Upload TexJEPA trained checkpoints and model card

LICENSE

@@ -0,0 +1,399 @@
+Attribution-NonCommercial 4.0 International
+
+=======================================================================
+
+Creative Commons Corporation ("Creative Commons") is not a law firm and
+does not provide legal services or legal advice. Distribution of
+Creative Commons public licenses does not create a lawyer-client or
+other relationship. Creative Commons makes its licenses and related
+information available on an "as-is" basis. Creative Commons gives no
+warranties regarding its licenses, any material licensed under their
+terms and conditions, or any related information. Creative Commons
+disclaims all liability for damages resulting from their use to the
+fullest extent possible.
+
+Using Creative Commons Public Licenses
+
+Creative Commons public licenses provide a standard set of terms and
+conditions that creators and other rights holders may use to share
+original works of authorship and other material subject to copyright
+and certain other rights specified in the public license below. The
+following considerations are for informational purposes only, are not
+exhaustive, and do not form part of our licenses.
+
+     Considerations for licensors: Our public licenses are
+     intended for use by those authorized to give the public
+     permission to use material in ways otherwise restricted by
+     copyright and certain other rights. Our licenses are
+     irrevocable. Licensors should read and understand the terms
+     and conditions of the license they choose before applying it.
+     Licensors should also secure all rights necessary before
+     applying our licenses so that the public can reuse the
+     material as expected. Licensors should clearly mark any
+     material not subject to the license. This includes other CC-
+     licensed material, or material used under an exception or
+     limitation to copyright. More considerations for licensors:
+	wiki.creativecommons.org/Considerations_for_licensors
+
+     Considerations for the public: By using one of our public
+     licenses, a licensor grants the public permission to use the
+     licensed material under specified terms and conditions. If
+     the licensor's permission is not necessary for any reason--for
+     example, because of any applicable exception or limitation to
+     copyright--then that use is not regulated by the license. Our
+     licenses grant only permissions under copyright and certain
+     other rights that a licensor has authority to grant. Use of
+     the licensed material may still be restricted for other
+     reasons, including because others have copyright or other
+     rights in the material. A licensor may make special requests,
+     such as asking that all changes be marked or described.
+     Although not required by our licenses, you are encouraged to
+     respect those requests where reasonable. More_considerations
+     for the public:
+	wiki.creativecommons.org/Considerations_for_licensees
+
+=======================================================================
+
+Creative Commons Attribution-NonCommercial 4.0 International Public
+License
+
+By exercising the Licensed Rights (defined below), You accept and agree
+to be bound by the terms and conditions of this Creative Commons
+Attribution-NonCommercial 4.0 International Public License ("Public
+License"). To the extent this Public License may be interpreted as a
+contract, You are granted the Licensed Rights in consideration of Your
+acceptance of these terms and conditions, and the Licensor grants You
+such rights in consideration of benefits the Licensor receives from
+making the Licensed Material available under these terms and
+conditions.
+
+Section 1 -- Definitions.
+
+  a. Adapted Material means material subject to Copyright and Similar
+     Rights that is derived from or based upon the Licensed Material
+     and in which the Licensed Material is translated, altered,
+     arranged, transformed, or otherwise modified in a manner requiring
+     permission under the Copyright and Similar Rights held by the
+     Licensor. For purposes of this Public License, where the Licensed
+     Material is a musical work, performance, or sound recording,
+     Adapted Material is always produced where the Licensed Material is
+     synched in timed relation with a moving image.
+
+  b. Adapter's License means the license You apply to Your Copyright
+     and Similar Rights in Your contributions to Adapted Material in
+     accordance with the terms and conditions of this Public License.
+
+  c. Copyright and Similar Rights means copyright and/or similar rights
+     closely related to copyright including, without limitation,
+     performance, broadcast, sound recording, and Sui Generis Database
+     Rights, without regard to how the rights are labeled or
+     categorized. For purposes of this Public License, the rights
