Skip to content

Commit 867b51a

Browse files
committed
research(daily): Apr 21 update — Day 12 plateau; PR openai#1758 pre-quant TTT; Recurrence Depth Curriculum; Parcae stable loops
- SOTA 1.0810 still holds (Day 12 plateau, longest in competition history) - PR openai#1758 (1.02840): pre-quant TTT — 6th attempt at illegal pattern, ignore - PR openai#1756 (1.06505): CaseOps + Recurrence Depth Curriculum (depth 1→3→4); has BOS bug; awaits Issue openai#1604 - PR openai#1755 (1.07462): CaseOps + Legal TTT; awaits Issue openai#1604 - New paper: Parcae (arXiv:2604.12946) — stable looped LMs via spectral norm constraint on injection params, relevant to Triple Loop stability - New paper: Gated Attention (arXiv:2505.06708, NeurIPS 2025) — backs PR openai#1667 Attention Output Gate - Added Session 18 lessons learned; Issue openai#1604 self-deadline Apr 24 - Primary action: implement PR openai#1586+openai#1667 immediately (9 days to deadline) https://claude.ai/code/session_0151v7YeWWUSnhmcC8U8NGUV
1 parent be05cc9 commit 867b51a

2 files changed

Lines changed: 150 additions & 5 deletions

File tree

CLAUDE.md

Lines changed: 20 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -112,14 +112,17 @@ torchrun --standalone --nproc_per_node=8 train_gpt.py
112112

