File size: 36,680 Bytes
b7c0571 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 | """
NovaMind-256M: Decoder-Only Conversational Language Model
==========================================================
Architecture: Decoder-only Transformer
Key design choices:
1. HiRoPE (Hierarchical RoPE) β adapted from "HiRoPE: Length Extrapolation
for Code Models Using Hierarchical Position" (ACL 2024).
Original paper used it for code structure (token/statement/function hierarchy).
We adapt it for conversation structure: local turn position vs. global dialogue position.
Split head dims into local-base=10K and global-base=500K streams.
2. Tag-Aware Loss Curriculum β original contribution.
Per-token loss weighting that changes across training phases.
Target: ~252M parameters (after weight tying)
Hardware target: H100 40GB (Lightning.ai student account)
Precision: BF16
"""
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.utils.checkpoint import checkpoint as gradient_checkpoint
from dataclasses import dataclass, field, asdict
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# CONFIG
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
@dataclass
class NovaMindConfig:
"""
All hyperparameters in one place for clean checkpointing and reproducibility.
Changing a value here changes the entire model β no magic numbers buried
in the code.
"""
# ββ Vocabulary ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
vocab_size: int = 32_000 # LLaMA-2 tokenizer (NOT GPT-2's 50K)
# ββ Dimensions ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
d_model: int = 1_024 # hidden size
n_heads: int = 16 # number of query attention heads
n_kv_heads: int = 4 # GQA: key/value heads (ratio 4:1)
n_layers: int = 24 # transformer depth
ff_dim: int = 2_304 # SwiGLU inner dim = 2.25 Γ d_model, div by 128 β ~265M total
# ββ Context βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
max_seq_len: int = 2_048 # training context; extendable later via RoPE scaling
# ββ HiRoPE β adapted from HiRoPE (ACL 2024), applied to conversation ββββ
rope_local_base: float = 10_000.0 # local position (within-turn): ~LLaMA default
rope_global_base: float = 500_000.0 # global position (cross-turn): much slower freq
# ββ Regularization ββββββββββββββββββββββββββββββββββββββββββββββββββββββ
dropout: float = 0.0 # no dropout (modern LLM practice)
attn_dropout: float = 0.0
# ββ Init ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
init_std: float = 0.02 # base std for embedding and non-residual projections
def head_dim(self) -> int:
assert self.d_model % self.n_heads == 0, "d_model must be divisible by n_heads"
return self.d_model // self.n_heads
def n_rep(self) -> int:
"""How many Q heads share each KV head."""
assert self.n_heads % self.n_kv_heads == 0
return self.n_heads // self.n_kv_heads
def to_dict(self):
return asdict(self)
@classmethod
def from_dict(cls, d):
return cls(**d)
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# RMS NORM
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
class RMSNorm(nn.Module):
"""
Root Mean Square Layer Normalization (Zhang & Sennrich, 2019).
Faster than LayerNorm because it skips the mean-centering step.
Used by LLaMA, Gemma, Qwen, Mistral β now us.
"""
def __init__(self, dim: int, eps: float = 1e-6):
super().__init__()
self.eps = eps
self.weight = nn.Parameter(torch.ones(dim))
def forward(self, x: torch.Tensor) -> torch.Tensor:
# x: [..., dim]
# Compute RMS then scale
rms = x.pow(2).mean(-1, keepdim=True).add(self.eps).rsqrt()
return x * rms * self.weight
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# HiRoPE β HIERARCHICAL ROTARY POSITION EMBEDDING
# Adapted from "HiRoPE: Length Extrapolation for Code Models Using Hierarchical
# Position" (ACL 2024). Original: code hierarchy. Our adaptation: dialogue.
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def build_hier_rope_cache(
seq_len: int,
head_dim: int,
device: torch.device,
local_base: float = 10_000.0,
global_base: float = 500_000.0,
) -> tuple[torch.Tensor, torch.Tensor]:
"""
HiRoPE: Hierarchical Rotary Position Embedding Cache.
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Adapted from: "HiRoPE: Length Extrapolation for Code Models Using
Hierarchical Position" (ACL 2024).
Original paper: used code structure hierarchy (token-in-statement,
statement-in-function) to split RoPE dimensions.