+     specified in Section 2(b)(1)-(2) are not Copyright and Similar
+     Rights.
+  d. Effective Technological Measures means those measures that, in the
+     absence of proper authority, may not be circumvented under laws
+     fulfilling obligations under Article 11 of the WIPO Copyright
+     Treaty adopted on December 20, 1996, and/or similar international
+     agreements.
+
+  e. Exceptions and Limitations means fair use, fair dealing, and/or
+     any other exception or limitation to Copyright and Similar Rights
+     that applies to Your use of the Licensed Material.
+
+  f. Licensed Material means the artistic or literary work, database,
+     or other material to which the Licensor applied this Public
+     License.
+
+  g. Licensed Rights means the rights granted to You subject to the
+     terms and conditions of this Public License, which are limited to
+     all Copyright and Similar Rights that apply to Your use of the
+     Licensed Material and that the Licensor has authority to license.
+
+  h. Licensor means the individual(s) or entity(ies) granting rights
+     under this Public License.
+
+  i. NonCommercial means not primarily intended for or directed towards
+     commercial advantage or monetary compensation. For purposes of
+     this Public License, the exchange of the Licensed Material for
+     other material subject to Copyright and Similar Rights by digital
+     file-sharing or similar means is NonCommercial provided there is
+     no payment of monetary compensation in connection with the
+     exchange.
+
+  j. Share means to provide material to the public by any means or
+     process that requires permission under the Licensed Rights, such
+     as reproduction, public display, public performance, distribution,
+     dissemination, communication, or importation, and to make material
+     available to the public including in ways that members of the
+     public may access the material from a place and at a time
+     individually chosen by them.
+
+  k. Sui Generis Database Rights means rights other than copyright
+     resulting from Directive 96/9/EC of the European Parliament and of
+     the Council of 11 March 1996 on the legal protection of databases,
+     as amended and/or succeeded, as well as other essentially
+     equivalent rights anywhere in the world.
+
+  l. You means the individual or entity exercising the Licensed Rights
+     under this Public License. Your has a corresponding meaning.
+
+Section 2 -- Scope.
+
+  a. License grant.
+
+       1. Subject to the terms and conditions of this Public License,
+          the Licensor hereby grants You a worldwide, royalty-free,
+          non-sublicensable, non-exclusive, irrevocable license to
+          exercise the Licensed Rights in the Licensed Material to:
+
+            a. reproduce and Share the Licensed Material, in whole or
+               in part, for NonCommercial purposes only; and
+
+            b. produce, reproduce, and Share Adapted Material for
+               NonCommercial purposes only.
+
+       2. Exceptions and Limitations. For the avoidance of doubt, where
+          Exceptions and Limitations apply to Your use, this Public
+          License does not apply, and You do not need to comply with
+          its terms and conditions.
+
+       3. Term. The term of this Public License is specified in Section
+          6(a).
+
+       4. Media and formats; technical modifications allowed. The
+          Licensor authorizes You to exercise the Licensed Rights in
+          all media and formats whether now known or hereafter created,
+          and to make technical modifications necessary to do so. The
+          Licensor waives and/or agrees not to assert any right or
+          authority to forbid You from making technical modifications
+          necessary to exercise the Licensed Rights, including
+          technical modifications necessary to circumvent Effective
+          Technological Measures. For purposes of this Public License,
+          simply making modifications authorized by this Section 2(a)
+          (4) never produces Adapted Material.
+
+       5. Downstream recipients.
+
+            a. Offer from the Licensor -- Licensed Material. Every
+               recipient of the Licensed Material automatically
+               receives an offer from the Licensor to exercise the
+               Licensed Rights under the terms and conditions of this
+               Public License.
+
+            b. No downstream restrictions. You may not offer or impose
+               any additional or different terms or conditions on, or
+               apply any Effective Technological Measures to, the
+               Licensed Material if doing so restricts exercise of the
+               Licensed Rights by any recipient of the Licensed
+               Material.
+
+       6. No endorsement. Nothing in this Public License constitutes or