113113
## Competition Strategy
114114

115-
**Merged leaderboard SOTA**: **1.0810 val_bpb** (bigbag, PR #1493, 2026-04-09) — NO CHANGE (confirmed Apr 19, Day 10 plateau — longest in competition history)
116-
**Best open legal PRs (Apr 19 update)**:
115+
**Merged leaderboard SOTA**: **1.0810 val_bpb** (bigbag, PR #1493, 2026-04-09) — NO CHANGE (confirmed Apr 21, **Day 12 plateau** — longest in competition history)
116+
**Best open legal PRs (Apr 21 update)**:
117+
- PR #1758 (kilojoules, **1.02840**): PR #1738 + Pre-Quant TTT LR=1e-3 + Unfrozen — **⚠️ LIKELY ILLEGAL** (pre-quant TTT, same adapt-then-score pattern as PR #1351/#1408/#1416/#1735). No reviews. Do NOT track.
117118
- PR #1698 (arsenis-cmd, **~~1.00995~~**): GatedDeltaNet (FLA) — **EFFECTIVELY DEAD**: BPB bug confirmed by dexhunter (double-count in `build_sentencepiece_luts`, actual ~1.189 BPB) + artifact size violation (16.47–16.60MB vs 16MB decimal limit). No organizer response; author lost GPU access. Do NOT track.
118119
- PR #1738 (alertcat, **1.03540**): CaseOps V15 + PR #1735 Pre-Quant TTT — **⚠️ BUILDS ON ILLEGAL PR #1735** (pre-quant AdamW TTT 21ep, flagged by dexhunter). No reviews. Score likely void once PR #1735 is rejected.
119-
- PR #1736 (dexhunter, **1.06549**): CaseOps bijective tokenizer + GatedAttn + QuantGate + SP8192 — **CLEANEST new PR**, no legality flags. Await Issue #1604 CaseOps ruling. QuantGate compensates artifact overhead.
120+
- PR #1756 (romeerp, **1.06505**): CaseOps + **Recurrence Depth Curriculum** (depth 1→3→4 over training thirds) + phased TTT + gated attn — NEW TODAY. ⚠️ Awaits Issue #1604 ruling. Has reproducibility bug: `prepare_caseops_data.py` missing BOS insertion → ZeroDivisionError in phased TTT eval path (@codemath3000 flagged). Training completes via fallback; eval crashes. Artifact ~15.985 MB. Watch for author fix.
121+
- PR #1755 (OE-GOD, **1.07462**): SP8192 + CaseOps + Legal TTT (no pre-quant) — NEW TODAY. ⚠️ Awaits Issue #1604 CaseOps ruling. z≈22.8, p≪0.0001 vs merged SOTA.
122+
- PR #1736 (dexhunter, **1.06549**): CaseOps bijective tokenizer + GatedAttn + QuantGate + SP8192 — **CLEANEST CaseOps PR**, no legality flags. Await Issue #1604 CaseOps ruling. QuantGate compensates artifact overhead.
120123
- PR #1693 (dexhunter, **1.05733**): Casefold V4 + AttnOutGate + SmearGate + Multi-Phase Global SGD TTT — **AWAIT CASEFOLD RULING (Issue #1604)**
121124
- PR #1729 (romeerp, **1.0678**): CaseOps bijective tokenizer + Tapered WD (50% at 70% training) — bijective/reversible, BPB via byte sidecar; await Issue #1604 ruling
122-
- PR #1667 (MarioPaerle, **1.07139**): SmearGate + Attention Output Gate (1,056 params, 12×8×11 heads) + Legal TTT — **CLEAN, no reviews, stack on #1586**
125+
- PR #1667 (MarioPaerle, **1.07139**): SmearGate + Attention Output Gate (1,056 params, 12×8×11 heads) + Legal TTT — **CLEAN, no reviews, stack on #1586**. Backed by NeurIPS 2025 arXiv:2505.06708.
123126
- PR #1727 (yahya010, **1.07217**): MP-SGD TTT 4 phases + QK-Gain 5.25 — **APPEARS LEGAL** (score-first per phase, explicit compliance notes); extends prior PR #1700 (3 phases); stackable