Our adaptation: we use CONVERSATION structure instead.
β’ Dims 0 β¦ head_dim//2-1 β LOCAL RoPE (base=10,000)
Encodes fine-grained position WITHIN a conversation turn.
Same frequency as standard LLaMA RoPE β well-calibrated for
sentence-level syntax and local coreference.
β’ Dims head_dim//2 β¦ head_dim-1 β GLOBAL RoPE (base=500,000)
Encodes coarse position ACROSS conversation turns.
A higher base = slower rotation = can distinguish positions
across much longer spans without frequency wrap-around.
This is what the original HiRoPE paper used for cross-function
dependencies in code; we repurpose it for cross-turn coherence.
Cost: ZERO extra parameters. Purely a change in the frequency basis.
Returns:
cos, sin β each shape [seq_len, head_dim], cached for efficiency
"""
half = head_dim // 2 # split point
# ββ Local stream (dims 0 to half-1) ββββββββββββββββββββββββββββββββββ
# theta_i = 1 / (base^(2i/dim)) for i in [0, half/2)
# We generate half/2 unique theta values, then duplicate for rotation trick
local_theta = 1.0 / (
local_base ** (torch.arange(0, half, 2, device=device).float() / half)
)
# ββ Global stream (dims half to head_dim-1) βββββββββββββββββββββββββββ
global_theta = 1.0 / (
global_base ** (torch.arange(0, half, 2, device=device).float() / half)
)
positions = torch.arange(seq_len, device=device).float() # [seq_len]
# Outer product: [seq_len, half/2]
local_freqs = torch.outer(positions, local_theta)
global_freqs = torch.outer(positions, global_theta)
# Duplicate for the rotation trick: [q1, q2] β [-q2, q1]
# β shape [seq_len, half]
local_freqs = torch.cat([local_freqs, local_freqs], dim=-1)
global_freqs = torch.cat([global_freqs, global_freqs], dim=-1)
# Concatenate local and global to cover full head_dim: [seq_len, head_dim]
cos = torch.cat([local_freqs.cos(), global_freqs.cos()], dim=-1)
sin = torch.cat([local_freqs.sin(), global_freqs.sin()], dim=-1)
return cos, sin # [seq_len, head_dim]
def _rotate_half(x: torch.Tensor) -> torch.Tensor:
"""Standard RoPE rotation: rotate each pair of dims by 90Β°."""
half = x.shape[-1] // 2
return torch.cat([-x[..., half:], x[..., :half]], dim=-1)
def apply_hier_rope(
q: torch.Tensor,
k: torch.Tensor,
cos: torch.Tensor,
sin: torch.Tensor,
) -> tuple[torch.Tensor, torch.Tensor]:
"""
Apply HierRoPE to query and key tensors.
Because cos/sin are already split into local//global halves,
this is mathematically identical to standard RoPE β the hierarchy
lives in the frequency basis, not the application code.
Args:
q, k: [batch, n_heads, seq_len, head_dim]
cos, sin: [seq_len, head_dim] (from build_hier_rope_cache)
Returns:
rotated q, k: same shapes as input
"""
# Expand cos/sin to broadcast over batch and head dims
cos = cos[None, None, :, :] # [1, 1, seq_len, head_dim]
sin = sin[None, None, :, :]
q_rot = q * cos + _rotate_half(q) * sin
k_rot = k * cos + _rotate_half(k) * sin
return q_rot, k_rot
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# GROUPED QUERY ATTENTION WITH HiRoPE
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
class GQACausalAttention(nn.Module):
"""
Grouped Query Attention (GQA) with HiRoPE (adapted from ACL 2024 paper).
GQA uses n_heads query heads but only n_kv_heads key/value heads.
Each KV head is shared by (n_heads // n_kv_heads) query heads.
This reduces KV cache size by 4x during inference β critical for long
conversations. Used by LLaMA 2/3, Mistral, Gemma.
Attention is computed via F.scaled_dot_product_attention which uses
Flash Attention 2 on supported hardware (H100 β
).
