diff --git a/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/README.md b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/README.md new file mode 100644 index 0000000000..83921c7454 --- /dev/null +++ b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/README.md @@ -0,0 +1,95 @@ +# Record: SP8192 + 3-Layer Recurrence + Parallel Residuals + QK-Gain 5.25 + Legal TTT + +**val_bpb (TTT) = 1.08083** (3-seed mean, std 0.00062) | **~15.97 MB** artifact | 8xH100 SXM + +Current SOTA PR #1413: **1.0810 BPB** (std 0.0002). Our mean lands 0.00017 below that, within our own std. + +## 3-Seed Results + +| Seed | Pre-Quant | Sliding | **TTT** | Artifact | +|------|-----------|---------|---------|----------| +| 1337 | 1.08609 | 1.08083 | **1.08032** | 15,971,929 | +| 42 | 1.08733 | 1.08214 | **1.08152** | 15,973,790 | +| 7 | 1.08635 | 1.08135 | **1.08064** | 15,971,863 | +| **Mean** | **1.08659** | **1.08144** | **1.08083** | **15,972,527** | +| **Std** | 0.00064 | 0.00066 | **0.00062** | | + +## Key Techniques + +1. **SP8192 + GPTQ SDClip:** int6 matrices (k=12.85), int8 embeddings (k=20.0), zero pruning (PR #1394 @clarkkev) +2. **3-Layer Depth Recurrence:** loops layers 3,4,5 twice, activates at frac=0.35 (PR #1331, #1437 @dexhunter) +3. **Parallel Residuals** (layers 7+): GPT-J style, attention and MLP read same input (PR #1412 @Robby955, PR #1204 @msisovic) +4. **QK-Gain 5.25:** learnable per-head query scaling (PR #1413 @dexhunter) +5. **Legal Score-First TTT:** SGD (lr=0.005, momentum=0.9), 3 epochs per 32K-token chunk, freeze first 9 blocks, cross-rank `dist.all_reduce` on gradients, cosine LR decay (PR #549 @abaybektursun, PR #1413 @dexhunter) +6. **Tuned Hyperparameters:** MUON_WD=0.095, MATRIX_LR=0.022, EMA=0.9965, WARMDOWN_FRAC=0.72 (PR #1445 @X-Abhishek-X) +7. **FA3/SDPA backend switch:** `USE_FA3=1` uses `flash_attn_3_func` on Hopper; `USE_FA3=0` falls back to `F.scaled_dot_product_attention(..., enable_gqa=True)` for non-Hopper GPUs. + +## Architecture + +11L × 512d, 8 heads / 4 KV heads, MLP 4×, LeakyReLU(0.5)², Partial RoPE (16/64 head dims), layerwise LN scale, tied embeddings, logit softcap 30.0. +Depth recurrence: encoder `[0,1,2,3,4,5,3,4]`, decoder `[5,3,4,5,6,7,8,9,10]`, loops layers 3-5 (three times total), activated at ~step 2016. +Parallel residuals from layer 7. Skip gates (sigmoid-gated U-Net connections). + +## Training + +MuonEq-R optimizer (row-normalized Muon, Newton-Schulz 5 steps) for matrices; AdamW for embeddings and scalars. **4550 steps in 588s** on 8×H100 SXM. Linear warmdown to LR=0 over the final 72% of training. EMA decay 0.9965. Data: LightSpeedUp public SP8192 pre-tokenized FineWeb mirror. + +## Quantization + +Full-Hessian GPTQ with SDClip (`clip = k × std(row)`): int6 for attention/MLP matrices (k=12.85), int8 for token embeddings (k=20.0). 64 calibration batches. Byte-shuffle + Brotli-11 compression. Zero pruning needed; model fits natively at ~15.97 MB. + +## TTT (Test-Time Training) + +Score-first, chunk-based SGD adaptation at eval time: + +- Val tokens divided into 32K-token chunks (~1240 chunks over full val) +- For each chunk: (1) score all sliding windows under `torch.no_grad()`, (2) SGD on chunk tokens with frozen first 9 blocks +- 3 epochs per chunk, cosine LR decay across chunks +- Gradient clip 1.0, cross-rank `dist.all_reduce` on gradients before every optimizer step +- Total TTT eval time ~420s per seed (within 600s eval budget) + +## Compliance (Issue #1017, Track B legal eval-time adaptation) + +- **Condition 1 (Causality):** Sliding-window eval is strictly causal. Each position scored from prefix tokens only. +- **Condition 2 (Normalized distribution):** Standard softmax over full vocab. No n-gram cache, no logit biasing. +- **Condition 3 (Score before update):** Each chunk fully scored under `torch.no_grad()` BEFORE any SGD update. Training only on already-scored tokens. +- **Condition 4 (Single pass):** Each token scored exactly once. No rescoring, no multi-pass selection. + +Additional: + +- No SLOT (standard or causal) +- No pre-quant TTT on val data (model quantized once during training, TTT adapts at eval time) +- No ETLB (eval-time logit bias) +- No n-gram cache or tilt +- All 3 seeds' artifacts under 16,000,000 bytes +- Training ≤ 600s on all 3 seeds (~588s actual) +- Eval (sliding + TTT) ≤ 600s on all 3 seeds (~545s actual) + +## Reproduction + +```bash +pip install brotli sentencepiece +# Hopper only (FA3). On non-Hopper GPUs skip this line and set USE_FA3=0 below. +pip install flash_attn_3 --no-deps --find-links https://windreamer.github.io/flash-attention3-wheels/cu128_torch291/ + +# Download LSU SP8192 pre-tokenized FineWeb into the expected layout: +python3 -c "from huggingface_hub import snapshot_download; snapshot_download('LightSpeedUp/parameter-golf-data', repo_type='dataset', local_dir='./data_lsu', allow_patterns=['fineweb_sp8192/*.bin','tokenizers/fineweb_8192_bpe.*'])" +mkdir -p data/datasets data/tokenizers +ln -s "$(pwd)/data_lsu/fineweb_sp8192" data/datasets/fineweb10B_sp8192 +ln -s "$(pwd)/data_lsu/tokenizers/fineweb_8192_bpe.model" data/tokenizers/fineweb_8192_bpe.model + +SEED=42 QK_GAIN_INIT=5.25 LOOP_START=3 LOOP_END=5 ENABLE_LOOPING_AT=0.35 \ +PARALLEL_RESIDUAL_START=7 MATRIX_LR=0.022 MUON_WD=0.095 EMBED_WD=0.095 \ +EMA_DECAY=0.9965 WARMDOWN_FRAC=0.72 TTT_ENABLED=1 TTT_EPOCHS=3 \ + torchrun --standalone --nproc_per_node=8 train_gpt.py +``` + +## Credits + +- **@clarkkev:** SP8192 + GPTQ Embeddings + SDClip + MuonEq-R + U-Net skips (PR #1394) +- **@dexhunter:** 3-layer depth recurrence (PR #1331, #1437), Legal TTT on SP8192 (PR #1413) +- **@abaybektursun:** Score-first TTT framework (PR #549) +- **@Robby955:** Parallel residuals on SP8192 (PR #1412) +- **@msisovic:** Parallel residuals concept (PR #1204) +- **@X-Abhishek-X:** Hyperparameter tuning WD=0.095, MLR=0.022, EMA=0.9965, warmdown=0.72 (PR #1445) +- **@LightSpeedUp:** Public SP8192 pre-tokenized FineWeb mirror (HuggingFace) diff --git a/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/submission.json b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/submission.json new file mode 100644 index 0000000000..915a2a07e3 --- /dev/null +++ b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/submission.json @@ -0,0 +1,32 @@ +{ + "track": "10min_16mb", + "date": "2026-04-22", + "name": "SP8192_ParResid_3LayerLoop_QK525_LegalTTT", + "author": "Anmar Hindi", + "github_id": "anmarhindi", + "val_bpb_note": "3-seed mean on 8xH100 SXM 600s, sliding-window eval stride=64, Legal Score-First TTT per PR #1413", + "val_bpb_sliding": 1.08144, + "val_bpb_ttt": 1.08083, + "val_bpb_std": 0.00062, + "artifact_bytes_mean": 15972527, + "artifact_bytes_max": 15973790, + "total_submission_bytes_mean": 16045285, + "total_submission_bytes_max": 16046548, + "seeds_run": 3, + "base": "PR #1394 (clarkkev) + PR #1413 (dexhunter)", + "architecture": "11L 512d GQA 8/4 MLP 4x, U-Net skips, 3-layer loop (3,4,5 @ frac=0.35), parallel residuals L7+, XSA-all-11, QK-gain 5.25", + "techniques": [ + "SP8192 LSU pre-tokenized FineWeb", + "U-Net encoder-decoder skip connections (skip_weights + skip_gates)", + "Per-block resid_mix + attn_scale + mlp_scale", + "Parallel residuals layers 7+ (GPT-J style)", + "3-layer depth recurrence (loops layers 3,4,5 twice, activated at frac=0.35)", + "QK-gain 5.25 (learnable per-head)", + "MuonEq-R optimizer (row-normalized Muon), momentum 0.99 warmup 0.92", + "EMA decay 0.9965, linear warmdown to 0 over final 72% of training", + "GPTQ int6 matrices (SDClip k=12.85) + int8 embeddings (SDClip k=20.0), 64 calibration batches", + "Byte-shuffle + Brotli-11 compression", + "Legal Score-First TTT (SGD lr=0.005 momentum=0.9, 3 epochs per 32K-token chunk, freeze first 9 blocks)" + ], + "compliance_notes": "Issue #1017 conditions 1-4 satisfied. TTT scores chunk fully under torch.no_grad BEFORE any SGD update; each token scored exactly once; single sliding-window pass; standard softmax over full vocab; causal attention." +} diff --git a/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_gpt.py b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_gpt.py new file mode 100644 index 0000000000..fd82556bcc --- /dev/null +++ b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_gpt.py @@ -0,0 +1,1692 @@ +import collections +import copy +import glob +import io +import lzma +import math +import os +from pathlib import Path +import random +import re +import subprocess +import sys +import time +import uuid + +import numpy as np +import sentencepiece as spm +import torch +import torch.distributed as dist +import torch.nn.functional as F +from torch.nn.parallel import DistributedDataParallel as DDP +from torch import Tensor, nn + +try: + from flash_attn_interface import flash_attn_func as _flash_attn_3_func + _HAS_FA3 = True +except ImportError: + _flash_attn_3_func = None + _HAS_FA3 = False + +def flash_attn_3_func(q, k, v, causal=True): + if _HAS_FA3 and os.environ.get("USE_FA3", "1") == "1": + return _flash_attn_3_func(q, k, v, causal=causal) + q_ = q.transpose(1, 2); k_ = k.transpose(1, 2); v_ = v.transpose(1, 2) + y = F.scaled_dot_product_attention(q_, k_, v_, is_causal=causal, enable_gqa=True) + return y.transpose(1, 2).contiguous() + +# ---------------------------------------- +# Hyperparameters +# ---------------------------------------- + +class Hyperparameters(): + # Experiment settings + data_dir = os.environ.get('DATA_DIR', './data/') + seed = int(os.environ.get('SEED', 1337)) + run_id = os.environ.get("RUN_ID", str(uuid.uuid4())) + + # Training length + iterations = int(os.environ.get('ITERATIONS', 20000)) + warmdown_frac = float(os.environ.get('WARMDOWN_FRAC', 0.667)) + warmup_steps = int(os.environ.get('WARMUP_STEPS', 20)) + train_batch_tokens = int(os.environ.get('TRAIN_BATCH_TOKENS', 2048 * 48 * 8)) + train_seq_len = int(os.environ.get('TRAIN_SEQ_LEN', 2048)) + train_log_every = int(os.environ.get('TRAIN_LOG_EVERY', 500)) + max_wallclock_seconds = float(os.environ.get('MAX_WALLCLOCK_SECONDS', 600.0)) + + # Validation/Evals + val_batch_tokens = int(os.environ.get('VAL_BATCH_TOKENS', 2048 * 32 * 8)) + eval_seq_len = int(os.environ.get('EVAL_SEQ_LEN', 2048)) + val_loss_every = int(os.environ.get('VAL_LOSS_EVERY', 4000)) + sliding_window_enabled = bool(int(os.environ.get('SLIDING_WINDOW_ENABLED', '1'))) + + # Legal Score-First TTT (Issue #1017). Score chunk under no_grad FIRST (counts + # toward BPB), then SGD-train on that chunk. Last chunk: score only. + # Matches PR #1413 recipe: SGD lr=0.005 momentum=0.9, 3 epochs per 32K-token chunk. + ttt_enabled = bool(int(os.environ.get('TTT_ENABLED', 0))) + ttt_lr = float(os.environ.get('TTT_LR', 0.005)) + ttt_momentum = float(os.environ.get('TTT_MOMENTUM', 0.9)) + ttt_epochs = int(os.environ.get('TTT_EPOCHS', 3)) + ttt_chunk = int(os.environ.get('TTT_CHUNK_TOKENS', 32768)) + ttt_freeze_blocks = int(os.environ.get('TTT_FREEZE_BLOCKS', 9)) + ttt_grad_clip = float(os.environ.get('TTT_GRAD_CLIP', 1.0)) + ttt_batch_seqs = int(os.environ.get('TTT_BATCH_SEQS', 4)) + + # Model architecture + vocab_size = int(os.environ.get('VOCAB_SIZE', 8192)) + num_layers = int(os.environ.get('NUM_LAYERS', 11)) + xsa_last_n = int(os.environ.get('XSA_LAST_N', 11)) + model_dim = int(os.environ.get('MODEL_DIM', 512)) + embedding_dim = int(os.environ.get('EMBEDDING_DIM', 512)) + num_kv_heads = int(os.environ.get('NUM_KV_HEADS', 4)) + num_heads = int(os.environ.get('NUM_HEADS', 8)) + mlp_mult = float(os.environ.get('MLP_MULT', 4.0)) + skip_gates_enabled = bool(int(os.environ.get('SKIP_GATES_ENABLED', '1'))) + tie_embeddings = bool(int(os.environ.get('TIE_EMBEDDINGS', '1'))) + logit_softcap = float(os.environ.get('LOGIT_SOFTCAP', 30.0)) + rope_base = float(os.environ.get('ROPE_BASE', 10000.0)) + rope_dims = int(os.environ.get('ROPE_DIMS', 16)) + rope_train_seq_len = int(os.environ.get('ROPE_TRAIN_SEQ_LEN', 2048)) + ln_scale = bool(int(os.environ.get('LN_SCALE', '1'))) + qk_gain_init = float(os.environ.get('QK_GAIN_INIT', 4.0)) + + # Layer looping + num_loops = int(os.environ.get('NUM_LOOPS', 2)) + loop_start = int(os.environ.get('LOOP_START', 4)) + loop_end = int(os.environ.get('LOOP_END', 5)) + enable_looping_at = float(os.environ.get('ENABLE_LOOPING_AT', 0.5)) + + # Parallel residuals (GPT-J style): layers >= this index use parallel attn+mlp