124127
- PR #1586 (dexhunter, **1.07493**): Per-Layer Adaptive GPTQ (MLP=12σ, Attn=13σ) + int7 Emb (15σ) + MLR=0.026 — **CLEAN, implement immediately**
125128
- PR #1732 (Victory963, **1.0785**): Hadamard Rotation + AWQ + Parallel Residuals — open, no reviews; new quantization approach
@@ -135,7 +138,7 @@ torchrun --standalone --nproc_per_node=8 train_gpt.py
135138
- PR #1647 (powerpratik, **1.0616**): SLOT-4 + TTT + 3-Layer Recurrence + Parallel Residuals — ⚠️ standard SLOT, no reviews
136139
**Best open with SLOT**: ~1.0616 val_bpb (PR #1647, powerpratik, SLOT-4) — no reviews yet
137140
**Best open (illegal)**: 1.0632 (PR #1517, RulinShao, Pre-Quant TTT 18ep — same ruling as #1351/#1416)
138-
**Target**: Beat 1.0810 merged SOTA by >=0.005 nats → need **≤1.0760 bpb**. Best reachable (legal, no CaseOps): ~1.068–1.072 (legal stack #1586+#1667+#1727+#1560). With CaseOps if ruled legal: ~1.065 (PR #1736). **11 days to deadline (Apr 30).**
141+
**Target**: Beat 1.0810 merged SOTA by >=0.005 nats → need **≤1.0760 bpb**. Best reachable (legal, no CaseOps): ~1.068–1.072 (legal stack #1586+#1667+#1727+#1560). With CaseOps if ruled legal: ~1.065 (PR #1736/#1756). **9 days to deadline (Apr 30). Issue #1604 self-imposed deadline: Apr 24 — act without ruling if no response by then.**
139142

140143
**CRITICAL LEGALITY UPDATES**:
141144
- **PR #771 REJECTED (2026-03-27)** — Our AdamW TTT 30ep was train-then-score. All 30-epoch TTT results void.
@@ -220,6 +223,8 @@ torchrun --standalone --nproc_per_node=8 train_gpt.py
220223
| MUD/MomentUm Decorrelation (arXiv:2603.17970) | +20-50% throughput | WATCH — triangular Cholesky whitening; 1.3–2.6× tokens/sec vs Muon |
221224
| Mousse (arXiv:2603.09697) | ~-0.002 to -0.003 | WATCH — Kronecker-factored preconditioning for Muon; ~12% fewer steps |
222225
| Infini-gram interpolation (arXiv:2401.17377) | large but legal unclear | WATCH — suffix array ∞-gram, normalized |
226+
| **Parcae stable loop injection (arXiv:2604.12946)** | **~6.3% lower perplexity vs prior looped models** | **WATCH — Apr 16, 2026 (UCSD + Together AI); constrains spectral norm of loop injection via negative diagonal parameterization; prevents residual explosion in our Triple Loop; may enable depth 4 or earlier activation. GitHub: github.com/sandyresearch/parcae** |
227+
| **Recurrence Depth Curriculum (arXiv:2511.07384)** | **unknown standalone** | **WATCH — PR #1756 implements depth 1→3→4 over training thirds; theoretical backing confirmed; await CaseOps ruling + BOS bug fix before adopting** |
223228
| AdamW TTT (30 ep, train-then-score) || **ILLEGAL (PR #771 rejected)** |
224229
| N-gram hash cache || **ILLEGAL (normalization, Issue #1017)** |
225230
| LoRA TTT | **+0.004 (HURTS)** | **Abandoned** |
@@ -404,3 +409,13 @@ _Updated: 2026-04-17 (v13.0 — PR #1698 GatedDeltaNet FLA 1.00995 flagged artif
404409
93. **Merged SOTA plateau now 10 days (Apr 9 → Apr 19) — deadline is 11 days away.** With 8+ open PRs in 1.062–1.078 range, a merge wave is overdue. Implementing #1586 immediately is critical — every day without it is wasted headroom.
405410