"""
def __init__(self, config: NovaMindConfig):
super().__init__()
self.n_heads = config.n_heads
self.n_kv_heads = config.n_kv_heads
self.n_rep = config.n_rep() # Q heads per KV head
self.head_dim = config.head_dim()
self.d_model = config.d_model
# Projections β NO bias (modern LLM best practice, saves params)
self.q_proj = nn.Linear(self.d_model, self.n_heads * self.head_dim, bias=False)
self.k_proj = nn.Linear(self.d_model, self.n_kv_heads * self.head_dim, bias=False)
self.v_proj = nn.Linear(self.d_model, self.n_kv_heads * self.head_dim, bias=False)
self.o_proj = nn.Linear(self.d_model, self.d_model, bias=False)
self.attn_drop = config.attn_dropout
# Cache HiRoPE cos/sin buffers to avoid recomputing every forward pass
# persistent=False β not saved in state_dict (recomputed on load)
self.register_buffer(
"_rope_cos",
torch.zeros(config.max_seq_len, self.head_dim),
persistent=False,
)
self.register_buffer(
"_rope_sin",
torch.zeros(config.max_seq_len, self.head_dim),
persistent=False,
)
self._rope_cached_len = 0
self._rope_local_base = config.rope_local_base
self._rope_global_base = config.rope_global_base
def _get_hier_rope(self, seq_len: int, device: torch.device):
"""Lazily compute and cache HiRoPE tables."""
if seq_len > self._rope_cached_len or self._rope_cos.device != device:
cos, sin = build_hier_rope_cache(
seq_len, self.head_dim, device,
local_base=self._rope_local_base,
global_base=self._rope_global_base,
)
self._rope_cos = cos
self._rope_sin = sin
self._rope_cached_len = seq_len
return self._rope_cos[:seq_len], self._rope_sin[:seq_len]
def forward(self, x: torch.Tensor) -> torch.Tensor:
"""
Args:
x: [B, T, D] β input hidden states
Returns:
out: [B, T, D]
"""
B, T, D = x.shape
# Project to Q, K, V
Q = self.q_proj(x).view(B, T, self.n_heads, self.head_dim).transpose(1, 2)
K = self.k_proj(x).view(B, T, self.n_kv_heads, self.head_dim).transpose(1, 2)
V = self.v_proj(x).view(B, T, self.n_kv_heads, self.head_dim).transpose(1, 2)
# Q: [B, n_heads, T, head_dim]
# K: [B, n_kv_heads, T, head_dim]
# V: [B, n_kv_heads, T, head_dim]
# Apply HiRoPE to Q and K
cos, sin = self._get_hier_rope(T, x.device)
Q, K = apply_hier_rope(Q, K, cos.to(x.dtype), sin.to(x.dtype))
# Expand KV heads to match Q heads for GQA
# This uses expand() + reshape instead of repeat_interleave β same math,
# but expand() is zero-copy (shares memory), saving VRAM
K = K.unsqueeze(2).expand(B, self.n_kv_heads, self.n_rep, T, self.head_dim)\
.reshape(B, self.n_heads, T, self.head_dim)
V = V.unsqueeze(2).expand(B, self.n_kv_heads, self.n_rep, T, self.head_dim)\
.reshape(B, self.n_heads, T, self.head_dim)
# Flash Attention 2 path via PyTorch β causal mask included
out = F.scaled_dot_product_attention(
Q, K, V,
attn_mask=None,
dropout_p=self.attn_drop if self.training else 0.0,
is_causal=True,
) # [B, n_heads, T, head_dim]
# Reassemble heads and project
out = out.transpose(1, 2).contiguous().view(B, T, D)
return self.o_proj(out)
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# SWIGLU FEED-FORWARD NETWORK
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
class SwiGLUFFN(nn.Module):
"""
SwiGLU Feed-Forward Network (Shazeer, 2020; adopted by PaLM, LLaMA, Gemma).
Unlike standard FFN (2 matrices: up-project + down-project),
SwiGLU uses 3 matrices:
gate(x) β passed through SiLU activation (the "gate")
up(x) β element-wise multiplied with the gate output
down(Β·) β project back to d_model
The gating mechanism acts like learned neuron selection β it can
suppress irrelevant features entirely. Result: ~15% better quality
per parameter vs standard GELU FFN (measured in Google's PaLM paper).