+          may be construed as permission to assert or imply that You
+          are, or that Your use of the Licensed Material is, connected
+          with, or sponsored, endorsed, or granted official status by,
+          the Licensor or others designated to receive attribution as
+          provided in Section 3(a)(1)(A)(i).
+
+  b. Other rights.
+
+       1. Moral rights, such as the right of integrity, are not
+          licensed under this Public License, nor are publicity,
+          privacy, and/or other similar personality rights; however, to
+          the extent possible, the Licensor waives and/or agrees not to
+          assert any such rights held by the Licensor to the limited
+          extent necessary to allow You to exercise the Licensed
+          Rights, but not otherwise.
+
+       2. Patent and trademark rights are not licensed under this
+          Public License.
+
+       3. To the extent possible, the Licensor waives any right to
+          collect royalties from You for the exercise of the Licensed
+          Rights, whether directly or through a collecting society
+          under any voluntary or waivable statutory or compulsory
+          licensing scheme. In all other cases the Licensor expressly
+          reserves any right to collect such royalties, including when
+          the Licensed Material is used other than for NonCommercial
+          purposes.
+
+Section 3 -- License Conditions.
+
+Your exercise of the Licensed Rights is expressly made subject to the
+following conditions.
+
+  a. Attribution.
+
+       1. If You Share the Licensed Material (including in modified
+          form), You must:
+
+            a. retain the following if it is supplied by the Licensor
+               with the Licensed Material:
+
+                 i. identification of the creator(s) of the Licensed
+                    Material and any others designated to receive
+                    attribution, in any reasonable manner requested by
+                    the Licensor (including by pseudonym if
+                    designated);
+
+                ii. a copyright notice;
+
+               iii. a notice that refers to this Public License;
+
+                iv. a notice that refers to the disclaimer of
+                    warranties;
+
+                 v. a URI or hyperlink to the Licensed Material to the
+                    extent reasonably practicable;
+
+            b. indicate if You modified the Licensed Material and
+               retain an indication of any previous modifications; and
+
+            c. indicate the Licensed Material is licensed under this
+               Public License, and include the text of, or the URI or
+               hyperlink to, this Public License.
+
+       2. You may satisfy the conditions in Section 3(a)(1) in any
+          reasonable manner based on the medium, means, and context in
+          which You Share the Licensed Material. For example, it may be
+          reasonable to satisfy the conditions by providing a URI or
+          hyperlink to a resource that includes the required
+          information.
+
+       3. If requested by the Licensor, You must remove any of the
+          information required by Section 3(a)(1)(A) to the extent
+          reasonably practicable.
+
+       4. If You Share Adapted Material You produce, the Adapter's
+          License You apply must not prevent recipients of the Adapted
+          Material from complying with this Public License.
+
+Section 4 -- Sui Generis Database Rights.
+
+Where the Licensed Rights include Sui Generis Database Rights that
+apply to Your use of the Licensed Material:
+
+  a. for the avoidance of doubt, Section 2(a)(1) grants You the right
+     to extract, reuse, reproduce, and Share all or a substantial
+     portion of the contents of the database for NonCommercial purposes
+     only;
+
+  b. if You include all or a substantial portion of the database
+     contents in a database in which You have Sui Generis Database
+     Rights, then the database in which You have Sui Generis Database
+     Rights (but not its individual contents) is Adapted Material; and
+
+  c. You must comply with the conditions in Section 3(a) if You Share
+     all or a substantial portion of the contents of the database.
+
+For the avoidance of doubt, this Section 4 supplements and does not
+replace Your obligations under this Public License where the Licensed
+Rights include other Copyright and Similar Rights.
+
+Section 5 -- Disclaimer of Warranties and Limitation of Liability.
+
+  a. UNLESS OTHERWISE SEPARATELY UNDERTAKEN BY THE LICENSOR, TO THE
+     EXTENT POSSIBLE, THE LICENSOR OFFERS THE LICENSED MATERIAL AS-IS
+     AND AS-AVAILABLE, AND MAKES NO REPRESENTATIONS OR WARRANTIES OF
+     ANY KIND CONCERNING THE LICENSED MATERIAL, WHETHER EXPRESS,
+     IMPLIED, STATUTORY, OR OTHER. THIS INCLUDES, WITHOUT LIMITATION,