406411
_Updated: 2026-04-19 (v14.0 — PR #1698 GDN effectively dead (BPB bug ~1.189 + artifact violation); CaseOps bijective tokenizer new community technique (#1729/#1736/#1738); PR #1735 pre-quant TTT flagged illegal; PR #1727 MP-SGD TTT 4-phase appears legal at 1.07217; merged SOTA 1.0810 Day 10; 11 days remaining)_
412+
413+
### Session 18 (2026-04-21)
414+
94. **Pre-quant TTT pattern (PR #1758, 1.02840) is the 6th attempt — still illegal.** kilojoules tuned PR #1738's pre-quant TTT with LR=1e-3 and unfrozen blocks to reach 1.02840. No reviews filed; no organizer response. Same adapt-then-score violation as PR #1351/#1408/#1416/#1423/#1735. The community keeps submitting; organizers keep rejecting. Do not track.
415+
95. **Recurrence Depth Curriculum (PR #1756, romeerp, 1.06505) is a new technique with theoretical backing.** Three-phase training schedule: depth 1 (first third) → depth 3 (second third) → depth 4 (final third), eval fixed at depth 4. Grounded in arXiv:2511.07384 (retrofitted recurrence curriculum). Two issues: (a) awaits Issue #1604 CaseOps ruling; (b) `prepare_caseops_data.py` missing BOS insertion causes ZeroDivisionError in phased TTT eval path (@codemath3000 flagged Apr 21). Watch for fix. If CaseOps ruled legal and bug fixed, this is a strong stack candidate.
416+
96. **Parcae (arXiv:2604.12946, UCSD + Together AI, Apr 16) directly addresses our Triple Loop instability.** Looped models suffer residual explosion from large spectral norms in injection parameters. Parcae constrains spectral norm via negative diagonal parameterization, achieving quality of a transformer **2× the size** at equal parameter count. Our PR #1493 Triple Loop (layers 4-5 × 3, activated at 0.35×) may be leaving performance on the table due to this same instability. If time permits after #1586+#1667+#1560 are validated, investigate adding Parcae-style spectral norm constraint on loop injection weights. GitHub: github.com/sandyresearch/parcae.
417+
97. **Attention Output Gate (PR #1667) is backed by NeurIPS 2025 research (arXiv:2505.06708).** Head-specific sigmoid SDPA output gate breaks the low-rank bottleneck of consecutive Wv/Wo projections, yielding up to 0.2 PPL reduction. Multiplicative form (as in PR #1667) is optimal. Confirms our implementation target is theoretically sound. Implement with #1586.
418+
98. **Issue #1604 (CaseOps/casefold ruling) has been open 8 days with no @valerio-oai response.** Self-impose a deadline: if no ruling by Apr 24 (6 days before competition close), proceed with the clean legal stack (#1586+#1667+#1560+#1727) rather than waiting. CaseOps has a stronger legal argument than casefold (bijective, lossless, BPB on original bytes), but the clock is running.
419+
99. **Merged SOTA Day 12 plateau is now confirmed longest in competition history.** No merges since Apr 9. The 8+ open PRs between 1.062–1.078 remain unreviewed. The organizers may be reviewing multiple PRs in batch. Expect a wave when rulings come (especially Issue #1604). Check leaderboard at the start of every session.
420+
421+
_Updated: 2026-04-21 (v15.1 — merged SOTA 1.0810 Day 12; PR #1758 pre-quant TTT 1.02840 likely illegal; PR #1756 CaseOps+Recurrence Depth Curriculum 1.06505 has BOS bug + Issue #1604 pending; PR #1755 CaseOps+Legal TTT 1.07462 pending Issue #1604; Parcae arXiv:2604.12946 relevant to Triple Loop stability; Attention Output Gate backed by arXiv:2505.06708; Issue #1604 self-deadline Apr 24; 9 days remaining)_

