JorgeAV
/

MR-JEPA

vision-language

Model card Files Files and versions

MR-JEPA / mr_jepa /models

70.1 kB

Ctrl+K

Ctrl+K

1 contributor

History: 5 commits

JorgeAV's picture

fix: target_encoder.py — respect config.jepa_loss_fn (smooth_l1/mse/cosine) instead of hardcoded MSE

ac52da3 verified about 1 month ago

__init__.py

472 Bytes
Initial MR-JEPA codebase: architecture, training, evaluation, and tests about 1 month ago
answer_heads.py

12.5 kB
Initial MR-JEPA codebase: architecture, training, evaluation, and tests about 1 month ago
backbones.py

7.99 kB
fix: backbones.py — DINOv3 via timm + Qwen3-Embedding-0.6B, add get_transform(), proper layer unfreezing for both architectures about 1 month ago
evidence_memory.py

11.4 kB
Initial MR-JEPA codebase: architecture, training, evaluation, and tests about 1 month ago
latent_rollout.py

11.4 kB
Initial MR-JEPA codebase: architecture, training, evaluation, and tests about 1 month ago
mr_jepa.py

14.7 kB
fix: mr_jepa.py — proper evidence gate disable via gate_type (no lambda hack), init ablation flags from config, update docstrings to current stack about 1 month ago
target_encoder.py

11.6 kB
fix: target_encoder.py — respect config.jepa_loss_fn (smooth_l1/mse/cosine) instead of hardcoded MSE about 1 month ago