mathias/jepa-fx-risk
1
0
Fork 0
generated from mathias/template-go-web
Code Issues 10 Pull Requests Actions Packages Projects Releases Wiki Activity
36 Commits 3 Branches 18 Tags
8eb0358c0149d9ba9380aa9f92f0796ec2f7aa3d
Commit Graph
5 Commits
Author SHA1 Message Date
mathiasandClaude Sonnet 4.6 b2bc01ba9e feat(phase1): warm-start joint encoder fine-tuning (Option B)
CD / Lint / Test / Vet (push) Successful in 3s
Details
CD / Build & Import (push) Failing after 7s
Details
CD / Deploy via GitOps (push) Has been skipped
Details 
Two-phase phase-1:
  1a. Frozen warmup: head trains on pre-computed embeddings for PHASE1_EPOCHS=200
  1b. Joint fine-tune: encoder + head for PHASE1_JOINT_EPOCHS=30 at PHASE1_ENCODER_LR=3e-6

Key design decisions:
- Warm start prevents catastrophic forgetting (PHASE1_JOINT=1 cold-start → -32 R²)
- Normalize live encoder output with FROZEN stats (mu_e/sd_e) so head sees same
  embedding distribution it was warmed up on
- head LR reduced 10× in joint phase to prevent head from racing ahead

HPO sweep: 30ep@3e-6=0.3962, 30ep@1e-5=0.3930, 50ep@3e-6=0.3923
Baseline (frozen): 0.3908. New best: phase1_r2=0.3962 (+0.0054 OOS).

New knobs: JEPA_PHASE1_JOINT (default 1), JEPA_PHASE1_JOINT_EPOCHS (default 30),
JEPA_PHASE1_ENCODER_LR (default 3e-6). 4 new tests (tests 15-18). 28/28 pass.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
 2026-06-26 13:25:09 +02:00
mathiasandClaude Sonnet 4.6 e739f84afd feat(hpo): env-var knob overrides + sweep script (18 configs)
CD / Lint / Test / Vet (push) Successful in 4s
Details
CD / Build & Import (push) Failing after 8s
Details
CD / Deploy via GitOps (push) Has been skipped
Details 
- train.py knobs all readable from JEPA_* env vars (JEPA_WINDOW, JEPA_D_MODEL,
  JEPA_DEPTH, etc.) so hpo_sweep.py can override without touching source
- scripts/hpo_sweep.py: 3×2×3 grid over D_MODEL × DEPTH × WINDOW,
  logs to results/hpo/hpo_results.jsonl with leaderboard at end
- 3 new tests: env override correctness, configs() schema validation
- 19/19 tests pass

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
 2026-06-26 12:37:27 +02:00
mathiasandClaude Sonnet 4.6 d282571c96 feat(phase1): MLP supervised head on frozen HEPA embeddings
CD / Lint / Test / Vet (push) Successful in 4s
Details
CD / Build & Import (push) Failing after 7s
Details
CD / Deploy via GitOps (push) Has been skipped
Details 
SupervisedHead: Linear(D→D/2)→GELU→Linear(D/2→1), trained on standardised
targets with proper epoch iteration (not random 200 batches) + weight_decay=1e-4.
Root cause of earlier -803 R²: unstandardised targets + ~1.2 effective passes.

Results on 2008-2023 hourly OOS (n=11,641):
  val_vol_r2 (linear probe): 0.3585
  phase1_r2  (MLP head):     0.3737  (+0.015 over probe)

New knobs: PHASE1_EPOCHS=200, PHASE1_LR=1e-3. 16/16 tests pass.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
 2026-06-26 12:33:28 +02:00
mathiasandClaude Sonnet 4.6 e31905dc43 feat(data): EUR/USD hourly pipeline + 2008-2023 M1 dataset (#2)
CD / Lint / Test / Vet (push) Successful in 4s
Details
CD / Build & Import (push) Failing after 8s
Details
CD / Deploy via GitOps (push) Has been skipped
Details 
- scripts/prepare_hourly.py: M1→hourly aggregation (realized_vol = sqrt(Σr²),
  MIN_BARS=30 threshold, no weekend rows, year-based split preserved)
- tests/test_prepare_hourly.py: 5 TDD tests, all green
- train.py: USE_HOURLY=True, WINDOW=240 (10-day), PATCH_LEN=24 (1-day patches);
  build() prefers eurusd_hourly.parquet, falls back to daily; EXPORT BLOCK updated
- Taskfile.yml: data:fetch:historical, data:prepare:hourly, data:prepare:all, data:test
- 98,591 hourly rows (2008-2023) covering GFC, Euro crisis, Brexit, COVID, Fed cycle

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
 2026-06-25 13:12:48 +02:00
mathiasandClaude Sonnet 4.6 bde651b0df feat(backbone): replace TS-JEPA+SIGReg with HEPA causal JEPA
CD / Lint / Test / Vet (push) Successful in 4s
Details
CD / Build & Import (push) Failing after 8s
Details
CD / Deploy via GitOps (push) Has been skipped
Details 
HEPA (Petersen et al., arXiv:2605.11130, ICML 2026 Spotlight):
- CausalEncoder: non-overlapping patches + per-patch LayerNorm +
  causal Transformer (generate_square_subsequent_mask) → all tokens (B, N, D)
- HorizonPredictor: MLP(cat(h_t, Δt)) → predicted future embedding;
  Δt sampled uniformly from [1, min(DELTA_T_MAX, N-1-c)] per epoch
- vicreg_loss: (1-α)·L1(norm(ĥ), norm(h*)) + α·(L_var + L_cov);
  joint training — no stop-gradient on target encoder
- Probe: last-token embedding [:, -1, :], fit on 2019-2021, eval on OOS

Results (true OOS 2022-2023):
  val_vol_r2: -0.45 (TS-JEPA+SIGReg) → +0.243/+0.276 (HEPA)
  effective_rank: 58.9/64 → 122.3/128 (near-full-rank, no collapse)
  Phase-0 gate on val_vol_r2: PASS ✓

Tests: 6/6 green (causal masking verified with non-uniform perturbation;
per-patch LayerNorm is mean-invariant so constant shifts are absorbed)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
 2026-06-25 08:05:33 +02:00