path. + # Set to num_layers to disable. PR #1412/#1413 uses 7. + parallel_residual_start = int(os.environ.get('PARALLEL_RESIDUAL_START', 999)) + + # Multi-Token Prediction (MTP). Speculative beyond-SOTA addition. + # Head k predicts token at t+1+k via a D->D transform sharing the tied embedding. + # MTP params are TRAINING-only: dropped before serialize, ignored at eval. + mtp_heads = int(os.environ.get('MTP_HEADS', 0)) + mtp_weight = float(os.environ.get('MTP_WEIGHT', 0.3)) + + # Optimizer + min_lr = float(os.environ.get('MIN_LR', 0.0)) + embed_lr = float(os.environ.get('EMBED_LR', 0.6)) + head_lr = float(os.environ.get('HEAD_LR', 0.008)) + tied_embed_lr = float(os.environ.get('TIED_EMBED_LR', 0.03)) + tied_embed_init_std = float(os.environ.get('TIED_EMBED_INIT_STD', 0.005)) + matrix_lr = float(os.environ.get('MATRIX_LR', 0.02)) + scalar_lr = float(os.environ.get('SCALAR_LR', 0.02)) + muon_momentum = float(os.environ.get('MUON_MOMENTUM', 0.99)) + muon_backend_steps = int(os.environ.get('MUON_BACKEND_STEPS', 5)) + muon_momentum_warmup_start = float(os.environ.get('MUON_MOMENTUM_WARMUP_START', 0.92)) + muon_momentum_warmup_steps = int(os.environ.get('MUON_MOMENTUM_WARMUP_STEPS', 1500)) + muon_row_normalize = bool(int(os.environ.get('MUON_ROW_NORMALIZE', '1'))) + beta1 = float(os.environ.get('BETA1', 0.9)) + beta2 = float(os.environ.get('BETA2', 0.95)) + adam_eps = float(os.environ.get('ADAM_EPS', 1e-8)) + grad_clip_norm = float(os.environ.get('GRAD_CLIP_NORM', 0.3)) + eval_stride = int(os.environ.get('EVAL_STRIDE', 64)) + muon_beta2 = float(os.environ.get('MUON_BETA2', 0.95)) + adam_wd = float(os.environ.get('ADAM_WD', 0.02)) + muon_wd = float(os.environ.get('MUON_WD', 0.085)) + embed_wd = float(os.environ.get('EMBED_WD', 0.085)) + ema_decay = float(os.environ.get('EMA_DECAY', 0.997)) + + # Quantization & Compression + compressor = os.environ.get('COMPRESSOR', 'brotli') + gptq_calibration_batches = int(os.environ.get('GPTQ_CALIBRATION_BATCHES', 64)) + gptq_reserve_seconds = float(os.environ.get('GPTQ_RESERVE_SECONDS', 12.0)) + matrix_bits = int(os.environ.get('MATRIX_BITS', 6)) + embed_bits = int(os.environ.get('EMBED_BITS', 8)) + matrix_clip_sigmas = float(os.environ.get('MATRIX_CLIP_SIGMAS', 12.85)) + embed_clip_sigmas = float(os.environ.get('EMBED_CLIP_SIGMAS', 20.0)) + + # Distributed setup + distributed = "RANK" in os.environ and "WORLD_SIZE" in os.environ + rank = int(os.environ.get("RANK", "0")) + world_size = int(os.environ.get("WORLD_SIZE", "1")) + local_rank = int(os.environ.get("LOCAL_RANK", "0")) + is_main_process = rank == 0 + grad_accum_steps = 8 // world_size + + # Data paths + datasets_dir = os.path.join(data_dir, 'datasets', f'fineweb10B_sp{vocab_size}') + train_files = os.path.join(datasets_dir, 'fineweb_train_*.bin') + val_files = os.path.join(datasets_dir, 'fineweb_val_*.bin') + tokenizer_path = os.path.join(data_dir, 'tokenizers', f'fineweb_{vocab_size}_bpe.model') + + # Experiment files + logfile = f"logs/{run_id}.txt" + model_path = "final_model.pt" + quantized_model_path = "final_model.int6.ptz" + +# ---------------------------------------- +# Global Logging Function +# ---------------------------------------- + +_logger_hparams = None + + +def set_logging_hparams(h: Hyperparameters) -> None: + global _logger_hparams + _logger_hparams = h + + +def log(msg, console: bool = True) -> None: + if _logger_hparams is None: + print(msg) + return + if _logger_hparams.is_main_process: + if console: + print(msg) + if _logger_hparams.logfile is not None: + with open(_logger_hparams.logfile, "a", encoding="utf-8") as f: + print(msg, file=f) + +# ---------------------------------------- +# Data Loading +# ---------------------------------------- + +class ValidationData: + def __init__(self, h: Hyperparameters, device: torch.device): + self.sp = spm.SentencePieceProcessor(model_file=h.tokenizer_path) + if int(self.sp.vocab_size()) != h.vocab_size: + raise ValueError( + f"VOCAB_SIZE={h.vocab_size} does not match tokenizer vocab_size={int(self.sp.vocab_size())}" + ) + self.val_tokens = load_validation_tokens(h.val_files, h.eval_seq_len) + # VAL_TOKENS_LIMIT lets pre-flight tests run on a tiny slice of val for + # fast end-to-end pipeline verification (not for real submissions). + val_limit = int(os.environ.get('VAL_TOKENS_LIMIT', 0)) + if val_limit > 0 and val_limit < self.val_tokens.numel(): + self.val_tokens = self.val_tokens[:val_limit].contiguous() + log(f"VAL_TOKENS_LIMIT={val_limit} applied — val_tokens truncated for fast test") + self.base_bytes_lut, self.has_leading_space_lut, self.is_boundary_token_lut = ( + build_sentencepiece_luts(self.sp, h.vocab_size, device)) + + +def build_sentencepiece_luts( + sp: spm.SentencePieceProcessor, vocab_size: int, device: torch.device +) -> tuple[Tensor, Tensor, Tensor]: + sp_vocab_size = int(sp.vocab_size()) + # The BPB calculation assumes "▁" is its own token so that leading-space bytes + # are counted correctly. See https://github.com/openai/parameter-golf/issues/897 + assert sp.piece_to_id("\u2581") != sp.unk_id(), \ + "Tokenizer must have '▁' (space) as its own token for correct BPB byte counting" + table_size = max(sp_vocab_size, vocab_size) + base_bytes_np = np.zeros((table_size,), dtype=np.int16) + has_leading_space_np = np.zeros((table_size,), dtype=np.bool_) + is_boundary_token_np = np.ones((table_size,), dtype=np.bool_) + for token_id in range(sp_vocab_size): + if sp.is_control(token_id) or sp.is_unknown(token_id) or sp.is_unused(token_id): + continue + is_boundary_token_np[token_id] = False + if sp.is_byte(token_id): + base_bytes_np[token_id] = 1 + continue + piece = sp.id_to_piece(token_id) + if piece.startswith("\u2581"): + has_leading_space_np[token_id] = True + piece = piece[1:] + base_bytes_np[token_id] = len(piece.encode("utf-8")) + return ( + torch.tensor(base_bytes_np, dtype=torch.int16, device=device), + torch.tensor(has_leading_space_np, dtype=torch.bool, device=device), + torch.tensor(is_boundary_token_np, dtype=torch.bool, device=device), + ) + + +def load_validation_tokens(pattern: str, seq_len: int) -> Tensor: + files = [Path(p) for p in sorted(glob.glob(pattern))] + if not files: + raise FileNotFoundError(f"No files found for pattern: {pattern}") + # The export pipeline writes the fixed first-50k-doc validation set to fineweb_val_*. + tokens = torch.cat([load_data_shard(file) for file in files]).contiguous() + usable = ((tokens.numel() - 1) // seq_len) * seq_len + if usable <= 0: + raise ValueError(f"Validation split is too short for TRAIN_SEQ_LEN={seq_len}") + return tokens[: usable + 1] + + +def load_data_shard(file: Path) -> Tensor: + header_bytes = 256 * np.dtype(" int: + key = str(file) + cached = _SHARD_NTOKENS_CACHE.get(key) + if cached is not None: + return cached + header = np.fromfile(file, dtype=" np.memmap: + key = str(file) + mm = _MMAP_CACHE.get(key) + if mm is not None: + return mm + n = _read_num_tokens(file) + mm = np.memmap(file, mode="r", dtype=" None: + max_phase = min(self.seq_len - 1, max(0, self.num_tokens[si] - self.seq_len - 1)) + phase = int(self.rng.integers(max_phase + 1)) if max_phase > 0 else 0 + num_sequences = (self.num_tokens[si] - 1 - phase) // self.seq_len + sequence_order = self.rng.permutation(num_sequences) + self.start_inds[si] = (phase + sequence_order * self.seq_len).tolist() + + def next_batch(self, global_tokens: int, grad_accum_steps: int) -> tuple[Tensor, Tensor]: + device_tokens = global_tokens // (self.world_size * grad_accum_steps) + device_batch_size = device_tokens // self.seq_len + remaining = np.array([len(s) for s in self.start_inds], dtype=np.float64) + x = torch.empty((device_batch_size, self.seq_len), dtype=torch.int64) + y = torch.empty((device_batch_size, self.seq_len), dtype=torch.int64) + for bi in range(device_batch_size): + total = remaining.sum() + if total <= 0: + for si in range(len(self.files)): + self._reset_shard(si) + remaining = np.array([len(s) for s in self.start_inds], dtype=np.float64) + total = remaining.sum() + probs = remaining / total + si = int(self.rng.choice(len(self.files), p=probs)) + start_ind = self.start_inds[si].pop() + remaining[si] -= 1 + mm = _get_shard_memmap(self.files[si]) + window = torch.as_tensor( + np.array(mm[start_ind:start_ind + self.seq_len + 1], dtype=np.int64)) + x[bi] = window[:-1] + y[bi] = window[1:] + return x.to(self.device, non_blocking=True), y.to(self.device, non_blocking=True) + +# ---------------------------------------- +# Model Architecture +# ---------------------------------------- + +class RMSNorm(nn.Module): + def __init__(self, eps: float | None = None): + super().__init__() + self.eps = eps + + def forward(self, x: Tensor) -> Tensor: + return F.rms_norm(x, (x.size(-1),), eps=self.eps) + + +class CastedLinear(nn.Linear): + def forward(self, x: Tensor) -> Tensor: + w = self.weight.to(x.dtype) + bias = self.bias.to(x.dtype) if self.bias is not None else None + return F.linear(x, w, bias) + + +class Rotary(nn.Module): + def __init__(self, dim: int, base: float = 10000.0, train_seq_len: int = 1024, rope_dims: int = 0): + super().__init__() + self.dim = dim + self.base = base + self.train_seq_len = train_seq_len + self.rope_dims = rope_dims if rope_dims > 0 else dim + inv_freq = 1.0 / (base ** (torch.arange(0, self.rope_dims, 2, dtype=torch.float32) / self.rope_dims)) + self.register_buffer("inv_freq", inv_freq, persistent=False) + self._seq_len_cached = 0 + self._cos_cached: Tensor | None = None + self._sin_cached: Tensor | None = None + + def forward(self, seq_len: int, device: torch.device, dtype: torch.dtype) -> tuple[Tensor, Tensor]: + if ( + self._cos_cached is None + or self._sin_cached is None + or self._seq_len_cached != seq_len + or self._cos_cached.device != device + ): + rd = self.rope_dims + if seq_len > self.train_seq_len: + scale = seq_len / self.train_seq_len + new_base = self.base * (scale ** (rd / (rd - 2))) + inv_freq = 1.0 / (new_base ** (torch.arange( + 0, rd, 2, dtype=torch.float32, device=device) / rd)) + else: + inv_freq = self.inv_freq.to(device) + t = torch.arange(seq_len, device=device, dtype=inv_freq.dtype) + freqs = torch.outer(t, inv_freq) + self._cos_cached = freqs.cos()[None, :, None, :] + self._sin_cached = freqs.sin()[None, :, None, :] + self._seq_len_cached = seq_len + return self._cos_cached.to(dtype=dtype), self._sin_cached.to(dtype=dtype) + + +def apply_rotary_emb(x: Tensor, cos: Tensor, sin: Tensor, rope_dims: int = 0) -> Tensor: + if rope_dims > 0 and rope_dims < x.size(-1): + x_rope, x_pass = x[..., :rope_dims], x[..., rope_dims:] + half = rope_dims // 2 + x1, x2 = x_rope[..., :half], x_rope[..., half:] + x_rope = torch.cat((x1 * cos + x2 * sin, x1 * (-sin) + x2 * cos), dim=-1) + return torch.cat((x_rope, x_pass), dim=-1) + half = x.size(-1) // 2 + x1, x2 = x[..., :half], x[..., half:] + return torch.cat((x1 * cos + x2 * sin, x1 * (-sin) + x2 * cos), dim=-1) + + +class CausalSelfAttention(nn.Module): + def __init__(self, dim: int, num_heads: int, num_kv_heads: int, + rope_base: float, qk_gain_init: float, train_seq_len: int): + super().__init__() + if dim % num_heads != 0: + raise ValueError("model_dim must be divisible by num_heads") + if num_heads % num_kv_heads != 0: + raise ValueError("num_heads must be divisible by num_kv_heads") + self.num_heads = num_heads + self.num_kv_heads = num_kv_heads + self.head_dim = dim // num_heads + if self.head_dim % 2 != 0: + raise ValueError("head_dim must be even for RoPE") + kv_dim = self.num_kv_heads * self.head_dim + self.c_q = CastedLinear(dim, dim, bias=False) + self.c_k = CastedLinear(dim, kv_dim, bias=False) + self.c_v = CastedLinear(dim, kv_dim, bias=False) + self.proj = CastedLinear(dim, dim, bias=False) + self.proj._zero_init = True + self.q_gain = nn.Parameter(torch.full((num_heads,), qk_gain_init, dtype=torch.float32)) + self.rope_dims = 0 + self.rotary = Rotary(self.head_dim, base=rope_base, train_seq_len=train_seq_len) + self.use_xsa = False + + def _xsa_efficient(self, y: Tensor, v: Tensor) -> Tensor: + B, T, H, D = y.shape + Hkv = v.size(-2) + group = H // Hkv + y_g = y.reshape(B, T, Hkv, group, D) + vn = F.normalize(v, dim=-1).unsqueeze(-2) + proj = (y_g * vn).sum(dim=-1, keepdim=True) * vn + return (y_g - proj).reshape(B, T, H, D) + + def forward(self, x: Tensor) -> Tensor: + bsz, seqlen, dim = x.shape + q = self.c_q(x).reshape(bsz, seqlen, self.num_heads, self.head_dim) + k = self.c_k(x).reshape(bsz, seqlen, self.num_kv_heads, self.head_dim) + v = self.c_v(x).reshape(bsz, seqlen, self.num_kv_heads, self.head_dim) + q = F.rms_norm(q, (q.size(-1),)) + k = F.rms_norm(k, (k.size(-1),)) + cos, sin = self.rotary(seqlen, x.device, q.dtype) + q = apply_rotary_emb(q, cos, sin, self.rope_dims) + k = apply_rotary_emb(k, cos, sin, self.rope_dims) + q = q * self.q_gain.to(dtype=q.dtype)[None, None, :, None] + y = flash_attn_3_func(q, k, v, causal=True) + if self.use_xsa: + y = self._xsa_efficient(y, v) + y = y.reshape(bsz, seqlen, dim) + return self.proj(y) + + +class MLP(nn.Module): + def __init__(self, dim: int, mlp_mult: int): + super().