+     WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR
+     PURPOSE, NON-INFRINGEMENT, ABSENCE OF LATENT OR OTHER DEFECTS,
+     ACCURACY, OR THE PRESENCE OR ABSENCE OF ERRORS, WHETHER OR NOT
+     KNOWN OR DISCOVERABLE. WHERE DISCLAIMERS OF WARRANTIES ARE NOT
+     ALLOWED IN FULL OR IN PART, THIS DISCLAIMER MAY NOT APPLY TO YOU.
+
+  b. TO THE EXTENT POSSIBLE, IN NO EVENT WILL THE LICENSOR BE LIABLE
+     TO YOU ON ANY LEGAL THEORY (INCLUDING, WITHOUT LIMITATION,
+     NEGLIGENCE) OR OTHERWISE FOR ANY DIRECT, SPECIAL, INDIRECT,
+     INCIDENTAL, CONSEQUENTIAL, PUNITIVE, EXEMPLARY, OR OTHER LOSSES,
+     COSTS, EXPENSES, OR DAMAGES ARISING OUT OF THIS PUBLIC LICENSE OR
+     USE OF THE LICENSED MATERIAL, EVEN IF THE LICENSOR HAS BEEN
+     ADVISED OF THE POSSIBILITY OF SUCH LOSSES, COSTS, EXPENSES, OR
+     DAMAGES. WHERE A LIMITATION OF LIABILITY IS NOT ALLOWED IN FULL OR
+     IN PART, THIS LIMITATION MAY NOT APPLY TO YOU.
+
+  c. The disclaimer of warranties and limitation of liability provided
+     above shall be interpreted in a manner that, to the extent
+     possible, most closely approximates an absolute disclaimer and
+     waiver of all liability.
+
+Section 6 -- Term and Termination.
+
+  a. This Public License applies for the term of the Copyright and
+     Similar Rights licensed here. However, if You fail to comply with
+     this Public License, then Your rights under this Public License
+     terminate automatically.
+
+  b. Where Your right to use the Licensed Material has terminated under
+     Section 6(a), it reinstates:
+
+       1. automatically as of the date the violation is cured, provided
+          it is cured within 30 days of Your discovery of the
+          violation; or
+
+       2. upon express reinstatement by the Licensor.
+
+     For the avoidance of doubt, this Section 6(b) does not affect any
+     right the Licensor may have to seek remedies for Your violations
+     of this Public License.
+
+  c. For the avoidance of doubt, the Licensor may also offer the
+     Licensed Material under separate terms or conditions or stop
+     distributing the Licensed Material at any time; however, doing so
+     will not terminate this Public License.
+
+  d. Sections 1, 5, 6, 7, and 8 survive termination of this Public
+     License.
+
+Section 7 -- Other Terms and Conditions.
+
+  a. The Licensor shall not be bound by any additional or different
+     terms or conditions communicated by You unless expressly agreed.
+
+  b. Any arrangements, understandings, or agreements regarding the
+     Licensed Material not stated herein are separate from and
+     independent of the terms and conditions of this Public License.
+
+Section 8 -- Interpretation.
+
+  a. For the avoidance of doubt, this Public License does not, and
+     shall not be interpreted to, reduce, limit, restrict, or impose
+     conditions on any use of the Licensed Material that could lawfully
+     be made without permission under this Public License.
+
+  b. To the extent possible, if any provision of this Public License is
+     deemed unenforceable, it shall be automatically reformed to the
+     minimum extent necessary to make it enforceable. If the provision
+     cannot be reformed, it shall be severed from this Public License
+     without affecting the enforceability of the remaining terms and
+     conditions.
+
+  c. No term or condition of this Public License will be waived and no
+     failure to comply consented to unless expressly agreed to by the
+     Licensor.
+
+  d. Nothing in this Public License constitutes or may be interpreted
+     as a limitation upon, or waiver of, any privileges and immunities
+     that apply to the Licensor or You, including from the legal
+     processes of any jurisdiction or authority.
+
+=======================================================================
+
+Creative Commons is not a party to its public
+licenses. Notwithstanding, Creative Commons may elect to apply one of
+its public licenses to material it publishes and in those instances
+will be considered the “Licensor.” The text of the Creative Commons
+public licenses is dedicated to the public domain under the CC0 Public
+Domain Dedication. Except for the limited purpose of indicating that
+material is shared under a Creative Commons public license or as
+otherwise permitted by the Creative Commons policies published at
+creativecommons.org/policies, Creative Commons does not authorize the
+use of the trademark "Creative Commons" or any other trademark or logo
+of Creative Commons without its prior written consent including,
+without limitation, in connection with any unauthorized modifications
+to any of its public licenses or any other arrangements,
+understandings, or agreements concerning use of licensed material. For
+the avoidance of doubt, this paragraph does not form part of the
+public licenses.
+
+Creative Commons may be contacted at creativecommons.org.

