docs: adoption-focused README, WHY_FROSSL, and reproducible CIFAR-10 comparison#2
Open
jacobsn wants to merge 4 commits into
Open
docs: adoption-focused README, WHY_FROSSL, and reproducible CIFAR-10 comparison#2jacobsn wants to merge 4 commits into
jacobsn wants to merge 4 commits into
Conversation
Reorganize the README to lead with the scientific contribution and practical advantages (elevator pitch, "Why use FroSSL?", comparison table, efficiency figure, method diagram) before installation. Add a prominent Quick Start, custom-dataset example, results takeaway, pretrained-model placeholder, and badges. Add docs/WHY_FROSSL.md with a deeper comparison against SimCLR, BYOL, VICReg, Barlow Twins, and W-MSE, including the math intuition and empirical evidence, so the README can stay concise. Co-authored-by: Cursor <cursoragent@cursor.com>
…rison - Add "Available in popular SSL libraries" section linking the in-progress lightly and solo-learn integration PRs. - Add a reproducible CIFAR-10 kNN comparison (FroSSL vs 7 SSL methods, same GPU and identical protocol from lightly's benchmark): table + convergence curve. FroSSL is highest accuracy, lowest peak memory, fastest-tier runtime.
Author
|
Not sure we want all of that added to the README, but it's a starting point. Let me know what you'd like to change. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR brings the documentation improvements from the MVRL fork back to the source repo. It is a docs-only change — no code or training behavior is modified.
What's included
README rewrite for adoption — clearer positioning (why/when to use FroSSL), a "how does it compare?" table, and sections on faster convergence and multiview.
docs/WHY_FROSSL.md— a deeper comparison of the math, tradeoffs, and evidence vs SimCLR, BYOL, VICReg, and Barlow Twins.Library integrations — links to the in-progress FroSSL integrations in lightly and solo-learn, so users can reach FroSSL from familiar APIs.
Reproducible CIFAR-10 comparison (
experiments/lightly_cifar10/) — FroSSL vs seven common SSL methods run through lightly's CIFAR-10 kNN benchmark under an identical protocol on the same GPU (ResNet-18, batch 512, 200 epochs, kNN k=200):FroSSL reaches the highest kNN accuracy at the lowest peak memory and in the fastest tier of runtimes, and matches every other method's final 200-epoch accuracy in fewer epochs. Single seed; runtime/memory comparable within this same-GPU run, kNN accuracy is hardware-independent.
Notes