logs/daily_research.md

Lines changed: 130 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -189,3 +189,133 @@ Rejected by @valerio-oai 2026-03-27. Train-then-score AdamW TTT 30ep on val toke
189189
---
190190

191191
_Updated: 2026-04-20 (v15.0 — Merged SOTA 1.0810 Day 11 plateau (longest ever); 5 new PRs today, none beat SOTA; PR #731 seeds still pending; Issue #1604 still unruled; PR #1749 GDN+Full-Hessian GPTQ incomplete; primary action overdue: implement PR #1586+#1667; 10 days to deadline)_
192+
193+
---
194+
195+
# Parameter Golf Daily Research - 2026-04-21
196+
197+
## PR #771 STATUS: CLOSED (ILLEGAL — confirmed, no change)
198+
199+
Rejected by @valerio-oai 2026-03-27. Train-then-score AdamW TTT 30ep on val tokens. No new comments.
200+
201+
---
202+
203+
## N-GRAM PR STATUS
204+
205+
| PR | Claimed BPB | Status | Notes |
206+
|----|-------------|--------|-------|
207+
| #727 | 0.9674 | **CLOSED (ILLEGAL)** | valerio-oai: target token in hash key = leaks eval tokens |
208+
| #758 | 1.0465 | **OPEN (effectively dead)** | Apr 12: XOR hash key includes target token, same violation as #727 |
209+
| #731 | 1.0400 | **OPEN — awaiting seeds 1337 + 2024** | Reviewer "LOOKS CLEAN". Dense count + Laplace, score-first per chunk. No movement since Apr 17. **9 days to deadline — if no seeds by Apr 24, this PR is unlikely to merge.** |
210+
211+
---
212+
213+
## Leaderboard
214+
215+
| | Score | Author | Date |
216+
|--|-------|--------|------|
217+
| **Merged SOTA** | **1.0810** | bigbag (PR #1493) | 2026-04-09 |
218+
| Best open (legal, no CaseOps) | **1.07139** | MarioPaerle (PR #1667) | |
219+
| Best open (CaseOps pending) | **1.06505** | romeerp (PR #1756) — new today | |
220+
| Best open (pre-quant TTT, likely illegal) | **1.02840** | kilojoules (PR #1758) — new today | |
221+
| Our PR #771 | 1.0705 | sunnypatneedi | CLOSED (illegal) |
222+
223+
**DAY 12 PLATEAU** — no new merges since Apr 9. Longest plateau in competition history. 9 days to deadline.
224+
225+
---
226+
227+
## What Changed (GitHub — Apr 21, 2026)
228+
229+
### New PRs filed today
230+
231+
| PR | Author | BPB | Technique | Legal? |
232+
|----|--------|-----|-----------|--------|
233+
| #1758 | kilojoules | **1.02840** | PR #1738 + Pre-Quant TTT LR=1e-3 + Unfrozen (`PREQUANT_TTT_FREEZE_BLOCKS=0`) | **⚠️ LIKELY ILLEGAL** — pre-quant TTT is same adapt-then-score pattern as PR #1735/#1351/#1408/#1416. Builds on PR #1738 (alertcat) which itself builds on PR #1735. No reviews yet. |
234+
| #1756 | romeerp | **1.06505** | CaseOps Tokenizer + **Recurrence Depth Curriculum** + phased TTT + gated attn | ⚠️ Awaits Issue #1604 CaseOps ruling. Has reproducibility bug: @codemath3000 found `prepare_caseops_data.py` missing BOS insertion → ZeroDivisionError in phased TTT eval path (training completes via fallback, but eval crashes). Artifact ~15.985 MB. |
235+
| #1755 | OE-GOD | **1.07462** | SP8192 + CaseOps + Legal TTT (no pre-quant explicitly excluded) | ⚠️ Awaits Issue #1604 CaseOps ruling. Statistically significant (-0.00638 BPB vs merged SOTA, z≈22.8). |
236+
| #1764 | gmn0105 || Non-record no-looping SOTA scaffold | Non-record, ignore |
237+
| #1763 | gmn0105 || Non-record SP8192 proxy stack | Non-record, ignore |
238+
| #1762 | frido22 | 1.5200 | Non-record Mac mini M4 | Non-record, ignore |
239+
| #1760 | BrandtChristian | 1.1863 | Non-record SP8192 + pre-quant TTT | Non-record |
240+
| #1759 | yijieyuan | 1.07994 | Non-record: LoRA on tied embedding (1 seed) | Non-record |
241+
242+
### Key open PRs — no status change since Apr 20
243+
244+
| PR | Author | Val BPB | Technique | Action |
245+
|----|--------|---------|-----------|--------|
246+
| #1586 | dexhunter | **1.07493** | Per-layer GPTQ (MLP=12σ, Attn=13σ) + int7 Emb@15σ + MLR=0.026 | **IMPLEMENT NOW** — no reviews, zero legality risk |
247+
| #1667 | MarioPaerle | **1.07139** | Attention Output Gate (1,056 params) + SmearGate (w=12) | **STACK ON #1586** — no reviews |
248+
| #1560 | dexhunter | **1.07406** | VarLen Attention (per-doc masking) + Doc-TTT (LoRA chunk=48) | Add after #1586+#1667 verified |
249+
| #1727 | yahya010 | **1.07217** | MP-SGD TTT 4 phases (score-first per phase) | Appears legal; stackable |
250+
| #1736 | dexhunter | **1.06549** | CaseOps bijective + GatedAttn + QuantGate | Awaits Issue #1604 |
251+
252+
**Issue #1604 (CaseOps/casefold legality)**: STILL OPEN. No @valerio-oai comment as of Apr 21. **Ruling deadline self-imposed: Apr 24.** If no ruling by then, proceed without CaseOps.
253+
254+
### New technique: Recurrence Depth Curriculum (PR #1756)
255+
256+
romeerp introduces a three-phase training schedule for depth recurrence:
257+
- Phase 1 (first third of training): loop depth = 1
258+
- Phase 2 (second third): loop depth = 3
259+
- Phase 3 (final third): loop depth = 4
260+
- Evaluation: always at depth 4
261+
262+
Hypothesis: "teach a useful shallow refinement operator first" before requiring deeper recurrence. This is consistent with arXiv:2511.07384 (retrofitted recurrence curriculum). **If CaseOps is ruled legal and the BOS reproducibility bug is fixed, this technique is worth stacking on our base.**
263+
264+
---
265+
266+
## New Research Papers
267+
268+