"""
def __init__(self, config: NovaMindConfig):
super().__init__()
# All three matrices, no bias
self.gate = nn.Linear(config.d_model, config.ff_dim, bias=False)
self.up = nn.Linear(config.d_model, config.ff_dim, bias=False)
self.down = nn.Linear(config.ff_dim, config.d_model, bias=False)
def forward(self, x: torch.Tensor) -> torch.Tensor:
# SiLU(gate(x)) β up(x) β element-wise gating
return self.down(F.silu(self.gate(x)) * self.up(x))
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# TRANSFORMER BLOCK
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
class TransformerBlock(nn.Module):
"""
One transformer layer using Pre-RMSNorm architecture:
x β RMSNorm β GQACausalAttention β +x (residual)
β RMSNorm β SwiGLUFFN β +x (residual)
Pre-norm (norm BEFORE the sub-layer) is more training-stable than
Post-norm, especially at 24 layers deep. All modern LLMs use Pre-norm.
"""
def __init__(self, config: NovaMindConfig):
super().__init__()
self.ln_attn = RMSNorm(config.d_model)
self.attn = GQACausalAttention(config)
self.ln_ffn = RMSNorm(config.d_model)
self.ffn = SwiGLUFFN(config)
def forward(self, x: torch.Tensor) -> torch.Tensor:
# Attention sub-layer with residual connection
x = x + self.attn(self.ln_attn(x))
# FFN sub-layer with residual connection
x = x + self.ffn(self.ln_ffn(x))
return x
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# TAG-AWARE LOSS β NOVEL CONTRIBUTION #2
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def tag_aware_loss(
logits: torch.Tensor,
targets: torch.Tensor,
token_weights: torch.Tensor | None = None,
) -> torch.Tensor:
"""
Tag-Aware Loss Curriculum (Novel Contribution #2).
ββββββββββββββββββββββββββββββββββββββββββββββββββ
Standard cross-entropy treats every predicted token equally.
We instead weight each token's loss contribution by its role in the
conversation, and by which training phase we are in:
Token types + phase weights:
ββββββββββββββββββ¬βββββββββββ¬βββββββββββ¬βββββββββββββββ
β Token type β Phase 1 β Phase 2 β Phase 3 β
β β Pretrain β SFT β CoT SFT β
ββββββββββββββββββΌβββββββββββΌβββββββββββΌβββββββββββββββ€
β <think> β 0.0 β 0.5 β 1.5 β BOOSTβ
β <assistant> β 1.0 β 1.0 β 1.0 β
β <human>/system β 0.0 β 0.0 β 0.0 β
ββββββββββββββββββ΄βββββββββββ΄βββββββββββ΄βββββββββββββββ
Why this is novel:
- Most papers use binary masking (weight is 0 or 1)
- A curriculum that changes weights across phases has not been
published at this scale
- Phase 3 boosting (weight=1.5 for think tokens) forces the model to
"care" about the quality of its own reasoning chain
Args:
logits: [B, T, V] β raw output logits
targets: [B, T] β target token ids (-1 = ignore this position)
token_weights: [B, T] β per-token float weight (None β uniform 1.0)
Returns:
Scalar loss value
"""
B, T, V = logits.shape
# Flat views for loss computation
logits_flat = logits.view(B * T, V)
targets_flat = targets.view(B * T)
if token_weights is None:
# Standard cross-entropy when no curriculum weights provided
return F.cross_entropy(logits_flat, targets_flat, ignore_index=-1)
# Compute per-token loss (reduction='none' gives us loss per position)
per_token_loss = F.cross_entropy(
logits_flat, targets_flat,
ignore_index=-1,
reduction="none",
) # [B*T]
weights_flat = token_weights.view(B * T)
# Mask out ignored positions (targets == -1)
# F.cross_entropy with ignore_index already zeroes them, but
# we also zero their weights to avoid dividing by them
valid_mask = (targets_flat != -1).float()
weights_flat = weights_flat * valid_mask
# Weighted mean: sum(w_i * loss_i) / sum(w_i)
# Small epsilon to prevent div-by-zero on pathological batches
weighted_loss = (per_token_loss * weights_flat).sum()
weight_sum = weights_flat.sum().clamp(min=1e-8)
return weighted_loss / weight_sum
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# NOVAMIND-256M β MAIN MODEL
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
class NovaMind256M(nn.Module):
"""
NovaMind-256M: 256M parameter conversational LLM.