README.md

@@ -0,0 +1,153 @@
+---
+license: cc-by-nc-4.0
+library_name: pytorch
+pipeline_tag: image-feature-extraction
+tags:
+  - chest-xray
+  - medical-imaging
+  - self-supervised-learning
+  - ijepa
+  - jepa
+  - vision-transformer
+  - pytorch
+  - mimic-cxr
+datasets:
+  - MIMIC-CXR-JPG
+  - VinBigData-CXR
+---
+
+# TexJEPA Model Weights
+
+This repository hosts the trained model weights for **TexJEPA** from:
+
+**When Texture Becomes the World: Texture-aware JEPA for Chest X-ray Representation Learning**
+
+Authors: **Yi Zhao\***, **Ruilang Wang\***, Bowen Liu, Donglong Chen†<br>
+\* Equal contribution. † Corresponding author.
+
+Code repository: [YiZhao-Jasper/TexJEPA](https://github.com/YiZhao-Jasper/TexJEPA)
+
+## Summary
+
+TexJEPA is a texture-aware extension of Image Joint-Embedding Predictive Architecture (I-JEPA) for chest X-ray representation learning. The models are trained on MIMIC-CXR-JPG with a ViT-H/14 backbone and an I-JEPA-style context-target predictive objective. The released weights are intended to reproduce the paper's pre-training lineage and downstream evaluation using the public TexJEPA codebase.
+
+The repository includes four checkpoints:
+
+| Model | File | Initialization | Training stage | Main purpose |
+| --- | --- | --- | --- | --- |
+| **I-JEPA-300** | `weights/I-JEPA-300/jepa-latest.pth.tar` | I-JEPA ViT-H/14 pre-training run | 300-epoch final/latest checkpoint | Baseline chest X-ray I-JEPA representation |
+| **TexJEPA-N** | `weights/TexJEPA-N/jepa-latest.pth.tar` | I-JEPA-300 lineage | +50 epoch texture-noise specialization | Learn invariance to nuisance texture corruption |
+| **TexJEPA-R** | `weights/TexJEPA-R/jepa-latest.pth.tar` | TexJEPA-N | +50 epoch register-token branch | Route high-norm texture artifacts through register tokens |
+| **TexJEPA-C** | `weights/TexJEPA-C/jepa-latest.pth.tar` | TexJEPA-N | +50 epoch covariance-regularized branch | Preserve local patch diversity and lesion-scale texture sensitivity |
+
+The historical internal experiment names are not used in this release. Please cite and report the public names above.
+
+## Files
+
+```text
+weights/
+  I-JEPA-300/jepa-latest.pth.tar
+  TexJEPA-N/jepa-latest.pth.tar
+  TexJEPA-R/jepa-latest.pth.tar
+  TexJEPA-C/jepa-latest.pth.tar
+configs/pretrain/
+  ijepa_300.yaml
+  texjepa_n.yaml
+  texjepa_r.yaml
+  texjepa_c.yaml
+checksums.sha256
+requirements.txt
+LICENSE
+```
+
+`checksums.sha256` contains SHA-256 hashes for the uploaded checkpoint files.
+
+## Model Differences
+
+### I-JEPA-300
+
+I-JEPA-300 is the chest X-ray baseline. It uses the original I-JEPA multi-block masking objective with a ViT-H/14 context encoder, EMA target encoder, and predictor. It is the reference model for clean representation quality and texture-sensitivity analysis.
+
+### TexJEPA-N
+
+TexJEPA-N adds **context-target asymmetric texture corruption**. The target branch receives clean chest radiographs, while the context branch receives mild stochastic texture perturbations such as Gaussian noise, Poisson noise, and JPEG-like compression. This trains the model to predict clean latent targets from corrupted context views, directly targeting texture robustness.
+
+### TexJEPA-R
+
+TexJEPA-R starts from TexJEPA-N and adds **register tokens** to the encoder. Register tokens participate in self-attention but are stripped before patch outputs are returned, preserving compatibility with the I-JEPA predictor. The branch also uses tighter local masking to encourage stronger local, lesion-scale structure modeling.
+
+### TexJEPA-C
+
+TexJEPA-C starts from TexJEPA-N and adds a **patch-token variance/covariance auxiliary loss**. The regularizer discourages patch-token collapse and helps preserve local texture diversity. The implementation computes the auxiliary loss per rank and avoids cross-rank all-gather, which is important because I-JEPA masks can produce variable visible-token counts across distributed workers.
+
+## Intended Use
+
+These weights are intended for:
+
+- Reproducing TexJEPA pre-training and downstream experiments with the GitHub repository.
+- Chest X-ray representation learning research.
+- Linear probing and fine-tuning on VinBigData/VinDr-style multi-label chest X-ray classification.
+- Texture robustness, lesion sensitivity, and model comparison experiments.
+
+They are **not** intended for clinical deployment or diagnostic use.
+
+## Reproduction
+
+Clone the code:
+
+```bash
+git clone https://github.com/YiZhao-Jasper/TexJEPA.git
+cd TexJEPA
+python -m pip install -r requirements.txt
+```
+
+Download this model repository and place or symlink the checkpoints under the paths expected by the configs:
+
+```text
+logs/ijepa_300/jepa-latest.pth.tar             <- weights/I-JEPA-300/jepa-latest.pth.tar
+logs/texjepa_n/jepa-latest.pth.tar             <- weights/TexJEPA-N/jepa-latest.pth.tar
+logs/texjepa_r/jepa-latest.pth.tar             <- weights/TexJEPA-R/jepa-latest.pth.tar
+logs/texjepa_c/jepa-latest.pth.tar             <- weights/TexJEPA-C/jepa-latest.pth.tar
+```
+
+The pre-training configs in this model repository mirror the GitHub release:
+
+```bash
+python scripts/sanity_check.py
+```
+
+For downstream evaluation:
+
+```bash
+export CHECKPOINT=logs/texjepa_n/jepa-latest.pth.tar
+export VINBIG_IMAGE_DIR=data/vinbig/images_1024/train
+export VINBIG_CSV=data/vinbig/annotations/train.csv
+bash downstream/run_downstream.sh all
+```
+
+## Data
+
+Pre-training uses **MIMIC-CXR-JPG**. This model repository does not redistribute MIMIC-CXR data. Researchers must obtain MIMIC-CXR-JPG through the official PhysioNet credentialing and data-use process.
+
+Downstream evaluation code supports VinBigData/VinDr-style multi-label chest X-ray annotations; those datasets are not redistributed here.
+
+## Checkpoint Format
+
+The checkpoints are PyTorch `.pth.tar` training checkpoints containing encoder, predictor, target encoder, optimizer metadata, and training metadata where available. Load only checkpoints from trusted sources because PyTorch checkpoint loading relies on Python pickle serialization.
+
+For representation extraction, the downstream code in the GitHub repository loads the `target_encoder` weights and supports ViT-H/14, ViT-L/14, and register-token variants by reading checkpoint metadata and tensor shapes.
+
+## Citation
+
+```bibtex
+@misc{zhao2026texjepa,
+  title  = {When Texture Becomes the World: Texture-aware JEPA for Chest X-ray Representation Learning},
+  author = {Zhao, Yi and Wang, Ruilang and Liu, Bowen and Chen, Donglong},
+  year   = {2026},
+  note   = {TexJEPA model weights}
+}
+```
+
+## License
+
+The released code and model weights are provided for non-commercial research use under the license included in this repository. MIMIC-CXR-JPG and downstream datasets remain governed by their own licenses and data-use agreements.

