-
Phase 4 final results with training analysis
-
Upload results/phase4_smollm2.json with huggingface_hub
-
Phase 4 results - SmolLM2 decoder, 7 epochs
-
Phase 4 (SmolLM2 decoder) checkpoint - Stage 2 Epoch 3
-
Add Phase 4 training: SmolLM2-135M decoder + bridge MLP
-
Add Phase 3.1 training: gen_weight 2.0, gen_len 32, scheduled sampling, beam search
-
Upload checkpoints/hybrid_main_phase3_1_best.pt with huggingface_hub
-
Upload results/hybrid_main_phase3_1.json with huggingface_hub
-
Upload results/hybrid_main_phase3.json with huggingface_hub
-
Add Phase 3 training results
-
Add Phase 3 best checkpoint (epoch 0)
-
Add complete Phase 3 training script with generative decoder + open-ended VQA
-
Add Phase 3 training script (generative decoder + enriched evidence)
-
Upload checkpoints/hybrid_main_phase3_best.pt with huggingface_hub
-
Upload results/hybrid_main_phase3.json with huggingface_hub
-
feat: add persistent Trackio Space for image logging (space_id + sync)
-
Upload results/hybrid_main_phase2.json with huggingface_hub
-
Fix TextEncoder.unfreeze_last: compatible with both AutoModel (Qwen3Model.layers) and ForCausalLM (model.model.layers)
-
Phase 2: Add SOTA visual diagnostics (13 types) with Trackio image logging
Visual diagnostics added:
1. PCA Feature Maps (V-JEPA 2.1 style, arxiv:2603.14482)
2. Multi-head Attention Heatmaps (DINO style, arxiv:2104.14294)
3. RankMe Score anti-collapse monitoring (arxiv:2210.02885)
4. Per-dimension Variance bars (VICReg style, arxiv:2105.04906)
5. Latent Trajectory PCA (z₀→z₁→z₂→z₃)
6. Temporal Straightness metric (LeWorldModel Eq.9, arxiv:2603.19312)
7. Evidence Gate Activation Heatmaps
8. Token Norm Maps (DINOv2+Registers, arxiv:2309.16588)
9. Cross-Attention Weights (Perceiver Resampler)
10. Eigenspectrum Plot (singular value distribution)
11. Feature Similarity Heatmap overlay on input images
12. Rollout PCA per step
13. Rollout Comparison Grid (image vs PCA per step)
All logged via trackio.Image(fig_to_pil(fig)) - tested and verified. -
Upload checkpoints/hybrid_main_best.pt with huggingface_hub
-
Upload results/hybrid_main.json with huggingface_hub
-
add: train_phase2.py — Phase 2 perception fine-tuning with 3-group differential LR optimizer
-
add: launch_ablations.py — complete ablation matrix with CLI commands for all 12 experiments
-
fix: test_architecture.py — use os.path.dirname(__file__) instead of hardcoded /app for sys.path
-
fix: target_encoder.py — respect config.jepa_loss_fn (smooth_l1/mse/cosine) instead of hardcoded MSE
-
fix: test_architecture.py — correct evidence_gate attribute check (gate_type='none' not gate=None), add dinov2 config test, compact formatting
-
fix: export get_dinov2_ablation_config from configs
-
fix: test_architecture.py — add comprehensive ablation tests (no_jepa, no_rollout, no_gate, K variants, loss_fn variants, SIGReg vs VICReg, purist config)
-
fix: README.md — complete ablation table with all 13 experiments and CLI flags, add no_sigreg/vicreg_only
-
fix: ARCHITECTURE.md — complete ablation table with all 13 experiments, CLI flags, dinov2/loss_fn/sigreg/vicreg ablations, footnote on no_rollout→no_jepa
-
fix: train_mrjepa.py — add VICReg loss, --no_sigreg/--use_vicreg CLI flags, log all ablation fields in results JSON, document no_rollout→no_jepa conflation
-
fix: mr_jepa.py — proper evidence gate disable via gate_type (no lambda hack), init ablation flags from config, update docstrings to current stack
-
fix: ablation.py — wire all disable flags, add missing experiments (dinov2, mse/cosine loss, no_sigreg, vicreg_only), add run()/load_results() methods, generate CLI commands
-
fix: backbones.py — DINOv3 via timm + Qwen3-Embedding-0.6B, add get_transform(), proper layer unfreezing for both architectures
-
fix: model_config.py — update defaults to DINOv3-L + Qwen3-Embedding, add jepa_loss_fn/use_jepa fields, add get_dinov2_ablation_config()
-
fix: SAM 3.1 → jetjodh/sam3.1 (non-gated mirror) in training script
-
fix: update ARCHITECTURE.md — SAM 3.1 → jetjodh/sam3.1, align with current component stack (DINOv3-L, Qwen3-Embedding)
-
fix: SAM 3.1 → jetjodh/sam3.1 (non-gated mirror)
-
Training script with DINOv3+Qwen3-Embedding, dtype fixes, all ablation flags
-
Update README with adapted component stack (DINOv3, Qwen3, SAM3.1, etc.)
-
Upload checkpoints/test_cpu_best.pt with huggingface_hub
-
Upload results/test_cpu.json with huggingface_hub
-
Remove __pycache__ files
-
Initial MR-JEPA codebase: architecture, training, evaluation, and tests
-
initial commit