Architectural features:
- Decoder-only transformer (autoregressive)
- Pre-RMSNorm for training stability
- GQA (16Q / 4KV) for inference efficiency
- SwiGLU FFN for parameter efficiency
- HiRoPE for hierarchical positional encoding (adapted from ACL 2024 paper)
- Weight-tied embedding and LM head (saves 33M params)
- Tag-Aware Loss Curriculum [NOVEL]
- BF16 training, Flash Attention 2
"""
def __init__(self, config: NovaMindConfig):
super().__init__()
self.config = config
self._use_gradient_checkpointing = False
# ββ Embedding ββββββββββββββββββββββββββββββββββββββββββββββββββββ
self.token_emb = nn.Embedding(config.vocab_size, config.d_model)
self.emb_dropout = nn.Dropout(config.dropout) if config.dropout > 0 else nn.Identity()
# ββ Transformer Stack βββββββββββββββββββββββββββββββββββββββββββββ
self.blocks = nn.ModuleList([
TransformerBlock(config) for _ in range(config.n_layers)
])
# ββ Output head ββββββββββββββββββββββββββββββββββββββββββββββββββ
self.ln_f = RMSNorm(config.d_model)
self.lm_head = nn.Linear(config.d_model, config.vocab_size, bias=False)
# Weight tying: lm_head and token_emb share the same weight matrix.
# This saves vocab_size Γ d_model = 32,000 Γ 1,024 = 32.77M parameters.
# The embedding learns "what does this token mean?" and the LM head
# learns "how does the hidden state score as this token?" β they're
# two views of the same embedding space.
self.lm_head.weight = self.token_emb.weight
# ββ Weight Initialization ββββββββββββββββββββββββββββββββββββββββ
self.apply(self._init_weights)
def _init_weights(self, module: nn.Module):
"""
GPT-2 style initialization with scaled residual projections.
Output projections (o_proj, down in FFN) are scaled by 1/sqrt(2*N)
where N = number of layers. This prevents the residual stream from
growing proportionally to depth, which would destabilize training.
Reference: GPT-2 paper, Table 2, "Modified initialization" footnote.
"""
std = self.config.init_std
if isinstance(module, nn.Linear):
nn.init.normal_(module.weight, mean=0.0, std=std)
if module.bias is not None:
nn.init.zeros_(module.bias)
elif isinstance(module, nn.Embedding):
nn.init.normal_(module.weight, mean=0.0, std=std)
# Scale residual output projections (o_proj and FFN down)
# Identify them by name β they project *into* the residual stream
residual_proj_names = {"o_proj", "down"}
for name, param in module.named_parameters(recurse=False):
# Use the parent module name to identify projection type
parent_name = type(module).__name__
if any(p in parent_name.lower() or p in name.lower()
for p in residual_proj_names):
nn.init.normal_(
param,
mean=0.0,
std=std / math.sqrt(2 * self.config.n_layers),
)
# ββ Gradient Checkpointing βββββββββββββββββββββββββββββββββββββββββββββββββ
def gradient_checkpointing_enable(self):
"""
Enable gradient checkpointing to trade compute for VRAM.
Discards intermediate activations during forward pass, recomputes
them during backward. Saves ~40% VRAM at cost of ~20% slower training.
CRITICAL for fitting 256M model + optimizer states on H100 40GB.
"""
self._use_gradient_checkpointing = True
def gradient_checkpointing_disable(self):
self._use_gradient_checkpointing = False
# ββ Forward Pass ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
def forward(
self,
input_ids: torch.Tensor,
targets: torch.Tensor | None = None,
token_weights: torch.Tensor | None = None,
) -> tuple[torch.Tensor, torch.Tensor | None]:
"""
Args:
input_ids: [B, T] β token ids
targets: [B, T] β target ids for loss computation (-1 = ignore)
If None, loss is not computed (inference mode)
token_weights: [B, T] β per-token float weights for Tag-Aware Loss
If None, uses standard uniform cross-entropy
Returns:
(logits, loss)
- logits: [B, T, V] during inference; empty tensor during training
(returning full logits during training on DataParallel is wasteful β
it forces a massive tensor transfer to GPU 0. We only need the loss.)