checksums.sha256

@@ -0,0 +1,4 @@
+5fbdd329af7e996e666ef361a8cf10512aa9788f04d0668e654442bd0e4e59b1  weights/I-JEPA-300/jepa-latest.pth.tar
+04216e3e1a44da04553c0e816edb1ca32034dd9e8ba5ea5169efe1ef4abda602  weights/TexJEPA-N/jepa-latest.pth.tar
+99dfe8992a49ad1a310e8eed331e5c2d1464bcaecbd43774a0b38e8d5f1117f4  weights/TexJEPA-R/jepa-latest.pth.tar
+7b3ff784a334c331313effb20ae9c76126e695f0d59a63e1e4ccf2049fff2239  weights/TexJEPA-C/jepa-latest.pth.tar

configs/pretrain/ijepa_300.yaml

@@ -0,0 +1,57 @@
+# I-JEPA-300 baseline pre-training for TexJEPA.
+#
+# This is the public 300-epoch baseline checkpoint lineage used before any
+# TexJEPA specialization. It follows the original I-JEPA joint-embedding
+# predictive objective with a ViT-H/14 context encoder, EMA target encoder,
+# 12-layer predictor, and multi-block masking on MIMIC-CXR-JPG chest radiographs.
+
+data:
+  dataset_type: mimic_cxr
+  root_path: data/mimic-cxr-384
+  use_zip: false
+  splits: null
+  batch_size: 128
+  crop_size: 224
+  crop_scale: [0.3, 1.0]
+  num_workers: 10
+  pin_mem: true
+  normalization: chest_xray
+  use_color_distortion: false
+  use_gaussian_blur: false
+  use_horizontal_flip: false
+  color_jitter_strength: 0.0
+
+logging:
+  folder: logs/ijepa_300
+  write_tag: jepa
+
+mask:
+  allow_overlap: false
+  aspect_ratio: [0.75, 1.5]
+  enc_mask_scale: [0.85, 1.0]
+  min_keep: 10
+  num_enc_masks: 1
+  num_pred_masks: 4
+  patch_size: 14
+  pred_mask_scale: [0.15, 0.2]
+
+meta:
+  load_checkpoint: false
+  model_name: vit_huge
+  patch_size: 14
+  pred_depth: 12
+  pred_emb_dim: 384
+  read_checkpoint: null
+  use_bfloat16: true
+  num_registers: 0
+
+optimization:
+  ema: [0.998, 1.0]
+  epochs: 300
+  final_lr: 1.0e-06
+  final_weight_decay: 0.4
+  ipe_scale: 1.25
+  lr: 2.0e-04
+  start_lr: 1.0e-05
+  warmup: 30
+  weight_decay: 0.04