__init__() + hidden = int(mlp_mult * dim) + self.fc = CastedLinear(dim, hidden, bias=False) + self.proj = CastedLinear(hidden, dim, bias=False) + self.proj._zero_init = True + + def forward(self, x: Tensor) -> Tensor: + return self.proj(F.leaky_relu(self.fc(x), negative_slope=0.5).square()) + + +class Block(nn.Module): + def __init__(self, dim: int, num_heads: int, num_kv_heads: int, mlp_mult: int, + rope_base: float, qk_gain_init: float, train_seq_len: int, + layer_idx: int = 0, ln_scale: bool = False, + parallel_residual: bool = False): + super().__init__() + self.attn_norm = RMSNorm() + self.mlp_norm = RMSNorm() + self.attn = CausalSelfAttention( + dim, num_heads, num_kv_heads, rope_base, qk_gain_init, train_seq_len) + self.mlp = MLP(dim, mlp_mult) + self.attn_scale = nn.Parameter(torch.ones(dim, dtype=torch.float32)) + self.mlp_scale = nn.Parameter(torch.ones(dim, dtype=torch.float32)) + self.resid_mix = nn.Parameter(torch.stack((torch.ones(dim), torch.zeros(dim))).float()) + self.ln_scale_factor = 1.0 / math.sqrt(layer_idx + 1) if ln_scale else 1.0 + self.parallel_residual = parallel_residual + + def forward(self, x: Tensor, x0: Tensor) -> Tensor: + mix = self.resid_mix.to(dtype=x.dtype) + x_in = mix[0][None, None, :] * x + mix[1][None, None, :] * x0 + if self.parallel_residual: + # GPT-J style: attention and MLP read from same input (x_in). + attn_out = self.attn(self.attn_norm(x_in) * self.ln_scale_factor) + mlp_out = self.mlp(self.mlp_norm(x_in) * self.ln_scale_factor) + x_out = x_in + self.attn_scale.to(dtype=x_in.dtype)[None, None, :] * attn_out + x_out = x_out + self.mlp_scale.to(dtype=x_out.dtype)[None, None, :] * mlp_out + return x_out + attn_out = self.attn(self.attn_norm(x_in) * self.ln_scale_factor) + x_out = x_in + self.attn_scale.to(dtype=x_in.dtype)[None, None, :] * attn_out + x_out = x_out + self.mlp_scale.to(dtype=x_out.dtype)[None, None, :] * self.mlp( + self.mlp_norm(x_out) * self.ln_scale_factor) + return x_out + + +class GPT(nn.Module): + def __init__(self, h: Hyperparameters): + super().__init__() + if h.logit_softcap <= 0.0: + raise ValueError(f"logit_softcap must be positive, got {h.logit_softcap}") + self.tie_embeddings = h.tie_embeddings + self.tied_embed_init_std = h.tied_embed_init_std + self.logit_softcap = h.logit_softcap + self.tok_emb = nn.Embedding(h.vocab_size, h.embedding_dim) + if h.embedding_dim != h.model_dim: + self.embed_proj = CastedLinear(h.embedding_dim, h.model_dim, bias=False) + self.head_proj = CastedLinear(h.model_dim, h.embedding_dim, bias=False) + else: + self.embed_proj = None + self.head_proj = None + self.num_encoder_layers = h.num_layers // 2 + self.num_decoder_layers = h.num_layers - self.num_encoder_layers + self.blocks = nn.ModuleList([ + Block(h.model_dim, h.num_heads, h.num_kv_heads, h.mlp_mult, h.rope_base, + h.qk_gain_init, h.train_seq_len, layer_idx=i, ln_scale=h.ln_scale, + parallel_residual=(i >= h.parallel_residual_start)) + for i in range(h.num_layers) + ]) + if h.rope_dims > 0: + head_dim = h.model_dim // h.num_heads + for block in self.blocks: + block.attn.rope_dims = h.rope_dims + block.attn.rotary = Rotary(head_dim, base=h.rope_base, train_seq_len=h.train_seq_len, rope_dims=h.rope_dims) + self.final_norm = RMSNorm() + self.lm_head = None if h.tie_embeddings else CastedLinear(h.embedding_dim, h.vocab_size, bias=False) + if self.lm_head is not None: + self.lm_head._zero_init = True + if h.xsa_last_n > 0: + for i in range(max(0, h.num_layers - h.xsa_last_n), h.num_layers): + self.blocks[i].attn.use_xsa = True + + # Layer looping + self.looping_active: bool = False + if h.num_loops > 0: + loop_seg = list(range(h.loop_start, h.loop_end + 1)) + all_indices = list(range(h.loop_start)) + for _ in range(h.num_loops + 1): + all_indices.extend(loop_seg) + all_indices.extend(range(h.loop_end + 1, h.num_layers)) + num_enc = len(all_indices) // 2 + self.encoder_indices: list[int] = all_indices[:num_enc] + self.decoder_indices: list[int] = all_indices[num_enc:] + else: + self.encoder_indices = list(range(self.num_encoder_layers)) + self.decoder_indices = list(range(self.num_encoder_layers, h.num_layers)) + self.num_skip_weights = min(len(self.encoder_indices), len(self.decoder_indices)) + self.skip_weights = nn.Parameter(torch.ones(self.num_skip_weights, h.model_dim, dtype=torch.float32)) + self.skip_gates = nn.Parameter(torch.zeros(self.num_skip_weights, h.model_dim, dtype=torch.float32)) if h.skip_gates_enabled else None + + # MTP auxiliary heads (training-only). Each head is a D->D transform sharing the + # tied embedding for final vocab projection. + self.mtp_heads = h.mtp_heads + self.mtp_weight = h.mtp_weight + if self.mtp_heads > 0: + self.mtp_projs = nn.ModuleList([ + CastedLinear(h.embedding_dim, h.embedding_dim, bias=False) + for _ in range(self.mtp_heads) + ]) + else: + self.mtp_projs = None + + self._init_weights() + + def _init_weights(self) -> None: + if self.tie_embeddings: + nn.init.normal_(self.tok_emb.weight, mean=0.0, std=self.tied_embed_init_std) + for name, module in self.named_modules(): + if isinstance(module, nn.Linear): + if getattr(module, "_zero_init", False): + nn.init.zeros_(module.weight) + elif (module.weight.ndim == 2 and module.weight.shape[0] >= 64 and + module.weight.shape[1] >= 64): + nn.init.orthogonal_(module.weight, gain=1.0) + + def _forward_trunk(self, input_ids: Tensor) -> Tensor: + x = self.tok_emb(input_ids) + x = F.rms_norm(x, (x.size(-1),)) + if self.embed_proj is not None: + x = self.embed_proj(x) + x0 = x + skips: list[Tensor] = [] + enc_iter = self.encoder_indices if self.looping_active else range(self.num_encoder_layers) + dec_iter = self.decoder_indices if self.looping_active else range(self.num_encoder_layers, self.num_encoder_layers + self.num_decoder_layers) + for i in enc_iter: + x = self.blocks[i](x, x0) + skips.append(x) + for skip_idx, i in enumerate(dec_iter): + if skip_idx < self.num_skip_weights and skips: + scaled_skip = self.skip_weights[skip_idx].to(dtype=x.dtype)[None, None, :] * skips.pop() + if self.skip_gates is not None: + g = torch.sigmoid(self.skip_gates[skip_idx].to(dtype=x.dtype))[None, None, :] + x = torch.lerp(scaled_skip, x, g) + else: + x = x + scaled_skip + x = self.blocks[i](x, x0) + return self.final_norm(x) + + def _project_logits(self, h: Tensor) -> Tensor: + if self.head_proj is not None: + h = self.head_proj(h) + if self.tie_embeddings: + logits = F.linear(h, self.tok_emb.weight) + else: + logits = self.lm_head(h) + return self.logit_softcap * torch.tanh(logits / self.logit_softcap) + + def forward_logits(self, input_ids: Tensor) -> Tensor: + return self._project_logits(self._forward_trunk(input_ids)) + + def forward(self, input_ids: Tensor, target_ids: Tensor) -> Tensor: + h = self._forward_trunk(input_ids) + primary_logits = self._project_logits(h) + primary_loss = F.cross_entropy( + primary_logits.reshape(-1, primary_logits.size(-1)).float(), + target_ids.reshape(-1), reduction="mean") + if self.mtp_projs is None or not self.training: + return primary_loss + total = primary_loss + # MTP aux heads: head k predicts token at t+1+(k+1). target_ids is already shifted + # by 1, so to predict t+1+k+1 we shift target_ids by (k+1) more positions. + for k, proj in enumerate(self.mtp_projs): + aux_logits = self._project_logits(proj(h)) + shift = k + 1 + if shift >= aux_logits.size(1): + continue + aux_logits = aux_logits[:, :-shift, :] + aux_targets = target_ids[:, shift:] + aux_loss = F.cross_entropy( + aux_logits.reshape(-1, aux_logits.size(-1)).float(), + aux_targets.reshape(-1), reduction="mean") + total = total + (self.mtp_weight / self.mtp_heads) * aux_loss + return total + + +def classify_param(name: str) -> str: + if "tok_emb" in name or "lm_head" in name: + return "embed" + if ".mlp." in name: + return "mlp" + if ".attn." in name or (".proj." in name and ".mlp." not in name): + return "attn" + return "other" + +# ---------------------------------------- +# Optimization +# ---------------------------------------- + +@torch.compile +def zeropower_via_newtonschulz5(G: Tensor, steps: int = 10, eps: float = 1e-7) -> Tensor: + a, b, c = (3.4445, -4.7750, 2.0315) + X = G.bfloat16() + X /= X.norm() + eps + transposed = G.size(0) > G.size(1) + if transposed: + X = X.T + for _ in range(steps): + A = X @ X.T + B = b * A + c * A @ A + X = a * X + B @ X + return X.T if transposed else X + + +class Muon(torch.optim.Optimizer): + def __init__(self, params, lr: float, momentum: float, backend_steps: int, + nesterov: bool = True, weight_decay: float = 0.0, + row_normalize: bool = False): + super().__init__( + params, + dict(lr=lr, momentum=momentum, backend_steps=backend_steps, + nesterov=nesterov, weight_decay=weight_decay, + row_normalize=row_normalize), + ) + + @torch.no_grad() + def step(self, closure=None): + loss = None + if closure is not None: + with torch.enable_grad(): + loss = closure() + distributed = dist.is_available() and dist.is_initialized() + world_size = dist.get_world_size() if distributed else 1 + rank = dist.get_rank() if distributed else 0 + for group in self.param_groups: + params = group["params"] + if not params: + continue + lr = group["lr"] + momentum = group["momentum"] + backend_steps = group["backend_steps"] + nesterov = group["nesterov"] + total_params = sum(int(p.numel()) for p in params) + updates_flat = torch.zeros(total_params, device=params[0].device, dtype=torch.bfloat16) + curr = 0 + for i, p in enumerate(params): + if i % world_size == rank and p.grad is not None: + g = p.grad + state = self.state[p] + if "momentum_buffer" not in state: + state["momentum_buffer"] = torch.zeros_like(g) + buf = state["momentum_buffer"] + buf.mul_(momentum).add_(g) + if nesterov: + g = g.add(buf, alpha=momentum) + if group.get("row_normalize", False): + row_norms = g.float().norm(dim=-1, keepdim=True).clamp_min(1e-07) + g = g / row_norms.to(g.dtype) + g = zeropower_via_newtonschulz5(g, steps=backend_steps) + g *= max(1, g.size(0) / g.size(1)) ** 0.5 + updates_flat[curr : curr + p.numel()] = g.reshape(-1) + curr += p.numel() + if distributed: + dist.all_reduce(updates_flat, op=dist.ReduceOp.SUM) + wd = group.get("weight_decay", 0.0) + curr = 0 + for p in params: + if wd > 0.0: + p.data.mul_(1.0 - lr * wd) + g = updates_flat[curr : curr + p.numel()].view_as(p).to(dtype=p.dtype) + p.add_(g, alpha=-lr) + curr += p.numel() + return loss + + +CONTROL_TENSOR_NAME_PATTERNS = tuple( + pattern + for pattern in os.environ.get( + "CONTROL_TENSOR_NAME_PATTERNS", + "attn_scale,attn_scales,mlp_scale,mlp_scales,resid_mix,resid_mixes,q_gain,skip_weight,skip_weights,skip_gates", + ).split(",") + if pattern +) + + +class Optimizers(): + def __init__(self, h: Hyperparameters, base_model: GPT): + block_named_params = list(base_model.blocks.named_parameters()) + matrix_params = [ + p + for name, p in block_named_params + if p.ndim == 2 and not any(pattern in name for pattern in + CONTROL_TENSOR_NAME_PATTERNS) + ] + scalar_params = [ + p + for name, p in block_named_params + if p.ndim < 2 or any(pattern in name for pattern in + CONTROL_TENSOR_NAME_PATTERNS) + ] + if base_model.skip_weights.numel() > 0: + scalar_params.append(base_model.skip_weights) + if base_model.skip_gates is not None and base_model.skip_gates.numel() > 0: + scalar_params.append(base_model.skip_gates) + # MTP aux projection matrices — optimized same as other matrices (Muon). + if base_model.mtp_projs is not None: + for p in base_model.mtp_projs.parameters(): + if p.ndim == 2: + matrix_params.append(p) + + token_lr = h.tied_embed_lr if h.tie_embeddings else h.embed_lr + tok_params = [{"params": [base_model.tok_emb.weight], "lr": token_lr, "base_lr": token_lr}] + self.optimizer_tok = torch.optim.AdamW( + tok_params, + betas=(h.beta1, h.beta2), + eps=h.adam_eps, + weight_decay=h.embed_wd, + fused=True, + ) + self.optimizer_muon = Muon( + matrix_params, + lr=h.matrix_lr, + momentum=h.muon_momentum, + backend_steps=h.muon_backend_steps, + weight_decay=h.muon_wd, + row_normalize=h.muon_row_normalize, + ) + for group in self.optimizer_muon.param_groups: + group["base_lr"] = h.matrix_lr + self.optimizer_scalar = torch.optim.AdamW( + [{"params": scalar_params, "lr": h.scalar_lr, "base_lr": h.scalar_lr}], + betas=(h.beta1, h.beta2), + eps=h.adam_eps, + weight_decay=h.adam_wd, + fused=True, + ) + self.optimizers = [self.optimizer_tok, self.optimizer_muon, self.optimizer_scalar] + if base_model.lm_head is not None: + self.optimizer_head = torch.optim.Adam( + [{"params": [base_model.lm_head.weight], "lr": h.head_lr, "base_lr": h.head_lr}], + betas=(h.beta1, h.beta2), + eps=h.adam_eps, + fused=True, + ) + self.optimizers.insert(1, self.optimizer_head) + else: + self.optimizer_head = None + + def __iter__(self): + return iter(self.optimizers) + + def zero_grad_all(self) -> None: + for opt in self.optimizers: + opt.zero_grad(set_to_none=True) + + def step(self): + for opt in self.optimizers: + opt.step() + self.zero_grad_all() + +# ---------------------------------------- +# Quantization +# ---------------------------------------- + +def restore_fp32_params(model: nn.Module) -> None: + for module in model.modules(): + if isinstance(module, CastedLinear): + module.float() + for name, param in model.named_parameters(): + if (param.ndim < 2 or any(pattern in name for pattern in CONTROL_TENSOR_NAME_PATTERNS)) and param.dtype != torch.float32: + param.data = param.data.float() + + +def collect_hessians( + model: nn.Module, + train_loader: ShuffledSequenceLoader, + h: Hyperparameters, + device: torch.device, + n_calibration_batches: int = 64, +) -> dict[str, Tensor]: + hessians: dict[str, Tensor] = {} + hooks = [] + + def make_hook(name: str): + def hook_fn(module, inp, out): + x = inp[0].detach().float() + if x.ndim == 3: + x = x.reshape(-1, x.shape[-1]) + if name not in hessians: + hessians[name] = torch.zeros( + x.shape[1], x.shape[1], dtype=torch.float32, device=device + ) + hessians[name].addmm_(x.T, x) + return hook_fn + + for name, module in model.named_modules(): + if isinstance(module, CastedLinear) and module.weight.numel() > 65536: + cat = classify_param(name + ".weight") + if cat in ("mlp", "attn"): + hooks.append(module.register_forward_hook(make_hook(name + ".weight"))) + + if model.tie_embeddings: + hook_module = model.head_proj if model.head_proj is not None else model.final_norm + def make_output_hook(name: str): + def hook_fn(module, inp, out): + x = out.detach().float() + if x.ndim == 3: + x = x.reshape(-1, x.shape[-1]) + if name not in hessians: + hessians[name] = torch.zeros( + x.shape[1], x.shape[1], dtype=torch.float32, device=device + ) + hessians[name].addmm_(x.T, x) + return hook_fn + hooks.append(hook_module.register_forward_hook(make_output_hook("tok_emb.weight"))) + + model.eval() + with torch.no_grad(): + for _ in range(n_calibration_batches): + x, _ = train_loader.next_batch(h.train_batch_tokens, h.grad_accum_steps) + model.forward_logits(x) + + for hook in hooks: + hook.remove() + + for name in hessians: + hessians[name] = hessians[name].cpu() / n_calibration_batches + + return hessians + + +def gptq_quantize_weight( + w: Tensor, + H: Tensor, + clip_sigmas: float = 3.0, + clip_range: int = 63, + block_size: int = 128, +) -> tuple[Tensor, Tensor]: + W_orig = w.float().clone() + rows, cols = W_orig.shape + H = H.float().clone() + + dead = torch.diag(H) == 0 + H[dead, dead] = 1 + damp = 0.01 * H.diag().mean() + H.diagonal().add_(damp) + + perm = torch.argsort(H.diag(), descending=True) + invperm = torch.argsort(perm) + W_perm = W_orig[:, perm].clone() + W_perm[:, dead[perm]] = 0 + H = H[perm][:, perm] + + Hinv = torch.cholesky_inverse(torch.linalg.cholesky(H)) + Hinv = torch.linalg.cholesky(Hinv, upper=True) + + row_std = W_orig.std(dim=1) + s = (clip_sigmas * row_std / clip_range).clamp_min(1e-10).to(torch.float16) + sf = s.float() + + Q = torch.zeros(rows, cols, dtype=torch.int8) + W_work = W_perm.clone() + for i1 in range(0, cols, block_size): + i2 = min(i1 + block_size, cols) + W_block = W_work[:, i1:i2].clone() + Hinv_block = Hinv[i1:i2, i1:i2] + Err = torch.zeros(rows, i2 - i1) + for j in range(i2 - i1): + w_col = W_block[:, j] + d = Hinv_block[j, j] + q_col = torch.clamp(torch.round(w_col / sf), -clip_range, clip_range) + Q[:, i1 + j] = q_col.to(torch.int8) + err = (w_col - q_col.float() * sf) / d + Err[:, j] = err + W_block[:, j:] -= err.unsqueeze(1) * Hinv_block[j, j:].unsqueeze(0) + if i2 < cols: + W_work[:, i2:] -= Err @ Hinv[i1:i2, i2:] + + return Q[:, invperm], s + + +def gptq_mixed_quantize( + state_dict: dict[str, Tensor], + hessians: dict[str, Tensor], + h: Hyperparameters, +) -> tuple[dict[str, Tensor], dict[str, object]]: + result: dict[str, Tensor] = {} + meta: dict[str, object] = {} + + for name, tensor in state_dict.items(): + t = tensor.detach().cpu().contiguous() + if not t.is_floating_point() or t.numel() <= 65536: + result[name] = t.to(torch.float16) if t.is_floating_point() else t + meta[name] = "passthrough (float16)" + continue + cs = h.embed_clip_sigmas if "tok_emb" in name else h.matrix_clip_sigmas + bits = h.embed_bits if "tok_emb" in name else h.matrix_bits + q, s = gptq_quantize_weight( + t, hessians[name], clip_sigmas=cs, clip_range=2**(bits - 1) - 1) + result[name + ".q"] = q + result[name + ".scale"] = s + meta[name] = f"gptq (int{bits})" + + categories = collections.defaultdict(set) + for name, cat in meta.items(): + short = re.sub(r'\.\d+$', '', re.sub(r'blocks\.\d+', 'blocks', name)) + categories[cat].add(short) + log("Quantized weights:") + for cat in sorted(categories): + log(f" {cat}: {', '.join(sorted(categories[cat]))}") + + return result, meta + + +def dequantize_mixed(result: dict[str, Tensor], meta: dict[str, object], + template_sd: dict[str, Tensor]) -> dict[str, Tensor]: + out: dict[str, Tensor] = {} + for name, orig in template_sd.items(): + info = meta.get(name) + if info is None: + continue + orig_dtype = orig.dtype + if "passthrough" in info: + t = result[name] + if t.dtype == torch.float16 and orig_dtype in (torch.float32, torch.bfloat16): + t = t.to(orig_dtype) + out[name] = t + continue + q, s = result[name + ".q"], result[name + ".scale"] + if s.ndim > 0: + out[name] = (q.float() * s.float().view(q.shape[0], *([1] * (q.ndim - 1)))).to(orig_dtype) + else: + out[name] = (q.float() * float(s.item())).to(orig_dtype) + return out + + +_BSHF_MAGIC = b"BSHF" + + +def _byte_shuffle(data: bytes, stride: int = 2) -> bytes: + if stride <= 1 or len(data) < stride: + return data + src = np.frombuffer(data, dtype=np.uint8) + n = len(src) + out = np.empty(n, dtype=np.uint8) + dest_off = 0 + for pos in range(stride): + chunk = src[pos::stride] + out[dest_off:dest_off + len(chunk)] = chunk + dest_off += len(chunk) + return _BSHF_MAGIC + bytes([stride]) + out.tobytes() + + +def _byte_unshuffle(data: bytes) -> bytes: + if len(data) < 5 or data[:4] != _BSHF_MAGIC: + return data + stride = data[4] + if stride < 2: + return data[5:] + payload = np.frombuffer(data, dtype=np.uint8, offset=5) + n = len(payload) + out = np.empty(n, dtype=np.uint8) + src_off = 0 + for pos in range(stride): + chunk_len = n // stride + (1 if pos < n % stride else 0) + out[pos::stride][:chunk_len] = payload[src_off:src_off + chunk_len] + src_off += chunk_len + return out.tobytes() + + +def _compress(data: bytes, compressor: str) -> bytes: + data = _byte_shuffle(data) + if compressor == "lzma": + return lzma.compress(data, preset=6) + elif compressor == "brotli": + import brotli + return brotli.compress(data, quality=11) + raise ValueError(f"Unknown compressor: {compressor!r}") + + +def _decompress(data: bytes, compressor: str) -> bytes: + if compressor == "lzma": + raw = lzma.decompress(data) + elif compressor == "brotli": + import brotli + raw = brotli.decompress(data) + else: + raise ValueError(f"Unknown compressor: {compressor!r}") + raw = _byte_unshuffle(raw) + return raw + + +def serialize(h: Hyperparameters, base_model: torch.nn.Module, code: str) -> tuple[int, int]: + code_bytes = len(code.encode("utf-8")) + if h.is_main_process: + torch.save(base_model.state_dict(), h.model_path) + model_bytes = os.path.getsize(h.model_path) + log(f"Serialized model: {model_bytes} bytes") + log(f"Code size: {code_bytes} bytes") + + # MTP aux heads are training-only — drop them before serialization so they don't + # eat into the 16MB artifact budget. + sd_cpu = {k: v.detach().cpu() for k, v in base_model.state_dict().items() + if not k.startswith("mtp_projs.")} + device = torch.device("cuda", h.local_rank) + log("GPTQ:collecting Hessians from calibration data...") + t0 = time.perf_counter() + calib_loader = ShuffledSequenceLoader(h, device) + hessians = collect_hessians( + base_model, calib_loader, h, device, + n_calibration_batches=h.gptq_calibration_batches, + ) + log(f"GPTQ:collected {len(hessians)} Hessians in {time.perf_counter() - t0:.1f}s") + quant_result, quant_meta = gptq_mixed_quantize(sd_cpu, hessians, h) + + quant_buf = io.BytesIO() + torch.save({"w": quant_result, "m": quant_meta}, quant_buf) + quant_raw = quant_buf.getvalue() + quant_blob = _compress(quant_raw, h.compressor) + quant_file_bytes = len(quant_blob) + bytes_total = quant_file_bytes + code_bytes + if h.is_main_process: + with open(h.quantized_model_path, "wb") as f: + f.write(quant_blob) + log(f"Serialized model quantized+{h.compressor}: {quant_file_bytes} bytes") + log(f"Total submission size quantized+{h.compressor}: {bytes_total} bytes") + return bytes_total, quant_file_bytes + + +def deserialize(h: Hyperparameters, device: torch.device) -> GPT: + eval_model = GPT(h).to(device).bfloat16() + # MTP aux heads are unused at eval; drop them to ensure clean load + no compute overhead. + if eval_model.mtp_projs is not None: + eval_model.mtp_projs = None + eval_model.mtp_heads = 0 + restore_fp32_params(eval_model) + sd_cpu = {k: v.detach().cpu() for k, v in eval_model.state_dict().items()} + + with open(h.quantized_model_path, "rb") as f: + quant_blob_disk = f.read() + quant_state = torch.load( + io.BytesIO(_decompress(quant_blob_disk, h.compressor)), + map_location="cpu", + ) + deq_state = dequantize_mixed(quant_state["w"], quant_state["m"], sd_cpu) + eval_model.load_state_dict(deq_state, strict=True) + + return eval_model + +# ---------------------------------------- +# Evaluation +# ---------------------------------------- + +def _loss_bpb(loss_sum, token_count, byte_count) -> tuple[float, float]: + val_loss = (loss_sum / token_count).item() + val_bpb = val_loss / math.log(2.0) * (token_count.item() / byte_count.item()) + return val_loss, val_bpb + + +def eval_val( + h: Hyperparameters, + device: torch.device, + val_data: ValidationData, + model: nn.Module +) -> tuple[float, float]: + seq_len = h.eval_seq_len + local_batch_tokens = h.val_batch_tokens // (h.world_size * h.grad_accum_steps) + if local_batch_tokens < seq_len: + raise ValueError( + "VAL_BATCH_SIZE must provide at least one sequence per rank; " + f"got VAL_BATCH_SIZE={h.val_batch_tokens}, WORLD_SIZE={h.world_size}, " + f"GRAD_ACCUM_STEPS={h.grad_accum_steps}, seq_len={seq_len}" + ) + local_batch_seqs = local_batch_tokens // seq_len + total_seqs = (val_data.val_tokens.numel() - 1) // seq_len + seq_start = (total_seqs * h.rank) // h.world_size + seq_end = (total_seqs * (h.rank + 1)) // h.world_size + val_loss_sum = torch.zeros((), device=device, dtype=torch.float64) + val_token_count = torch.zeros((), device=device, dtype=torch.float64) + val_byte_count = torch.zeros((), device=device, dtype=torch.float64) + + model.eval() + # Use no_grad instead of inference_mode: inference_mode taints cached rotary + # tensors and makes subsequent TTT (which needs autograd) crash. + with torch.no_grad(): + for batch_seq_start in range(seq_start, seq_end, local_batch_seqs): + batch_seq_end = min(batch_seq_start + local_batch_seqs, seq_end) + raw_start = batch_seq_start * seq_len + raw_end = batch_seq_end * seq_len + 1 + local = val_data.val_tokens[raw_start:raw_end].to( + device=device, dtype=torch.int64, non_blocking=True) + x = local[:-1].reshape(-1, seq_len) + y = local[1:].reshape(-1, seq_len) + with torch.autocast(device_type="cuda", dtype=torch.bfloat16, enabled=True): + batch_loss = model(x, y).detach() + batch_token_count = float(y.numel()) + val_loss_sum += batch_loss.to(torch.float64) * batch_token_count + val_token_count += batch_token_count + prev_ids = x.reshape(-1) + tgt_ids = y.reshape(-1) + token_bytes = val_data.base_bytes_lut[tgt_ids].to(dtype=torch.int16) + token_bytes += (val_data.has_leading_space_lut[tgt_ids] & + ~val_data.is_boundary_token_lut[prev_ids]).to(dtype=torch.int16) + val_byte_count += token_bytes.to(torch.float64).sum() + + if dist.is_available() and dist.is_initialized(): + dist.all_reduce(val_loss_sum, op=dist.ReduceOp.SUM) + dist.all_reduce(val_token_count, op=dist.ReduceOp.SUM) + dist.all_reduce(val_byte_count, op=dist.ReduceOp.SUM) + + model.train() + return _loss_bpb(val_loss_sum, val_token_count, val_byte_count) + + +def eval_val_sliding( + h: Hyperparameters, + device: torch.device, + val_data: ValidationData, + base_model: nn.Module, + batch_seqs: int = 32 +) -> tuple[float, float]: + base_model.eval() + logits_fn = torch.compile(base_model.forward_logits, dynamic=False, fullgraph=True) + + seq_len = h.eval_seq_len + context_size = seq_len - h.eval_stride + total_tokens = val_data.val_tokens.numel() - 1 + + window_starts = [ws for ws in range(0, total_tokens, h.eval_stride) + if ws + context_size < total_tokens] + + total_windows = len(window_starts) + my_s = (total_windows * h.rank) // h.world_size + my_e = (total_windows * (h.rank + 1)) // h.world_size + my_windows = window_starts[my_s:my_e] + + loss_sum = torch.zeros((), device=device, dtype=torch.float64) + token_count = torch.zeros((), device=device, dtype=torch.float64) + byte_count = torch.zeros((), device=device, dtype=torch.float64) + + # Use no_grad: inference_mode taints cached rotary tensors across the + # subsequent TTT eval which needs autograd. + with torch.no_grad(): + for bi in range(0, len(my_windows), batch_seqs): + batch_ws = my_windows[bi:bi + batch_seqs] + bsz = len(batch_ws) + + x_batch = torch.zeros(bsz, seq_len, dtype=torch.int64, device=device) + y_batch = torch.zeros(bsz, seq_len, dtype=torch.int64, device=device) + wlens: list[int] = [] + + for i, ws in enumerate(batch_ws): + we = min(ws + seq_len, total_tokens) + wlen = we - ws + wlens.append(wlen) + chunk = val_data.val_tokens[ws:we + 1].to(dtype=torch.int64, device=device) + x_batch[i, :wlen] = chunk[:-1] + y_batch[i, :wlen] = chunk[1:] + + with torch.autocast(device_type="cuda", dtype=torch.bfloat16): + logits = logits_fn(x_batch) + + nll = F.cross_entropy( + logits.reshape(-1, logits.size(-1)).float(), + y_batch.reshape(-1), + reduction="none", + ).reshape(bsz, seq_len) + + for i, ws in enumerate(batch_ws): + wlen = wlens[i] + s = 0 if ws == 0 else context_size + scored_nll = nll[i, s:wlen].to(torch.float64) + loss_sum += scored_nll.sum() + token_count += float(wlen - s) + tgt = y_batch[i, s:wlen] + prev = x_batch[i, s:wlen] + tb = val_data.base_bytes_lut[tgt].to(torch.float64) + tb += (val_data.has_leading_space_lut[tgt] & + ~val_data.is_boundary_token_lut[prev]).to(torch.float64) + byte_count += tb.sum() + + if dist.is_available() and dist.is_initialized(): + dist.all_reduce(loss_sum, op=dist.ReduceOp.SUM) + dist.all_reduce(token_count, op=dist.ReduceOp.SUM) + dist.all_reduce(byte_count, op=dist.ReduceOp.SUM) + + base_model.train() + return _loss_bpb(loss_sum, token_count, byte_count) + + +def eval_val_sliding_ttt( + h: Hyperparameters, + device: torch.device, + val_data: ValidationData, + base_model: nn.Module, + batch_seqs: int = 32, +) -> tuple[float, float]: + """Legal Score-First TTT (Issue #1017 compliant): + For each chunk of the val stream: (1) score all sliding windows whose final + position falls in the chunk under torch.no_grad (this is what counts toward BPB), + (2) SGD-train on the scored chunk tokens. Last chunk is score-only. + + PR #1413 recipe: SGD(lr=0.005, momentum=0.9), 3 epochs per 32K-token chunk, + cosine LR decay across chunks, freeze first N blocks, grad clip 1.0. + """ + seq_len = h.eval_seq_len + total_tokens = val_data.val_tokens.numel() - 1 + context_size = seq_len - h.eval_stride + stride = h.eval_stride + + # Build sliding window starts exactly as eval_val_sliding does. + window_starts = [ws for ws in range(0, total_tokens, stride) + if ws + context_size < total_tokens] + + # Chunk by the position of the last-scored token in each window. + # For ws==0 we score the full seq; otherwise we score positions [context_size..wlen). + # The chunk index is based on the END of the scored region. + num_chunks = (total_tokens + h.ttt_chunk - 1) // h.ttt_chunk + chunk_windows: list[list[int]] = [[] for _ in range(num_chunks)] + for ws in window_starts: + end_pos = min(ws + seq_len, total_tokens) + ci = min((end_pos - 1) // h.ttt_chunk, num_chunks - 1) + chunk_windows[ci].append(ws) + + # Freeze first N blocks; unfreeze the rest for TTT. + frozen_ids = set(range(min(h.ttt_freeze_blocks, len(base_model.blocks)))) + ttt_params: list[Tensor] = [] + for name, p in base_model.named_parameters(): + # Match "blocks.{bi}." prefix (and only those). + frozen = any(name.startswith(f"blocks.{bi}.") for bi in frozen_ids) + p.requires_grad_(not frozen) + if not frozen: + ttt_params.append(p) + optimizer = torch.optim.SGD(ttt_params, lr=h.ttt_lr, momentum=h.ttt_momentum) + + loss_sum = torch.zeros((), device=device, dtype=torch.float64) + token_count = torch.zeros((), device=device, dtype=torch.float64) + byte_count = torch.zeros((), device=device, dtype=torch.float64) + + # Distribute chunks across ranks (each rank scores its windows under no_grad, + # then trains locally on its chunk — no cross-rank param sync needed for correctness). + for ci in range(num_chunks): + windows = chunk_windows[ci] + # Split windows across ranks + my_s = (len(windows) * h.rank) // h.world_size + my_e = (len(windows) * (h.rank + 1)) // h.world_size + my_windows = windows[my_s:my_e] + + # SCORE (no grad, counts toward BPB) + base_model.eval() + with torch.no_grad(): + for bi in range(0, len(my_windows), batch_seqs): + batch_ws = my_windows[bi:bi + batch_seqs] + bsz = len(batch_ws) + if bsz == 0: + continue + x_batch = torch.zeros(bsz, seq_len, dtype=torch.int64, device=device) + y_batch = torch.zeros(bsz, seq_len, dtype=torch.int64, device=device) + wlens: list[int] = [] + for i, ws in enumerate(batch_ws): + we = min(ws + seq_len, total_tokens) + wlen = we - ws + wlens.append(wlen) + chunk = val_data.val_tokens[ws:we + 1].to(dtype=torch.int64, device=device) + x_batch[i, :wlen] = chunk[:-1] + y_batch[i, :wlen] = chunk[1:] + with torch.autocast(device_type="cuda", dtype=torch.bfloat16): + logits = base_model.forward_logits(x_batch) + nll = F.cross_entropy( + logits.reshape(-1, logits.size(-1)).float(), + y_batch.reshape(-1), + reduction="none", + ).reshape(bsz, seq_len) + for i, ws in enumerate(batch_ws): + wlen = wlens[i] + s = 0 if ws == 0 else context_size + scored_nll = nll[i, s:wlen].to(torch.float64) + loss_sum += scored_nll.sum() + token_count += float(wlen - s) + tgt = y_batch[i, s:wlen] + prev = x_batch[i, s:wlen] + tb = val_data.base_bytes_lut[tgt].to(torch.float64) + tb += (val_data.has_leading_space_lut[tgt] & + ~val_data.is_boundary_token_lut[prev]).to(torch.float64) + byte_count += tb.sum() + + # TRAIN on the scored chunk (skip last chunk) + is_last = (ci == num_chunks - 1) + if is_last or h.ttt_epochs <= 0: + continue + chunk_start = ci * h.ttt_chunk + chunk_end = min((ci + 1) * h.ttt_chunk, total_tokens) + chunk_seqs = (chunk_end - chunk_start) // seq_len + if chunk_seqs <= 0: + continue + # Shard chunk seqs across ranks for training. + # CRITICAL: use even sharding (same count per rank) so all ranks execute + # the SAME number of optimizer steps -> the all_reduce below never hangs. + seqs_per_rank = chunk_seqs // max(h.world_size, 1) + if seqs_per_rank <= 0: + continue # deterministic skip: depends only on chunk_seqs+world_size + my_s = h.rank * seqs_per_rank + my_chunk_seqs = list(range(my_s, my_s + seqs_per_rank)) + # Cosine LR decay across chunks + cos_lr = h.ttt_lr * 0.5 * (1.0 + math.cos(math.pi * ci / max(num_chunks - 1, 1))) + for pg in optimizer.param_groups: + pg['lr'] = cos_lr + + base_model.train() + is_dist = dist.is_available() and dist.is_initialized() and h.world_size > 1 + for _ep in range(h.ttt_epochs): + for bs in range(0, len(my_chunk_seqs), h.ttt_batch_seqs): + be = min(bs + h.ttt_batch_seqs, len(my_chunk_seqs)) + seq_idxs = my_chunk_seqs[bs:be] + if not seq_idxs: + continue + start_tok = chunk_start + seq_idxs[0] * seq_len + end_tok = chunk_start + (seq_idxs[-1] + 1) * seq_len + 1 + if end_tok > val_data.val_tokens.numel(): + continue + local = val_data.val_tokens[start_tok:end_tok].to(device=device, dtype=torch.int64) + x = local[:-1].reshape(-1, seq_len) + y = local[1:].reshape(-1, seq_len) + optimizer.zero_grad(set_to_none=True) + with torch.autocast(device_type="cuda", dtype=torch.bfloat16): + loss = base_model(x, y) + loss.backward() + # CRITICAL: all-reduce grads across ranks so each optimizer step moves + # every rank's model identically (otherwise 8 ranks diverge into 8 models + # and subsequent chunk scoring becomes inconsistent). + if is_dist: + for p in ttt_params: + if p.grad is not None: + dist.all_reduce(p.grad, op=dist.ReduceOp.SUM) + p.grad.div_(h.world_size) + if h.ttt_grad_clip > 0: + torch.nn.utils.clip_grad_norm_(ttt_params, h.ttt_grad_clip) + optimizer.step() + + if dist.is_available() and dist.is_initialized(): + dist.all_reduce(loss_sum, op=dist.ReduceOp.SUM) + dist.all_reduce(token_count, op=dist.ReduceOp.SUM) + dist.all_reduce(byte_count, op=dist.ReduceOp.SUM) + + # Restore grad state + for p in base_model.parameters(): + p.requires_grad_(True) + base_model.eval() + return _loss_bpb(loss_sum, token_count, byte_count) + + +def timed_eval(label: str, fn, *args, **kwargs) -> tuple[float, float]: + torch.cuda.synchronize() + t0 = time.perf_counter() + val_loss, val_bpb = fn(*args, **kwargs) + torch.cuda.synchronize() + elapsed_ms = 1000.0 * (time.perf_counter() - t0) + log(f"{label} val_loss:{val_loss:.8f} val_bpb:{val_bpb:.8f} eval_time:{elapsed_ms:.0f}ms") + return val_loss, val_bpb + + +# ----------------------------- +# Training +# ----------------------------- + +def train_model(h: Hyperparameters, device: torch.device, val_data: ValidationData): + # Set up model + base_model = GPT(h).to(device).bfloat16() + restore_fp32_params(base_model) + compiled_model = torch.compile(base_model, dynamic=False, fullgraph=True) + if h.distributed: + model = DDP(compiled_model, device_ids=[h.local_rank], broadcast_buffers=False) + else: + model = compiled_model + log(f"model_params:{sum(p.numel() for p in base_model.parameters())}") + + # Set up optimizer and load train data + optimizers = Optimizers(h, base_model) + train_loader = ShuffledSequenceLoader(h, device) + + # Helper functions for training + max_wallclock_ms = 1000.0 * h.max_wallclock_seconds if h.max_wallclock_seconds > 0 else None + if max_wallclock_ms is not None: + max_wallclock_ms -= h.gptq_reserve_seconds * 1000.0 + log(f"gptq:reserving {h.gptq_reserve_seconds:.0f}s, effective={max_wallclock_ms:.0f}ms") + + def training_frac(step: int, elapsed_ms: float) -> float: + if max_wallclock_ms is None: + return step / max(h.iterations, 1) + return elapsed_ms / max(max_wallclock_ms, 1e-9) + + def lr_mul(frac: float) -> float: + if h.warmdown_frac <= 0: + return 1.0 + if frac >= 1.0 - h.warmdown_frac: + return max((1.0 - frac) / h.warmdown_frac, h.min_lr) + return 1.0 + + def step_fn(step, lr_scale): + optimizers.zero_grad_all() + train_loss = torch.zeros((), device=device) + for micro_step in range(h.grad_accum_steps): + if h.distributed: + model.require_backward_grad_sync = micro_step == h.grad_accum_steps - 1 + x, y = train_loader.next_batch(h.train_batch_tokens, h.grad_accum_steps) + with torch.autocast(device_type="cuda", dtype=torch.bfloat16, enabled=True): + loss = model(x, y) + train_loss += loss.detach() + (loss / h.grad_accum_steps).backward() + train_loss /= h.grad_accum_steps + + frac = min(step / h.muon_momentum_warmup_steps, 1.0) if h.muon_momentum_warmup_steps > 0 else 1.0 + muon_momentum = (1 - frac) * h.muon_momentum_warmup_start + frac * h.muon_momentum + for group in optimizers.optimizer_muon.param_groups: + group["momentum"] = muon_momentum + + for opt in optimizers: + for group in opt.param_groups: + group["lr"] = group["base_lr"] * lr_scale + + if h.grad_clip_norm > 0: + torch.nn.utils.clip_grad_norm_(base_model.parameters(), h.grad_clip_norm) + + optimizers.step() + return train_loss + + # Model warmup + if h.warmup_steps > 0: + initial_model_state = {name: tensor.detach().cpu().clone() + for name, tensor in base_model.state_dict().items()} + initial_optimizer_states = [copy.deepcopy(opt.state_dict()) for opt in optimizers] + model.train() + for warmup_step in range(h.warmup_steps): + step_fn(warmup_step, 1.0) + if warmup_step <= 5 or (warmup_step + 1) % 10 == 0 or warmup_step + 1 == h.warmup_steps: + log(f"warmup_step: {warmup_step + 1}/{h.warmup_steps}") + if h.num_loops > 0: + base_model.looping_active = True + log(f"loop_warmup:enabled encoder:{base_model.encoder_indices} decoder:{base_model.decoder_indices}") + for warmup_step in range(h.warmup_steps): + step_fn(warmup_step, 1.0) + if warmup_step <= 5 or (warmup_step + 1) % 10 == 0 or warmup_step + 1 == h.warmup_steps: + log(f"loop_warmup_step: {warmup_step + 1}/{h.warmup_steps}") + base_model.looping_active = False + base_model.load_state_dict(initial_model_state, strict=True) + for opt, state in zip(optimizers, initial_optimizer_states, strict=True): + opt.load_state_dict(state) + optimizers.zero_grad_all() + if h.distributed: + model.require_backward_grad_sync = True + train_loader = ShuffledSequenceLoader(h, device) + + # Training loop + ema_state = {name: t.detach().float().clone() for name, t in base_model.state_dict().items()} + ema_decay = h.ema_decay + + training_time_ms = 0.0 + stop_after_step: int | None = None + torch.cuda.synchronize() + t0 = time.perf_counter() + + step = 0 + while True: + last_step = step == h.iterations or (stop_after_step is not None and step >= stop_after_step) + + should_validate = last_step or (h.val_loss_every > 0 and step % h.val_loss_every == 0) + if should_validate: + torch.cuda.synchronize() + training_time_ms += 1000.0 * (time.perf_counter() - t0) + val_loss, val_bpb = eval_val(h, device, val_data, model) + log(f"{step}/{h.iterations} val_loss: {val_loss:.4f} val_bpb: {val_bpb:.4f}") + torch.cuda.synchronize() + t0 = time.perf_counter() + + if last_step: + if stop_after_step is not None and step < h.iterations: + log( + f"stopping_early: wallclock_cap train_time: {training_time_ms:.0f}ms " + f"step: {step}/{h.iterations}" + ) + break + + elapsed_ms = training_time_ms + 1000.0 * (time.perf_counter() - t0) + frac = training_frac(step, elapsed_ms) + scale = lr_mul(frac) + if h.num_loops > 0 and not base_model.looping_active and frac >= h.enable_looping_at: + base_model.looping_active = True + log(f"layer_loop:enabled step:{step} frac:{frac:.3f} encoder:{base_model.encoder_indices} decoder:{base_model.decoder_indices}") + train_loss = step_fn(step, scale) + + with torch.no_grad(): + for name, t in base_model.state_dict().items(): + ema_state[name].mul_(ema_decay).add_(t.detach().float(), alpha=1.0 - ema_decay) + + step += 1 + approx_training_time_ms = training_time_ms + 1000.0 * (time.perf_counter() - t0) + + should_log_train = ( + h.train_log_every > 0 + and (step <= 5 or step % h.train_log_every == 0 or stop_after_step is not None) + ) + if should_log_train: + tok_per_sec = step * h.train_batch_tokens / (approx_training_time_ms / 1000.0) + log( + f"{step}/{h.iterations} train_loss: {train_loss.item():.4f} " + f"train_time: {approx_training_time_ms / 60000:.1f}m tok/s: {tok_per_sec:.0f}" + ) + + reached_cap = max_wallclock_ms is not None and approx_training_time_ms >= max_wallclock_ms + if h.distributed and max_wallclock_ms is not None: + reached_cap_tensor = torch.tensor(int(reached_cap), device=device) + dist.all_reduce(reached_cap_tensor, op=dist.ReduceOp.MAX) + reached_cap = bool(reached_cap_tensor.item()) + if stop_after_step is None and reached_cap: + stop_after_step = step + + log( + f"peak memory allocated: {torch.cuda.max_memory_allocated() // 1024 // 1024} MiB " + f"reserved: {torch.cuda.max_memory_reserved() // 1024 // 1024} MiB" + ) + + # Weight averaging + log("ema:applying EMA weights") + current_state = base_model.state_dict() + avg_state = {name: t.to(dtype=current_state[name].dtype) for name, t in ema_state.items()} + base_model.load_state_dict(avg_state, strict=True) + + return base_model, compiled_model + + +def train_and_eval(h: Hyperparameters, device: torch.device) -> None: + random.seed(h.seed) + np.random.seed(h.seed) + torch.manual_seed(h.seed) + torch.cuda.manual_seed_all(h.seed) + + val_data = ValidationData(h, device) + log(f"train_shards: {len(list(Path(h.datasets_dir).resolve().glob('fineweb_train_*.bin')))}") + log(f"val_tokens: {val_data.val_tokens.numel() - 1}") + + base_model, compiled_model = train_model(h, device, val_data) + torch._dynamo.reset() + timed_eval("pre-quantization post-ema", eval_val, h, device, val_data, compiled_model) + + serialize(h, base_model, Path(__file__).read_text(encoding="utf-8")) + if h.distributed: + dist.barrier() + eval_model = deserialize(h, device) + if h.num_loops > 0: + eval_model.looping_active = True + + compiled_model = torch.compile(eval_model, dynamic=False, fullgraph=True) + timed_eval("quantized", eval_val, h, device, val_data, compiled_model) + if h.sliding_window_enabled: + timed_eval("quantized_sliding_window", eval_val_sliding, h, device, val_data, eval_model) + if h.ttt_enabled: + # TTT mutates model weights. Run AFTER the score-only sliding eval above so + # the unadapted sliding BPB is still reported for comparison. + log(f"ttt:enabled lr={h.ttt_lr} momentum={h.ttt_momentum} epochs={h.ttt_epochs} " + f"chunk={h.ttt_chunk} freeze_blocks={h.ttt_freeze_blocks}") + # Reset any cached rotary buffers so they're autograd-friendly (prior sliding + # eval cached them under no_grad, but safer to rebuild fresh for TTT training). + for block in eval_model.blocks: + if hasattr(block.attn, "rotary") and block.attn.rotary is not None: + if hasattr(block.attn.rotary, "cached_cos"): + block.attn.rotary.cached_cos = None + if hasattr(block.attn.rotary, "cached_sin"): + block.attn.rotary.cached_sin = None + try: + timed_eval("quantized_sliding_window_ttt", eval_val_sliding_ttt, + h, device, val_data, eval_model) + except Exception as e: + log(f"ttt:FAILED {type(e).__name__}: {e}") + log("ttt:falling back — sliding BPB above is the final score for this seed") + if dist.is_available() and dist.is_initialized(): + dist.barrier() + + +def main(): + world_size = int(os.environ.get("WORLD_SIZE", "1")) + local_rank = int(os.environ.get("LOCAL_RANK", "0")) + distributed = "RANK" in os.environ and "WORLD_SIZE" in os.environ + + if not torch.cuda.is_available(): + raise RuntimeError("CUDA is required") + if world_size <= 0: + raise ValueError(f"WORLD_SIZE must be positive, got {world_size}") + if 8 % world_size != 0: + raise ValueError(f"WORLD_SIZE={world_size} must divide 8 so grad_accum_steps stays integral") + + device = torch.device("cuda", local_rank) + torch.cuda.set_device(device) + if distributed: + dist.init_process_group(backend="nccl", device_id=device) + dist.barrier() + + torch.backends.cuda.matmul.allow_tf32 = True + torch.backends.cudnn.allow_tf32 = True + torch.set_float32_matmul_precision("high") + from torch.backends.cuda import enable_cudnn_sdp, enable_flash_sdp, enable_math_sdp, enable_mem_efficient_sdp + + enable_cudnn_sdp(False) + enable_flash_sdp(True) + enable_mem_efficient_sdp(False) + enable_math_sdp(False) + torch._dynamo.config.optimize_ddp = False + + h = Hyperparameters() + set_logging_hparams(h) + if h.is_main_process: + os.makedirs("logs", exist_ok=True) + log(100 * "=", console=False) + log("Hyperparameters:", console=True) + for k, v in sorted(vars(type(h)).items()): + if not k.startswith("_"): + log(f" {k}: {v}", console=True) + log("=" * 100, console=False) + log(f"Running Python {sys.version}", console=False) + log(f"Running PyTorch {torch.__version__}", console=False) + log( + subprocess.run(["nvidia-smi"], stdout=subprocess.PIPE, stderr=subprocess.PIPE, + text=True, check=False).stdout, + console=False, + ) + log("=" * 100, console=False) + + train_and_eval(h, device) + + if distributed: + dist.destroy_process_group() + + +if __name__ == "__main__": + main() diff --git a/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_seed1337.log b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_seed1337.log new file mode 100644 index 0000000000..209124f37f --- /dev/null +++ b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_seed1337.log @@ -0,0 +1,229 @@ +W0422 12:45:52.072000 32785 torch/distributed/run.py:803] +W0422 12:45:52.072000 32785 torch/distributed/run.py:803] ***************************************** +W0422 12:45:52.072000 32785 torch/distributed/run.py:803] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0422 12:45:52.072000 32785 torch/distributed/run.py:803] ***************************************** +Hyperparameters: + adam_eps: 1e-08 + adam_wd: 0.02 + beta1: 0.9 + beta2: 0.95 + compressor: brotli + data_dir: ./data_lsu_shim/ + datasets_dir: ./data_lsu_shim/datasets/fineweb10B_sp8192 + distributed: True + ema_decay: 0.9965 + embed_bits: 8 + embed_clip_sigmas: 20.0 + embed_lr: 0.6 + embed_wd: 0.095 + embedding_dim: 512 + enable_looping_at: 0.35 + eval_seq_len: 2048 + eval_stride: 64 + gptq_calibration_batches: 64 + gptq_reserve_seconds: 12.0 + grad_accum_steps: 1 + grad_clip_norm: 0.3 + head_lr: 0.008 + is_main_process: True + iterations: 20000 + ln_scale: True + local_rank: 0 + logfile: logs/submit_ref_s1337.txt + logit_softcap: 30.0 + loop_end: 5 + loop_start: 3 + matrix_bits: 6 + matrix_clip_sigmas: 12.85 + matrix_lr: 0.022 + max_wallclock_seconds: 600.0 + min_lr: 0.0 + mlp_mult: 4.0 + model_dim: 512 + model_path: final_model.pt + mtp_heads: 0 + mtp_weight: 0.3 + muon_backend_steps: 5 + muon_beta2: 0.95 + muon_momentum: 0.99 + muon_momentum_warmup_start: 0.92 + muon_momentum_warmup_steps: 1500 + muon_row_normalize: True + muon_wd: 0.095 + num_heads: 8 + num_kv_heads: 4 + num_layers: 11 + num_loops: 2 + parallel_residual_start: 7 + qk_gain_init: 5.25 + quantized_model_path: final_model.int6.ptz + rank: 0 + rope_base: 10000.0 + rope_dims: 16 + rope_train_seq_len: 2048 + run_id: submit_ref_s1337 + scalar_lr: 0.02 + seed: 1337 + skip_gates_enabled: True + sliding_window_enabled: True + tie_embeddings: True + tied_embed_init_std: 0.005 + tied_embed_lr: 0.03 + tokenizer_path: ./data_lsu_shim/tokenizers/fineweb_8192_bpe.model + train_batch_tokens: 786432 + train_files: ./data_lsu_shim/datasets/fineweb10B_sp8192/fineweb_train_*.bin + train_log_every: 500 + train_seq_len: 2048 + ttt_batch_seqs: 4 + ttt_chunk: 32768 + ttt_enabled: True + ttt_epochs: 3 + ttt_freeze_blocks: 9 + ttt_grad_clip: 1.0 + ttt_lr: 0.005 + ttt_momentum: 0.9 + val_batch_tokens: 524288 + val_files: ./data_lsu_shim/datasets/fineweb10B_sp8192/fineweb_val_*.bin + val_loss_every: 4000 + vocab_size: 8192 + warmdown_frac: 0.72 + warmup_steps: 20 + world_size: 8 + xsa_last_n: 11 +train_shards: 80 +val_tokens: 40548352 +model_params:35944536 +gptq:reserving 12s, effective=588000ms +warmup_step: 1/20 +warmup_step: 2/20 +warmup_step: 3/20 +warmup_step: 4/20 +warmup_step: 5/20 +warmup_step: 6/20 +warmup_step: 10/20 +warmup_step: 20/20 +loop_warmup:enabled encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +loop_warmup_step: 1/20 +loop_warmup_step: 2/20 +loop_warmup_step: 3/20 +loop_warmup_step: 4/20 +loop_warmup_step: 5/20 +loop_warmup_step: 6/20 +loop_warmup_step: 10/20 +loop_warmup_step: 20/20 +0/20000 val_loss: 9.0047 val_bpb: 3.4867 +1/20000 train_loss: 9.0080 train_time: 0.0m tok/s: 8066570 +2/20000 train_loss: 12.2992 train_time: 0.0m tok/s: 8109891 +3/20000 train_loss: 11.0440 train_time: 0.0m tok/s: 8043550 +4/20000 train_loss: 9.4087 train_time: 0.0m tok/s: 8012978 +5/20000 train_loss: 8.3248 train_time: 0.0m tok/s: 7998156 +500/20000 train_loss: 3.3404 train_time: 0.8m tok/s: 7773362 +1000/20000 train_loss: 3.2133 train_time: 1.7m tok/s: 7764854 +1500/20000 train_loss: 3.1034 train_time: 2.5m tok/s: 7769144 +2000/20000 train_loss: 3.0206 train_time: 3.4m tok/s: 7769320 +layer_loop:enabled step:2034 frac:0.350 encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +2500/20000 train_loss: 3.0034 train_time: 4.6m tok/s: 7149424 +3000/20000 train_loss: 3.0418 train_time: 5.8m tok/s: 6758023 +3500/20000 train_loss: 2.9228 train_time: 7.1m tok/s: 6504182 +4000/20000 train_loss: 2.9591 train_time: 8.3m tok/s: 6326334 +4000/20000 val_loss: 2.8764 val_bpb: 1.1138 +4500/20000 train_loss: 2.7576 train_time: 9.5m tok/s: 6195392 +4614/20000 val_loss: 2.8081 val_bpb: 1.0873 +stopping_early: wallclock_cap train_time: 588121ms step: 4614/20000 +peak memory allocated: 39046 MiB reserved: 39070 MiB +ema:applying EMA weights +pre-quantization post-ema val_loss:2.80491496 val_bpb:1.08609269 eval_time:6363ms +Serialized model: 135431033 bytes +Code size: 72758 bytes +GPTQ:collecting Hessians from calibration data... +GPTQ:collected 67 Hessians in 12.7s +Quantized weights: + gptq (int6): blocks.attn.c_k.weight, blocks.attn.c_q.weight, blocks.attn.c_v.weight, blocks.attn.proj.weight, blocks.mlp.fc.weight, blocks.mlp.proj.weight + gptq (int8): tok_emb.weight + passthrough (float16): blocks.attn.q_gain, blocks.attn_scale, blocks.mlp_scale, blocks.resid_mix, skip_gates, skip_weights +Serialized model quantized+brotli: 15971929 bytes +Total submission size quantized+brotli: 16044687 bytes +quantized val_loss:2.83464461 val_bpb:1.09760432 eval_time:21738ms +quantized_sliding_window val_loss:2.79132641 val_bpb:1.08083105 eval_time:124542ms +ttt:enabled lr=0.005 momentum=0.9 epochs=3 chunk=32768 freeze_blocks=9 +quantized_sliding_window_ttt val_loss:2.78999706 val_bpb:1.08031632 eval_time:423060ms +--------------------------------+------------------------+----------------------+ +| 5 NVIDIA H100 80GB HBM3 On | 00000000:BB:00.0 Off | 0 | +| N/A 34C P0 120W / 700W | 1521MiB / 81559MiB | 0% Default | +| | | Disabled | ++-----------------------------------------+------------------------+----------------------+ +| 6 NVIDIA H100 80GB HBM3 On | 00000000:CB:00.0 Off | 0 | +| N/A 37C P0 117W / 700W | 1521MiB / 81559MiB | 3% Default | +| | | Disabled | ++-----------------------------------------+------------------------+----------------------+ +| 7 NVIDIA H100 80GB HBM3 On | 00000000:DB:00.0 Off | 0 | +| N/A 34C P0 117W / 700W | 1521MiB / 81559MiB | 3% Default | +| | | Disabled | ++-----------------------------------------+------------------------+----------------------+ + ++-----------------------------------------------------------------------------------------+ +| Processes: | +| GPU GI CI PID Type Process name GPU Memory | +| ID ID Usage | +|=========================================================================================| +| No running processes found | ++-----------------------------------------------------------------------------------------+ + +==================================================================================================== +train_shards: 80 +val_tokens: 40548352 +model_params:35944536 +gptq:reserving 12s, effective=588000ms +warmup_step: 1/20 +warmup_step: 2/20 +warmup_step: 3/20 +warmup_step: 4/20 +warmup_step: 5/20 +warmup_step: 6/20 +warmup_step: 10/20 +warmup_step: 20/20 +loop_warmup:enabled encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +loop_warmup_step: 1/20 +loop_warmup_step: 2/20 +loop_warmup_step: 3/20 +loop_warmup_step: 4/20 +loop_warmup_step: 5/20 +loop_warmup_step: 6/20 +loop_warmup_step: 10/20 +loop_warmup_step: 20/20 +0/20000 val_loss: 9.0047 val_bpb: 3.4867 +1/20000 train_loss: 9.0080 train_time: 0.0m tok/s: 8066570 +2/20000 train_loss: 12.2992 train_time: 0.0m tok/s: 8109891 +3/20000 train_loss: 11.0440 train_time: 0.0m tok/s: 8043550 +4/20000 train_loss: 9.4087 train_time: 0.0m tok/s: 8012978 +5/20000 train_loss: 8.3248 train_time: 0.0m tok/s: 7998156 +500/20000 train_loss: 3.3404 train_time: 0.8m tok/s: 7773362 +1000/20000 train_loss: 3.2133 train_time: 1.7m tok/s: 7764854 +1500/20000 train_loss: 3.1034 train_time: 2.5m tok/s: 7769144 +2000/20000 train_loss: 3.0206 train_time: 3.4m tok/s: 7769320 +layer_loop:enabled step:2034 frac:0.350 encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +2500/20000 train_loss: 3.0034 train_time: 4.6m tok/s: 7149424 +3000/20000 train_loss: 3.0418 train_time: 5.8m tok/s: 6758023 +3500/20000 train_loss: 2.9228 train_time: 7.1m tok/s: 6504182 +4000/20000 train_loss: 2.9591 train_time: 8.3m tok/s: 6326334 +4000/20000 val_loss: 2.8764 val_bpb: 1.1138 +4500/20000 train_loss: 2.7576 train_time: 9.5m tok/s: 6195392 +4614/20000 val_loss: 2.8081 val_bpb: 1.0873 +stopping_early: wallclock_cap train_time: 588121ms step: 4614/20000 +peak memory allocated: 39046 MiB reserved: 39070 MiB +ema:applying EMA weights +pre-quantization post-ema val_loss:2.80491496 val_bpb:1.08609269 eval_time:6363ms +Serialized model: 135431033 bytes +Code size: 72758 bytes +GPTQ:collecting Hessians from calibration data... +GPTQ:collected 67 Hessians in 12.7s +Quantized weights: + gptq (int6): blocks.attn.c_k.weight, blocks.attn.c_q.weight, blocks.attn.c_v.weight, blocks.attn.proj.weight, blocks.mlp.fc.weight, blocks.mlp.proj.weight + gptq (int8): tok_emb.weight + passthrough (float16): blocks.attn.q_gain, blocks.attn_scale, blocks.mlp_scale, blocks.resid_mix, skip_gates, skip_weights +Serialized model quantized+brotli: 15971929 bytes +Total submission size quantized+brotli: 16044687 bytes +quantized val_loss:2.83464461 val_bpb:1.09760432 eval_time:21738ms +quantized_sliding_window val_loss:2.79132641 val_bpb:1.08083105 eval_time:124542ms +ttt:enabled lr=0.005 momentum=0.9 epochs=3 chunk=32768 freeze_blocks=9 +quantized_sliding_window_ttt val_loss:2.78999706 val_bpb:1.08031632 eval_time:423060ms diff --git a/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_seed42.log b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_seed42.log new file mode 100644 index 0000000000..47642bea7f --- /dev/null +++ b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_seed42.log @@ -0,0 +1,229 @@ +W0422 13:08:55.260000 58377 torch/distributed/run.py:803] +W0422 13:08:55.260000 58377 torch/distributed/run.py:803] ***************************************** +W0422 13:08:55.260000 58377 torch/distributed/run.py:803] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0422 13:08:55.260000 58377 torch/distributed/run.py:803] ***************************************** +Hyperparameters: + adam_eps: 1e-08 + adam_wd: 0.02 + beta1: 0.9 + beta2: 0.95 + compressor: brotli + data_dir: ./data_lsu_shim/ + datasets_dir: ./data_lsu_shim/datasets/fineweb10B_sp8192 + distributed: True + ema_decay: 0.9965 + embed_bits: 8 + embed_clip_sigmas: 20.0 + embed_lr: 0.6 + embed_wd: 0.095 + embedding_dim: 512 + enable_looping_at: 0.35 + eval_seq_len: 2048 + eval_stride: 64 + gptq_calibration_batches: 64 + gptq_reserve_seconds: 12.0 + grad_accum_steps: 1 + grad_clip_norm: 0.3 + head_lr: 0.008 + is_main_process: True + iterations: 20000 + ln_scale: True + local_rank: 0 + logfile: logs/submit_ref_s42.txt + logit_softcap: 30.0 + loop_end: 5 + loop_start: 3 + matrix_bits: 6 + matrix_clip_sigmas: 12.85 + matrix_lr: 0.022 + max_wallclock_seconds: 600.0 + min_lr: 0.0 + mlp_mult: 4.0 + model_dim: 512 + model_path: final_model.pt + mtp_heads: 0 + mtp_weight: 0.3 + muon_backend_steps: 5 + muon_beta2: 0.95 + muon_momentum: 0.99 + muon_momentum_warmup_start: 0.92 + muon_momentum_warmup_steps: 1500 + muon_row_normalize: True + muon_wd: 0.095 + num_heads: 8 + num_kv_heads: 4 + num_layers: 11 + num_loops: 2 + parallel_residual_start: 7 + qk_gain_init: 5.25 + quantized_model_path: final_model.int6.ptz + rank: 0 + rope_base: 10000.0 + rope_dims: 16 + rope_train_seq_len: 2048 + run_id: submit_ref_s42 + scalar_lr: 0.02 + seed: 42 + skip_gates_enabled: True + sliding_window_enabled: True + tie_embeddings: True + tied_embed_init_std: 0.005 + tied_embed_lr: 0.03 + tokenizer_path: ./data_lsu_shim/tokenizers/fineweb_8192_bpe.model + train_batch_tokens: 786432 + train_files: ./data_lsu_shim/datasets/fineweb10B_sp8192/fineweb_train_*.bin + train_log_every: 500 + train_seq_len: 2048 + ttt_batch_seqs: 4 + ttt_chunk: 32768 + ttt_enabled: True + ttt_epochs: 3 + ttt_freeze_blocks: 9 + ttt_grad_clip: 1.0 + ttt_lr: 0.005 + ttt_momentum: 0.9 + val_batch_tokens: 524288 + val_files: ./data_lsu_shim/datasets/fineweb10B_sp8192/fineweb_val_*.bin + val_loss_every: 4000 + vocab_size: 8192 + warmdown_frac: 0.72 + warmup_steps: 20 + world_size: 8 + xsa_last_n: 11 +train_shards: 80 +val_tokens: 40548352 +model_params:35944536 +gptq:reserving 12s, effective=588000ms +warmup_step: 1/20 +warmup_step: 2/20 +warmup_step: 3/20 +warmup_step: 4/20 +warmup_step: 5/20 +warmup_step: 6/20 +warmup_step: 10/20 +warmup_step: 20/20 +loop_warmup:enabled encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +loop_warmup_step: 1/20 +loop_warmup_step: 2/20 +loop_warmup_step: 3/20 +loop_warmup_step: 4/20 +loop_warmup_step: 5/20 +loop_warmup_step: 6/20 +loop_warmup_step: 10/20 +loop_warmup_step: 20/20 +0/20000 val_loss: 9.0091 val_bpb: 3.4884 +1/20000 train_loss: 9.0116 train_time: 0.0m tok/s: 8329684 +2/20000 train_loss: 12.3389 train_time: 0.0m tok/s: 8158276 +3/20000 train_loss: 11.1024 train_time: 0.0m tok/s: 8071410 +4/20000 train_loss: 9.3879 train_time: 0.0m tok/s: 8033798 +5/20000 train_loss: 8.2843 train_time: 0.0m tok/s: 8000878 +500/20000 train_loss: 3.3367 train_time: 0.8m tok/s: 7747278 +1000/20000 train_loss: 3.2140 train_time: 1.7m tok/s: 7756816 +1500/20000 train_loss: 3.1054 train_time: 2.5m tok/s: 7763477 +2000/20000 train_loss: 3.0217 train_time: 3.4m tok/s: 7767333 +layer_loop:enabled step:2033 frac:0.350 encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +2500/20000 train_loss: 3.0066 train_time: 4.6m tok/s: 7144538 +3000/20000 train_loss: 3.0442 train_time: 5.8m tok/s: 6752969 +3500/20000 train_loss: 2.9300 train_time: 7.1m tok/s: 6486565 +4000/20000 train_loss: 2.9568 train_time: 8.3m tok/s: 6311046 +4000/20000 val_loss: 2.8787 val_bpb: 1.1146 +4500/20000 train_loss: 2.7594 train_time: 9.5m tok/s: 6181387 +4605/20000 val_loss: 2.8113 val_bpb: 1.0886 +stopping_early: wallclock_cap train_time: 588087ms step: 4605/20000 +peak memory allocated: 39046 MiB reserved: 39070 MiB +ema:applying EMA weights +pre-quantization post-ema val_loss:2.80810780 val_bpb:1.08732899 eval_time:6903ms +Serialized model: 135431033 bytes +Code size: 72758 bytes +GPTQ:collecting Hessians from calibration data... +GPTQ:collected 67 Hessians in 12.8s +Quantized weights: + gptq (int6): blocks.attn.c_k.weight, blocks.attn.c_q.weight, blocks.attn.c_v.weight, blocks.attn.proj.weight, blocks.mlp.fc.weight, blocks.mlp.proj.weight + gptq (int8): tok_emb.weight + passthrough (float16): blocks.attn.q_gain, blocks.attn_scale, blocks.mlp_scale, blocks.resid_mix, skip_gates, skip_weights +Serialized model quantized+brotli: 15973790 bytes +Total submission size quantized+brotli: 16046548 bytes +quantized val_loss:2.83775400 val_bpb:1.09880831 eval_time:4741ms +quantized_sliding_window val_loss:2.79470070 val_bpb:1.08213762 eval_time:91780ms +ttt:enabled lr=0.005 momentum=0.9 epochs=3 chunk=32768 freeze_blocks=9 +quantized_sliding_window_ttt val_loss:2.79311423 val_bpb:1.08152332 eval_time:429656ms +----------------------------------+------------------------+----------------------+ +| 5 NVIDIA H100 80GB HBM3 On | 00000000:BB:00.0 Off | 0 | +| N/A 41C P0 122W / 700W | 1521MiB / 81559MiB | 0% Default | +| | | Disabled | ++-----------------------------------------+------------------------+----------------------+ +| 6 NVIDIA H100 80GB HBM3 On | 00000000:CB:00.0 Off | 0 | +| N/A 51C P0 123W / 700W | 1521MiB / 81559MiB | 0% Default | +| | | Disabled | ++-----------------------------------------+------------------------+----------------------+ +| 7 NVIDIA H100 80GB HBM3 On | 00000000:DB:00.0 Off | 0 | +| N/A 40C P0 120W / 700W | 1521MiB / 81559MiB | 3% Default | +| | | Disabled | ++-----------------------------------------+------------------------+----------------------+ + ++-----------------------------------------------------------------------------------------+ +| Processes: | +| GPU GI CI PID Type Process name GPU Memory | +| ID ID Usage | +|=========================================================================================| +| No running processes found | ++-----------------------------------------------------------------------------------------+ + +==================================================================================================== +train_shards: 