and the changelog there is one sentence: the maintainers say they are endeavored to constantly update FSL-Mate, and hope it can make few-shot learning easier.
The record underneath it is three dates. The list was extended in July 2025, again in October 2025, and again in April 2026, with the newest batch covering five venues and the one before it covering three. So the update is real and it is periodic rather than constant, and the newest batch is the smallest of the three when measured in venues named.
For a reading list that is a reasonable shape. Conference cycles do not line up, and the effort of curating a batch of a dozen venues is not the same as curating three. The point is only that a reader deciding whether to rely on this list should read the changelog rather than the sentence above it, because the changelog is where the gaps and the cadence actually are.
The last recorded change to the repository is dated 2026-04-18. Between then and the rest of 2026 there may be more to add, and nothing in the repository says there will be.
Editorial conclusion
Treat FSL-Mate as a pointer rather than a toolkit. FewShotPapers is worth bookmarking if you want the field's reading list in one place and can tolerate the gaps, which are visible in the changelog itself, and PaddleFSL is worth opening only if PaddlePaddle is already your stack, since the repository says nothing about installing it. Two things to check before you build on either half: there is no licence file and no declared licence, which is unusual for something holding both a curated list and a library, and there are no releases, so a library taken from the default branch has no version to record. If you need a citable few-shot learning reference, the survey this project asks you to cite is a better anchor than the repository itself.
Frequently asked questions
What does the tata1661/FSL-Mate repository contain?
Two directories. FewShotPapers is a running paper list tracking research advances in few-shot learning, and PaddleFSL is a PaddlePaddle-based Python library for the same field. The root holds only a README, a .gitignore and a logo image.
How often are new papers added to FSL-Mate?
The changelog records three batches: 2025-07-17, 2025-10-14 and 2026-04-18. Each names the venues covered, from a long run of conferences in the first to five venues in the most recent. The last recorded change to the repository is dated 2026-04-18.
How do I install PaddleFSL from FSL-Mate?
The repository does not say. PaddleFSL is described in one line as a PaddlePaddle-based Python library for few-shot learning, with no install command, no package name on an index, no version and no expected PaddlePaddle release. There is also no release published for it.
What should I cite if I use FSL-Mate?
The repository asks you to cite a 2020 survey, Generalizing from a few examples, by Wang, Yao, Kwok and Ni, in ACM Computing Surveys volume 53 number 3, pages 1 to 34. It supplies a BibTeX entry and links to the ACM Digital Library record.
Is FSL-Mate licensed?
No licence is declared and no LICENSE file appears among the top-level entries, which hold only a .gitignore, the two directories, the README and a logo image. Nothing in the tree states what may be done with the paper list or with the library.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/tata1661-fsl-mate)