configs/pretrain/texjepa_c.yaml

@@ -0,0 +1,72 @@
+# TexJEPA-C: covariance-regularized texture consistency.
+#
+# Warm-start from TexJEPA-N and train for 50 additional epochs. TexJEPA-C keeps
+# the TexJEPA-N asymmetric context corruption objective and adds a mild
+# patch-level variance/covariance auxiliary loss. The auxiliary term discourages
+# patch-token collapse and preserves diversity in local chest texture features.
+
+data:
+  dataset_type: mimic_cxr
+  root_path: data/mimic-cxr-384
+  use_zip: false
+  splits: null
+  batch_size: 128
+  crop_size: 224
+  crop_scale: [0.3, 1.0]
+  num_workers: 10
+  pin_mem: true
+  normalization: chest_xray
+  use_color_distortion: false
+  use_gaussian_blur: false
+  use_horizontal_flip: false
+  color_jitter_strength: 0.0
+  context_noise:
+    enabled: true
+    p_gauss: 0.5
+    p_poisson: 0.3
+    p_jpeg: 0.2
+    sigma_range: [0.005, 0.05]
+    peak_range: [50, 200]
+    quality_range: [40, 80]
+    warmup_epochs: 5
+
+logging:
+  folder: logs/texjepa_c
+  write_tag: jepa
+
+mask:
+  allow_overlap: false
+  aspect_ratio: [0.75, 1.5]
+  enc_mask_scale: [0.85, 1.0]
+  min_keep: 10
+  num_enc_masks: 1
+  num_pred_masks: 4
+  patch_size: 14
+  pred_mask_scale: [0.15, 0.2]
+
+meta:
+  load_checkpoint: true
+  model_name: vit_huge
+  patch_size: 14
+  pred_depth: 12
+  pred_emb_dim: 384
+  read_checkpoint: ../texjepa_n/jepa-latest.pth.tar
+  use_bfloat16: true
+  num_registers: 0
+
+optimization:
+  ema: [0.999, 1.0]
+  epochs: 50
+  final_lr: 1.0e-06
+  final_weight_decay: 0.4
+  ipe_scale: 1.25
+  lr: 3.0e-05
+  start_lr: 1.0e-06
+  warmup: 3
+  weight_decay: 0.04
+  vicreg:
+    enabled: true
+    var_weight: 1.0
+    cov_weight: 0.04
+    gamma: 1.0
+    warmup_epochs: 5

configs/pretrain/texjepa_n.yaml

@@ -0,0 +1,67 @@
+# TexJEPA-N: noise-invariant texture adaptation.
+#
+# Warm-start from the I-JEPA-300 baseline checkpoint and train for 50 more
+# epochs. The target branch sees clean chest radiographs while the context
+# branch receives mild stochastic texture corruptions. This asymmetric
+# context-target view teaches invariance to nuisance texture while preserving
+# the I-JEPA predictive objective.
+
+data:
+  dataset_type: mimic_cxr
+  root_path: data/mimic-cxr-384
+  use_zip: false
+  splits: null
+  batch_size: 128
+  crop_size: 224
+  crop_scale: [0.3, 1.0]
+  num_workers: 10
+  pin_mem: true
+  normalization: chest_xray
+  use_color_distortion: false
+  use_gaussian_blur: false
+  use_horizontal_flip: false
+  color_jitter_strength: 0.0
+  context_noise:
+    enabled: true
+    p_gauss: 0.5
+    p_poisson: 0.3
+    p_jpeg: 0.2
+    sigma_range: [0.005, 0.05]
+    peak_range: [50, 200]
+    quality_range: [40, 80]
+    warmup_epochs: 5
+
+logging:
+  folder: logs/texjepa_n
+  write_tag: jepa
+
+mask:
+  allow_overlap: false
+  aspect_ratio: [0.75, 1.5]
+  enc_mask_scale: [0.85, 1.0]
+  min_keep: 10
+  num_enc_masks: 1
+  num_pred_masks: 4
+  patch_size: 14
+  pred_mask_scale: [0.15, 0.2]
+
+meta:
+  load_checkpoint: true
+  model_name: vit_huge
+  patch_size: 14
+  pred_depth: 12
+  pred_emb_dim: 384
+  read_checkpoint: ../ijepa_300/jepa-latest.pth.tar
+  use_bfloat16: true
+  num_registers: 0
+
+optimization:
+  ema: [0.999, 1.0]
+  epochs: 50
+  final_lr: 1.0e-06
+  final_weight_decay: 0.4
+  ipe_scale: 1.25
+  lr: 5.0e-05
+  start_lr: 1.0e-06
+  warmup: 3
+  weight_decay: 0.04