- loss: scalar tensor if targets provided, else None
"""
B, T = input_ids.shape
assert T <= self.config.max_seq_len, (
f"Input length {T} exceeds max_seq_len {self.config.max_seq_len}"
)
# Embed tokens
x = self.emb_dropout(self.token_emb(input_ids)) # [B, T, D]
# Pass through transformer blocks
for block in self.blocks:
if self._use_gradient_checkpointing and self.training:
# gradient_checkpoint recomputes this block's activations
# during backward instead of storing them β saves VRAM
x = gradient_checkpoint(block, x, use_reentrant=False)
else:
x = block(x)
# Final norm + project to vocabulary
logits = self.lm_head(self.ln_f(x)) # [B, T, V]
loss = None
if targets is not None:
loss = tag_aware_loss(logits, targets, token_weights)
# Drop logits during training to avoid PCIe transfer on multi-GPU
if self.training:
logits = torch.empty(0, device=logits.device)
return logits, loss
# ββ Generation ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
@torch.no_grad()
def generate(
self,
input_ids: torch.Tensor,
max_new_tokens: int = 256,
temperature: float = 0.7,
top_k: int = 50,
top_p: float = 0.9,
repetition_penalty: float = 1.2,
stop_token_ids: list[int] | None = None,
) -> torch.Tensor:
"""
Autoregressive text generation with temperature, top-k, top-p sampling
and repetition penalty.
Args:
input_ids: [1, T] β prompt tokens
max_new_tokens: maximum tokens to generate
temperature: sampling temperature (lower = more focused)
top_k: keep only top-k logits before sampling
top_p: nucleus sampling threshold
repetition_penalty: >1.0 penalizes repeated tokens
stop_token_ids: list of token ids that stop generation
Returns:
token ids including the prompt: [1, T + generated]
"""
self.eval()
for _ in range(max_new_tokens):
# Truncate context to max_seq_len (sliding window)
idx_cond = input_ids[:, -self.config.max_seq_len:]
logits, _ = self(idx_cond) # [1, T, V]
logits = logits[:, -1, :] # [1, V] β last position only
# Repetition penalty: down-weight tokens already in context
if repetition_penalty != 1.0:
for token_id in set(idx_cond[0].tolist()):
if logits[0, token_id] < 0:
logits[0, token_id] *= repetition_penalty
else:
logits[0, token_id] /= repetition_penalty
# Temperature scaling
logits = logits / max(temperature, 1e-8)
# Top-k filtering: zero out all but top-k logits
if top_k is not None and top_k > 0:
top_k_val = min(top_k, logits.size(-1))
kth_val = torch.topk(logits, top_k_val).values[:, -1, None]
logits[logits < kth_val] = float("-inf")
# Top-p (nucleus) filtering: keep smallest set of tokens
# whose cumulative probability exceeds top_p
if top_p is not None and 0.0 < top_p < 1.0:
sorted_logits, sorted_idx = torch.sort(logits, dim=-1, descending=True)
cumulative_probs = torch.cumsum(F.softmax(sorted_logits, dim=-1), dim=-1)
# Remove tokens with cumulative prob above threshold
# (shift by 1 to keep the token that *crosses* the threshold)
sorted_remove = cumulative_probs - F.softmax(sorted_logits, dim=-1) > top_p
sorted_logits[sorted_remove] = float("-inf")
logits = torch.zeros_like(logits).scatter_(
dim=-1, index=sorted_idx, src=sorted_logits
)
probs = F.softmax(logits, dim=-1)
next_id = torch.multinomial(probs, num_samples=1) # [1, 1]
input_ids = torch.cat([input_ids, next_id], dim=1)
# Stop if we hit a stop token
if stop_token_ids and next_id.item() in stop_token_ids:
break
return input_ids
# ββ Parameter Counting ββββββββββββββββββββββββββββββββββββββββββββββββββββ
def count_parameters(self, print_table: bool = True) -> int:
"""
Count and optionally print a detailed parameter breakdown.