80 +val_tokens: 40548352 +model_params:35944536 +gptq:reserving 12s, effective=588000ms +warmup_step: 1/20 +warmup_step: 2/20 +warmup_step: 3/20 +warmup_step: 4/20 +warmup_step: 5/20 +warmup_step: 6/20 +warmup_step: 10/20 +warmup_step: 20/20 +loop_warmup:enabled encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +loop_warmup_step: 1/20 +loop_warmup_step: 2/20 +loop_warmup_step: 3/20 +loop_warmup_step: 4/20 +loop_warmup_step: 5/20 +loop_warmup_step: 6/20 +loop_warmup_step: 10/20 +loop_warmup_step: 20/20 +0/20000 val_loss: 9.0091 val_bpb: 3.4884 +1/20000 train_loss: 9.0116 train_time: 0.0m tok/s: 8329684 +2/20000 train_loss: 12.3389 train_time: 0.0m tok/s: 8158276 +3/20000 train_loss: 11.1024 train_time: 0.0m tok/s: 8071410 +4/20000 train_loss: 9.3879 train_time: 0.0m tok/s: 8033798 +5/20000 train_loss: 8.2843 train_time: 0.0m tok/s: 8000878 +500/20000 train_loss: 3.3367 train_time: 0.8m tok/s: 7747278 +1000/20000 train_loss: 3.2140 train_time: 1.7m tok/s: 7756816 +1500/20000 train_loss: 3.1054 train_time: 2.5m tok/s: 7763477 +2000/20000 train_loss: 3.0217 train_time: 3.4m tok/s: 7767333 +layer_loop:enabled step:2033 frac:0.350 encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +2500/20000 train_loss: 3.0066 train_time: 4.6m tok/s: 7144538 +3000/20000 train_loss: 3.0442 train_time: 5.8m tok/s: 6752969 +3500/20000 train_loss: 2.9300 train_time: 7.1m tok/s: 6486565 +4000/20000 train_loss: 2.9568 train_time: 8.3m tok/s: 6311046 +4000/20000 val_loss: 2.8787 val_bpb: 1.1146 +4500/20000 train_loss: 2.7594 train_time: 9.5m tok/s: 6181387 +4605/20000 val_loss: 2.8113 val_bpb: 1.0886 +stopping_early: wallclock_cap train_time: 588087ms step: 4605/20000 +peak memory allocated: 39046 MiB reserved: 39070 MiB +ema:applying EMA weights +pre-quantization post-ema val_loss:2.80810780 val_bpb:1.08732899 eval_time:6903ms +Serialized model: 135431033 bytes +Code size: 72758 bytes +GPTQ:collecting Hessians from calibration data... +GPTQ:collected 67 Hessians in 12.8s +Quantized weights: + gptq (int6): blocks.attn.c_k.weight, blocks.attn.c_q.weight, blocks.attn.c_v.weight, blocks.attn.proj.weight, blocks.mlp.fc.weight, blocks.mlp.proj.weight + gptq (int8): tok_emb.weight + passthrough (float16): blocks.attn.q_gain, blocks.attn_scale, blocks.mlp_scale, blocks.resid_mix, skip_gates, skip_weights +Serialized model quantized+brotli: 15973790 bytes +Total submission size quantized+brotli: 16046548 bytes +quantized val_loss:2.83775400 val_bpb:1.09880831 eval_time:4741ms +quantized_sliding_window val_loss:2.79470070 val_bpb:1.08213762 eval_time:91780ms +ttt:enabled lr=0.005 momentum=0.9 epochs=3 chunk=32768 freeze_blocks=9 +quantized_sliding_window_ttt val_loss:2.79311423 val_bpb:1.08152332 eval_time:429656ms diff --git a/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_seed7.log b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_seed7.log new file mode 100644 index 0000000000..8accae5971 --- /dev/null +++ b/records/track_10min_16mb/2026-04-22_SP8192_ParResid_3LayerLoop_QK525_LegalTTT/train_seed7.log @@ -0,0 +1,229 @@ +W0422 13:30:25.269000 64062 torch/distributed/run.py:803] +W0422 13:30:25.269000 64062 torch/distributed/run.py:803] ***************************************** +W0422 13:30:25.269000 64062 torch/distributed/run.py:803] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0422 13:30:25.269000 64062 torch/distributed/run.py:803] ***************************************** +Hyperparameters: + adam_eps: 1e-08 + adam_wd: 0.02 + beta1: 0.9 + beta2: 0.95 + compressor: brotli + data_dir: ./data_lsu_shim/ + datasets_dir: ./data_lsu_shim/datasets/fineweb10B_sp8192 + distributed: True + ema_decay: 0.9965 + embed_bits: 8 + embed_clip_sigmas: 20.0 + embed_lr: 0.6 + embed_wd: 0.095 + embedding_dim: 512 + enable_looping_at: 0.35 + eval_seq_len: 2048 + eval_stride: 64 + gptq_calibration_batches: 64 + gptq_reserve_seconds: 12.0 + grad_accum_steps: 1 + grad_clip_norm: 0.3 + head_lr: 0.008 + is_main_process: True + iterations: 20000 + ln_scale: True + local_rank: 0 + logfile: logs/submit_ref_s7.txt + logit_softcap: 30.0 + loop_end: 5 + loop_start: 3 + matrix_bits: 6 + matrix_clip_sigmas: 12.85 + matrix_lr: 0.022 + max_wallclock_seconds: 600.0 + min_lr: 0.0 + mlp_mult: 4.0 + model_dim: 512 + model_path: final_model.pt + mtp_heads: 0 + mtp_weight: 0.3 + muon_backend_steps: 5 + muon_beta2: 0.95 + muon_momentum: 0.99 + muon_momentum_warmup_start: 0.92 + muon_momentum_warmup_steps: 1500 + muon_row_normalize: True + muon_wd: 0.095 + num_heads: 8 + num_kv_heads: 4 + num_layers: 11 + num_loops: 2 + parallel_residual_start: 7 + qk_gain_init: 5.25 + quantized_model_path: final_model.int6.ptz + rank: 0 + rope_base: 10000.0 + rope_dims: 16 + rope_train_seq_len: 2048 + run_id: submit_ref_s7 + scalar_lr: 0.02 + seed: 7 + skip_gates_enabled: True + sliding_window_enabled: True + tie_embeddings: True + tied_embed_init_std: 0.005 + tied_embed_lr: 0.03 + tokenizer_path: ./data_lsu_shim/tokenizers/fineweb_8192_bpe.model + train_batch_tokens: 786432 + train_files: ./data_lsu_shim/datasets/fineweb10B_sp8192/fineweb_train_*.bin + train_log_every: 500 + train_seq_len: 2048 + ttt_batch_seqs: 4 + ttt_chunk: 32768 + ttt_enabled: True + ttt_epochs: 3 + ttt_freeze_blocks: 9 + ttt_grad_clip: 1.0 + ttt_lr: 0.005 + ttt_momentum: 0.9 + val_batch_tokens: 524288 + val_files: ./data_lsu_shim/datasets/fineweb10B_sp8192/fineweb_val_*.bin + val_loss_every: 4000 + vocab_size: 8192 + warmdown_frac: 0.72 + warmup_steps: 20 + world_size: 8 + xsa_last_n: 11 +train_shards: 80 +val_tokens: 40548352 +model_params:35944536 +gptq:reserving 12s, effective=588000ms +warmup_step: 1/20 +warmup_step: 2/20 +warmup_step: 3/20 +warmup_step: 4/20 +warmup_step: 5/20 +warmup_step: 6/20 +warmup_step: 10/20 +warmup_step: 20/20 +loop_warmup:enabled encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +loop_warmup_step: 1/20 +loop_warmup_step: 2/20 +loop_warmup_step: 3/20 +loop_warmup_step: 4/20 +loop_warmup_step: 5/20 +loop_warmup_step: 6/20 +loop_warmup_step: 10/20 +loop_warmup_step: 20/20 +0/20000 val_loss: 9.0086 val_bpb: 3.4882 +1/20000 train_loss: 9.0106 train_time: 0.0m tok/s: 8311938 +2/20000 train_loss: 12.3305 train_time: 0.0m tok/s: 8182813 +3/20000 train_loss: 11.0986 train_time: 0.0m tok/s: 8056030 +4/20000 train_loss: 9.4791 train_time: 0.0m tok/s: 8001984 +5/20000 train_loss: 8.3710 train_time: 0.0m tok/s: 7973045 +500/20000 train_loss: 3.3306 train_time: 0.8m tok/s: 7762760 +1000/20000 train_loss: 3.2126 train_time: 1.7m tok/s: 7760347 +1500/20000 train_loss: 3.0987 train_time: 2.5m tok/s: 7763231 +2000/20000 train_loss: 3.0163 train_time: 3.4m tok/s: 7766942 +layer_loop:enabled step:2033 frac:0.350 encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +2500/20000 train_loss: 3.0026 train_time: 4.6m tok/s: 7143995 +3000/20000 train_loss: 3.0386 train_time: 5.8m tok/s: 6752728 +3500/20000 train_loss: 2.9241 train_time: 7.1m tok/s: 6487269 +4000/20000 train_loss: 2.9575 train_time: 8.3m tok/s: 6312086 +4000/20000 val_loss: 2.8762 val_bpb: 1.1137 +4500/20000 train_loss: 2.7573 train_time: 9.5m tok/s: 6181769 +4605/20000 val_loss: 2.8088 val_bpb: 1.0876 +stopping_early: wallclock_cap train_time: 588045ms step: 4605/20000 +peak memory allocated: 39046 MiB reserved: 39070 MiB +ema:applying EMA weights +pre-quantization post-ema val_loss:2.80558297 val_bpb:1.08635134 eval_time:6924ms +Serialized model: 135431033 bytes +Code size: 72758 bytes +GPTQ:collecting Hessians from calibration data... +GPTQ:collected 67 Hessians in 12.8s +Quantized weights: + gptq (int6): blocks.attn.c_k.weight, blocks.attn.c_q.weight, blocks.attn.c_v.weight, blocks.attn.proj.weight, blocks.mlp.fc.weight, blocks.mlp.proj.weight + gptq (int8): tok_emb.weight + passthrough (float16): blocks.attn.q_gain, blocks.attn_scale, blocks.mlp_scale, blocks.resid_mix, skip_gates, skip_weights +Serialized model quantized+brotli: 15971863 bytes +Total submission size quantized+brotli: 16044621 bytes +quantized val_loss:2.83557581 val_bpb:1.09796489 eval_time:4761ms +quantized_sliding_window val_loss:2.79265390 val_bpb:1.08134507 eval_time:91774ms +ttt:enabled lr=0.005 momentum=0.9 epochs=3 chunk=32768 freeze_blocks=9 +quantized_sliding_window_ttt val_loss:2.79082941 val_bpb:1.08063861 eval_time:417914ms +----------------------------------+------------------------+----------------------+ +| 5 NVIDIA H100 80GB HBM3 On | 00000000:BB:00.0 Off | 0 | +| N/A 41C P0 123W / 700W | 1521MiB / 81559MiB | 0% Default | +| | | Disabled | ++-----------------------------------------+------------------------+----------------------+ +| 6 NVIDIA H100 80GB HBM3 On | 00000000:CB:00.0 Off | 0 | +| N/A 51C P0 126W / 700W | 1521MiB / 81559MiB | 0% Default | +| | | Disabled | ++-----------------------------------------+------------------------+----------------------+ +| 7 NVIDIA H100 80GB HBM3 On | 00000000:DB:00.0 Off | 0 | +| N/A 40C P0 120W / 700W | 1521MiB / 81559MiB | 0% Default | +| | | Disabled | ++-----------------------------------------+------------------------+----------------------+ + ++-----------------------------------------------------------------------------------------+ +| Processes: | +| GPU GI CI PID Type Process name GPU Memory | +| ID ID Usage | +|=========================================================================================| +| No running processes found | ++-----------------------------------------------------------------------------------------+ + +==================================================================================================== +train_shards: 80 +val_tokens: 40548352 +model_params:35944536 +gptq:reserving 12s, effective=588000ms +warmup_step: 1/20 +warmup_step: 2/20 +warmup_step: 3/20 +warmup_step: 4/20 +warmup_step: 5/20 +warmup_step: 6/20 +warmup_step: 10/20 +warmup_step: 20/20 +loop_warmup:enabled encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +loop_warmup_step: 1/20 +loop_warmup_step: 2/20 +loop_warmup_step: 3/20 +loop_warmup_step: 4/20 +loop_warmup_step: 5/20 +loop_warmup_step: 6/20 +loop_warmup_step: 10/20 +loop_warmup_step: 20/20 +0/20000 val_loss: 9.0086 val_bpb: 3.4882 +1/20000 train_loss: 9.0106 train_time: 0.0m tok/s: 8311938 +2/20000 train_loss: 12.3305 train_time: 0.0m tok/s: 8182813 +3/20000 train_loss: 11.0986 train_time: 0.0m tok/s: 8056030 +4/20000 train_loss: 9.4791 train_time: 0.0m tok/s: 8001984 +5/20000 train_loss: 8.3710 train_time: 0.0m tok/s: 7973045 +500/20000 train_loss: 3.3306 train_time: 0.8m tok/s: 7762760 +1000/20000 train_loss: 3.2126 train_time: 1.7m tok/s: 7760347 +1500/20000 train_loss: 3.0987 train_time: 2.5m tok/s: 7763231 +2000/20000 train_loss: 3.0163 train_time: 3.4m tok/s: 7766942 +layer_loop:enabled step:2033 frac:0.350 encoder:[0, 1, 2, 3, 4, 5, 3, 4] decoder:[5, 3, 4, 5, 6, 7, 8, 9, 10] +2500/20000 train_loss: 3.0026 train_time: 4.6m tok/s: 7143995 +3000/20000 train_loss: 3.0386 train_time: 5.8m tok/s: 6752728 +3500/20000 train_loss: 2.9241 train_time: 7.1m tok/s: 6487269 +4000/20000 train_loss: 2.9575 train_time: 8.3m tok/s: 6312086 +4000/20000 val_loss: 2.8762 val_bpb: 1.1137 +4500/20000 train_loss: 2.7573 train_time: 9.5m tok/s: 6181769 +4605/20000 val_loss: 2.8088 val_bpb: 1.0876 +stopping_early: wallclock_cap train_time: 588045ms step: 4605/20000 +peak memory allocated: 39046 MiB reserved: 39070 MiB +ema:applying EMA weights +pre-quantization post-ema val_loss:2.80558297 val_bpb:1.08635134 eval_time:6924ms +Serialized model: 135431033 bytes +Code size: 72758 bytes +GPTQ:collecting Hessians from calibration data... +GPTQ:collected 67 Hessians in 12.8s +Quantized weights: + gptq (int6): blocks.attn.c_k.weight, blocks.attn.c_q.weight, blocks.attn.c_v.weight, blocks.attn.proj.weight, blocks.mlp.fc.weight, blocks.mlp.proj.weight + gptq (int8): tok_emb.weight + passthrough (float16): blocks.attn.q_gain, blocks.attn_scale, blocks.mlp_scale, blocks.resid_mix, skip_gates, skip_weights +Serialized model quantized+brotli: 15971863 bytes +Total submission size quantized+brotli: 16044621 bytes +quantized val_loss:2.83557581 val_bpb:1.09796489 eval_time:4761ms +quantized_sliding_window val_loss:2.79265390 val_bpb:1.08134507 eval_time:91774ms +ttt:enabled lr=0.005 momentum=0.9 epochs=3 chunk=32768 freeze_blocks=9 +quantized_sliding_window_ttt val_loss:2.79082941 val_bpb:1.08063861 eval_time:417914ms