configs/pretrain/texjepa_r.yaml

@@ -0,0 +1,66 @@
+# TexJEPA-R: register-token texture routing.
+#
+# Warm-start from TexJEPA-N and train for 50 additional epochs. TexJEPA-R keeps
+# the context-only texture corruption curriculum, adds four learnable register
+# tokens to the encoder, and tightens the masking distribution so the encoder
+# must rely on more local, lesion-scale structure.
+
+data:
+  dataset_type: mimic_cxr
+  root_path: data/mimic-cxr-384
+  use_zip: false
+  splits: null
+  batch_size: 128
+  crop_size: 224
+  crop_scale: [0.3, 1.0]
+  num_workers: 10
+  pin_mem: true
+  normalization: chest_xray
+  use_color_distortion: false
+  use_gaussian_blur: false
+  use_horizontal_flip: false
+  color_jitter_strength: 0.0
+  context_noise:
+    enabled: true
+    p_gauss: 0.5
+    p_poisson: 0.3
+    p_jpeg: 0.2
+    sigma_range: [0.005, 0.05]
+    peak_range: [50, 200]
+    quality_range: [40, 80]
+    warmup_epochs: 5
+
+logging:
+  folder: logs/texjepa_r
+  write_tag: jepa
+
+mask:
+  allow_overlap: false
+  aspect_ratio: [0.5, 2.0]
+  enc_mask_scale: [0.65, 0.85]
+  min_keep: 10
+  num_enc_masks: 1
+  num_pred_masks: 6
+  patch_size: 14
+  pred_mask_scale: [0.15, 0.25]
+
+meta:
+  load_checkpoint: true
+  model_name: vit_huge_reg
+  num_registers: 4
+  patch_size: 14
+  pred_depth: 12
+  pred_emb_dim: 384
+  read_checkpoint: ../texjepa_n/jepa-latest.pth.tar
+  use_bfloat16: true
+
+optimization:
+  ema: [0.999, 1.0]
+  epochs: 50
+  final_lr: 1.0e-06
+  final_weight_decay: 0.4
+  ipe_scale: 1.25
+  lr: 5.0e-05
+  start_lr: 1.0e-06
+  warmup: 5
+  weight_decay: 0.04

requirements.txt

@@ -0,0 +1,8 @@
+torch>=2.0
+torchvision>=0.15
+numpy
+pyyaml
+Pillow
+scikit-learn
+opencv-python
+submitit

weights/I-JEPA-300/jepa-latest.pth.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5fbdd329af7e996e666ef361a8cf10512aa9788f04d0668e654442bd0e4e59b1
+size 10358172879

weights/README.md

@@ -0,0 +1,10 @@
+# Weight Files
+
+The four subdirectories contain the released TexJEPA checkpoint lineage:
+
+- `I-JEPA-300`: baseline/latest I-JEPA chest X-ray checkpoint.
+- `TexJEPA-N`: noise-invariant texture specialization.
+- `TexJEPA-R`: register-token specialization from TexJEPA-N.
+- `TexJEPA-C`: covariance-regularized specialization from TexJEPA-N.
+
+See the repository model card for usage and checksums.

weights/TexJEPA-C/jepa-latest.pth.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b3ff784a334c331313effb20ae9c76126e695f0d59a63e1e4ccf2049fff2239
+size 10358172943

weights/TexJEPA-N/jepa-latest.pth.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04216e3e1a44da04553c0e816edb1ca32034dd9e8ba5ea5169efe1ef4abda602
+size 10358172943

weights/TexJEPA-R/jepa-latest.pth.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99dfe8992a49ad1a310e8eed331e5c2d1464bcaecbd43774a0b38e8d5f1117f4
+size 10358256378