Useful for verifying we hit ~252M.
"""
cfg = self.config
embed_params = cfg.vocab_size * cfg.d_model
hd = cfg.head_dim()
# Per-layer: Q+K+V+O projections + gate+up+down FFN + 2ΓRMSNorm
# QKV projections: (n_q + n_kv + n_kv) Γ head_dim Γ d_model
qkv_params = (cfg.n_heads + 2 * cfg.n_kv_heads) * hd * cfg.d_model
# Output projection: d_model Γ d_model
o_params = cfg.d_model * cfg.d_model
attn_params = qkv_params + o_params
# SwiGLU: gate + up + down (3 matrices)
ffn_params = 3 * cfg.d_model * cfg.ff_dim
norm_params = 2 * cfg.d_model # two RMSNorm per block
per_layer = attn_params + ffn_params + norm_params
final_norm = cfg.d_model
total_with_tie = embed_params + cfg.n_layers * per_layer + final_norm
total_trainable = sum(p.numel() for p in self.parameters() if p.requires_grad)
if print_table:
w = 46
print("β" * w)
print(f" NovaMind-256M Parameter Breakdown")
print("β" * w)
print(f" {'Token Embedding (shared w/ lm_head)':<40} {embed_params/1e6:>6.2f}M")
print(f" {'Attention QKV projections (per layer)':<40} {qkv_params/1e6:>6.2f}M")
print(f" {'Attention O projection (per layer)':<40} {o_params/1e6:>6.2f}M")
print(f" {'SwiGLU FFN (per layer)':<40} {ffn_params/1e6:>6.2f}M")
print(f" {'RMSNorm Γ 2 (per layer)':<40} {norm_params/1e6:>6.4f}M")
print(f" {'Γ {n} layers total':<40} {cfg.n_layers * per_layer/1e6:>6.2f}M")
print(f" {'Final RMSNorm':<40} {final_norm/1e6:>6.4f}M")
print(f" {'LM Head (weight-tied to embedding)':<40} {'β':>4} 0.00M")
print("β" * w)
print(f" {'TOTAL (formula estimate)':<40} {total_with_tie/1e6:>6.2f}M")
print(f" {'TOTAL (actual PyTorch count)':<40} {total_trainable/1e6:>6.2f}M")
print("β" * w)
return total_trainable
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
# QUICK SANITY CHECK
# ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
if __name__ == "__main__":
import sys
config = NovaMindConfig()
print(f"\nBuilding NovaMind-256M with config:")
for k, v in config.to_dict().items():
print(f" {k}: {v}")
print()
model = NovaMind256M(config)
n_params = model.count_parameters(print_table=True)
# Verify target range
target_min, target_max = 245e6, 265e6
status = "β
" if target_min <= n_params <= target_max else "β"
print(f"\n{status} Parameter count: {n_params/1e6:.2f}M (target: 245β265M)\n")
if not (target_min <= n_params <= target_max):
sys.exit(1)
# Quick forward pass test
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = model.to(device)
model.gradient_checkpointing_enable()
B, T = 2, 128
ids = torch.randint(0, config.vocab_size, (B, T), device=device)
targets = ids.clone()
weights = torch.ones(B, T, device=device)
weights[:, :10] = 0.0 # mask first 10 tokens (simulate user tokens)
weights[:, 20:40] = 1.5 # boost tokens 20-40 (simulate think tokens in Phase 3)
with torch.autocast(device_type=device.type, dtype=torch.bfloat16):
logits, loss = model(ids, targets, weights)
print(f"Forward pass OK β loss: {loss.item():.4f}")
print(f"Logits shape during training: {logits.shape} (should be empty [0])")
# Test generation
prompt = torch.randint(0, config.vocab_size, (1, 10), device=device)
with torch.autocast(device_type=device.type, dtype=torch.bfloat16):
generated = model.generate(prompt, max_new_tokens=20, temperature=0.8)
print(f"Generation OK β output shape: {generated.shape}")
print("\nβ
All checks passed. model.py is ready.\n")
|