Skip to content

Record: SP8192 + Pre-Quant TTT + QK-Gain 5.0 + Depth Recurrence + MuonEq-R — val_bpb 1.0791 (3-seed mean) #1423

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
aryanbhosale wants to merge 1 commit into
base: main
Choose a base branch
from
+2,752 −0
Open

Record: SP8192 + Pre-Quant TTT + QK-Gain 5.0 + Depth Recurrence + MuonEq-R — val_bpb 1.0791 (3-seed mean) #1423

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
38 changes: 38 additions & 0 deletions records/track_10min_16mb/2026-04-06_SP8192_PreQuantTTT_QKGain5/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
# Record: SP8192 + Pre-Quant TTT + QK-Gain 5.0 — val_bpb 1.0791 (3-seed mean)

**val_bpb = 1.0791** (3-seed mean, std 0.0012) | **~15.12 MB** | 8xH100 SXM

## 3-Seed Results (8xH100 80GB SXM, PyTorch 2.9.1+cu128)

| Seed | **Sliding BPB** | Artifact |
|------|-----------------|----------|
| 42 | **1.0802** | 15,123,918 |
| 314 | **1.0778** | 15,118,254 |
| 999 | **1.0794** | 15,127,567 |
| **Mean** | **1.0791** | |

Merged SOTA (PR #1019): **1.1147 BPB**. Delta: **-0.0356 BPB**.

## Key Change: QK-Gain 5.0 on the SP8192 + Pre-Quant TTT stack

Takes PR #1394 (@clarkkev) + PR #1364 pre-quant TTT and adds QK-Gain 5.0 (from 4.0). The base stack: SP8192, MLP 4x, depth recurrence (loop 4,5), MuonEq-R, SDClip, GPTQ embeddings, sigmoid-gated U-Net skips, brotli.

## Compliance (Track A — Fixed Predictor)

- No eval-time adaptation — model frozen after training + pre-quant TTT + GPTQ
- No SLOT, no n-gram cache
- Pre-quant TTT adapts EMA weights BEFORE GPTQ quantization (baked into artifact)
- Standard sliding-window eval (stride=64)
- All four conditions from Issue #1017 satisfied

## Reproduction

```bash
pip install brotli
MATCHED_FINEWEB_REPO_ID=kevclark/parameter-golf python3 data/cached_challenge_fineweb.py --variant sp8192 --skip-manifest
SEED=42 QK_GAIN_INIT=5.0 torchrun --standalone --nproc_per_node=8 train_gpt.py
```

## Credits

PR #1394 @clarkkev, PR #1364 @stukenov, PR #1416 @erichroepke, PR #1217 @bigbag, PR #1204 @msisovic, PR #1260 @dexhunter, PR #1019 @abaybektursun
20 changes: 20 additions & 0 deletions records/track_10min_16mb/2026-04-06_SP8192_PreQuantTTT_QKGain5/submission.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,20 @@
{
"author": "aryanbhosale",
"github_id": "aryanbhosale",
"name": "SP8192 + Pre-Quant TTT + QK-Gain 5.0 + Depth Recurrence + MuonEq-R + SDClip",
"date": "2026-04-06",
"track": "10min_16mb",
"val_bpb": 1.07912726,
"val_bpb_std": 0.00123751,
"seeds": [42, 314, 999],
"seed_results": {
"42": {"val_bpb": 1.08020079, "artifact_bytes": 15123918},
"314": {"val_bpb": 1.07777375, "artifact_bytes": 15118254},
"999": {"val_bpb": 1.07940724, "artifact_bytes": 15127567}
},
"comparison_baseline_pr": 1019,
"delta_vs_pr1019_bpb": -0.03560783,
"hardware": "8xH100 80GB SXM",
"pytorch_version": "2.9.1+cu128",
"technique_summary": "SP8192 + MLP 4x + Pre-Quant AdamW TTT (6 epochs) + QK-Gain 5.0 + Depth Recurrence (loop 4,5) + MuonEq-R + SDClip + GPTQ Embeddings + Brotli"
}
Loading