### arXiv:2604.12946 — Parcae: Stable Looped Language Models (Apr 16, 2026)
269+
**UCSD + Together AI.** Addresses instability in looped LMs caused by residual explosion (large spectral norms in injection parameters). Solution: constrain spectral norm via "negative diagonal parameterization" of injection parameters, recast as a nonlinear time-variant dynamical system.
270+
271+
Key results:
272+
- 6.3% lower val perplexity vs prior looped models at same parameter count
273+
- Achieves quality of transformer **2× the size**
274+
- At 1.3B params: +2.99/+1.18 CORE/Core-Extended points vs Transformer baseline under fixed budget
275+
- Predicts looping and training data should scale **in tandem** (not independently)
276+
277+
**Relevance to Parameter Golf**: Our Triple Loop architecture (layers 4-5 repeated 3×, activated at 0.35× training, from PR #1493) may suffer from residual explosion instability. Parcae's spectral norm constraint on injection parameters could stabilize our loops and allow deeper/more aggressive recurrence. Implementation complexity: moderate (add norm constraint to loop injection weights). **Watch for competition PRs implementing Parcae stabilization on the SP8192 stack.** GitHub: github.com/sandyresearch/parcae.
278+
279+
### arXiv:2511.07384 — Teaching Pretrained LMs to Think Deeper with Retrofitted Recurrence (Nov 2025)
280+
Proposes curriculum over recurrence depth during training (depth increases from shallow to deep). Exactly the mechanism PR #1756 implements. Validates romeerp's approach with theoretical grounding.
281+
282+
**Relevance**: If CaseOps is ruled legal, adopting Recurrence Depth Curriculum (depth 1→3→4 curriculum) on our own stack is a natural experiment. Expected gain: unclear standalone from depth curriculum alone; PR #1756 bundles it with CaseOps. Low priority until CaseOps ruling.
283+
284+
### arXiv:2505.06708 — Gated Attention: Non-linearity, Sparsity, Attention-Sink-Free (NeurIPS 2025)
285+
Head-specific sigmoid gate after SDPA output (`g = sigmoid(Wg * x)`, multiply attention output element-wise). Key findings:
286+
- Up to 0.2 PPL reduction, +2 MMLU points
287+
- Multiplicative gating > additive gating
288+
- Element-wise + head-specific is optimal balance
289+
- Improves training stability (reduces loss spikes)
290+
- Gating scores are sparse (<0.5 for most heads)
291+
292+
**Relevance**: This is the theoretical backing for PR #1667's Attention Output Gate. Confirms that our target stack element (#1667) is theoretically sound and from published NeurIPS work. Also explains *why* it works: breaks the low-rank bottleneck of consecutive Wv/Wo projections.
293+
294+
---
295+
296+
## HuggingFace / Community Discoveries
297+
298+
- **Pre-quant TTT pattern continues**: PR #1758 (1.02840) is the 6th pre-quant TTT attempt (after #1351, #1408, #1416, #1423, #1735). Community keeps trying; organizers keep rejecting. Ignore.
299+
- **Recurrence Depth Curriculum is emerging**: PR #1756 is the first competition PR to implement it. Has a reproducibility bug (BOS missing) — watch for author fix.
300+
- **No new GDN attempts with corrected BPB**: PR #1749 (GDN + Full-Hessian GPTQ) from Apr 20 still awaits full 8xH100 run.
301+
- **Parcae architecture from Together AI** (arXiv:2604.12946) could inspire stable loop injection technique — first paper to address exactly the instability pattern our depth recurrence faces.
302+
303+
---
304+
305+
## Recommended Actions (priority order)
306+
307+
1. **IMPLEMENT PR #1586 TODAY.** 9 days to deadline. Per-layer GPTQ (MLP=12σ, Attn=13σ, Emb int7@15σ), MLR=0.026. Config-level change, -0.013 nats, zero legality risk. This is critically overdue.
308+
309+
2. **STACK PR #1667 IN THE SAME RUN.** Attention Output Gate (1,056 params, init zero) + SmearGate (w=12). Combined expected: ~-0.019 nats total over merged SOTA base. Backed by NeurIPS 2025 paper (arXiv:2505.06708).
310+
311+
3. **ADD VarLen Attention + Doc-TTT (PR #1560 approach) in next run.** ~-0.007 bpb vs merged SOTA. Per-document causal masking + score-first LoRA TTT (chunk=48). dexhunter-authored; reliable.
312+
313+
4. **AWAIT Issue #1604 until Apr 24 then act.** If CaseOps ruled legal before Apr 24: add bijective CaseOps from PR #1736/PR #1755 stack. If no ruling by Apr 24: submit without CaseOps. Do not wait past Apr 24 — 6 days will remain for 3-seed runs.
314+
315+
5. **DO NOT IMPLEMENT**: Pre-quant TTT (#1758/#1735), casefold without ruling, SLOT, GDN without corrected BPB.
316+
317+
6. **INVESTIGATE Parcae stabilization for Triple Loop**: If time permits after #1586+#1667+#1560 are in, look at whether spectral norm constraint on loop injection parameters can enable a 4th loop depth or earlier activation (currently at 0.35× training). Read github.com/sandyresearch/parcae.
318+
319+
---
320+
321+
_Updated: 2026-04-21 (v15.1 — Merged SOTA 1.0810 Day 12 plateau (longest ever); PR #1758 pre-quant TTT 1.02840 likely illegal; PR #1756 CaseOps+Recurrence Depth Curriculum 1.06505 awaits BOS fix + Issue #1604; PR #1755 CaseOps+Legal TTT 1.07462 awaits Issue #1604; Parcae stable looped LM paper arXiv:2604.12946 relevant to Triple Loop stability; 9 days to deadline)_

0 commit comments

Comments
 (0)