| """Modern BERTc v3 — ModernBERT release-aligned + Cramming-style ScaledSinusoidal PE. |
| |
| 主要按 release `modernbert-base-pretrain.yaml` 对齐(除 Alt Attn 和 PE): |
| - 架构: 22L / 768H / 1152I (GLU) / 12 heads,head_dim=64 |
| - **ScaledSinusoidal 位置编码**(Hua et al. 2022 FLASH;Cramming 实测短 seq 比 |
| RoPE 更值:计算几乎免费,RoPE 收益被 5-10% 速度损失抵消) |
| - GeGLU FFN(glu + gelu) |
| - LayerNorm 无 bias(eps=1e-5),非 RMSNorm |
| - pre-norm 布局 + skip_first_prenorm |
| - embed_norm + final_norm |
| - Megatron-style init:残差层 W 缩 1/sqrt(2L) |
| - 全无 Linear bias |
| - Dropout: 全 0(Cramming 论据:short single-epoch 无 overfit risk) |
| - tied word embedding |
| - flex_attention compiled(支持 cross-doc 隔离 via seg_ids) |
| |
| 不上的 ModernBERT 特性: |
| - Alternating Attention(我们走全局 attention) |
| - Unpadded packing + cu_seqlens(我们定长 pack) |
| - RoPE(换 ScaledSinusoidal,见 Cramming Section 4.2) |
| |
| 参数量(默认 22L/768H/1152I,V=12536): |
| emb (tied) : 12536 × 768 ≈ 9.6M |
| embed_norm : 768 × 1 ≈ 1K(no bias) |
| per layer: |
| norm1/2 : 768 × 2 ≈ 1.5K |
| Q K V O : 4 × 768² ≈ 2.36M |
| GeGLU (W_in=2I, W_out): 768×2304 + 1152×768 ≈ 2.66M |
| total per layer : ≈ 5.0M |
| × 22 layers : ≈ 110M |
| final_norm : 768 × 1 |
| head: dense + norm + gelu: 768×768 + 768 ≈ 0.59M |
| head_bias (V,) : 12.5K |
| total ≈ 130M |
| """ |
| from dataclasses import dataclass, field |
| from typing import Optional |
| import math |
|
|
| import torch |
| import torch.nn as nn |
| import torch.nn.functional as F |
| from torch.nn.attention.flex_attention import ( |
| flex_attention as _flex_attention_raw, |
| create_block_mask, |
| ) |
|
|
| |
| |
| |
| |
| _flex_attention = torch.compile(_flex_attention_raw) |
|
|
|
|
| |
|
|
| @dataclass |
| class ModernBertConfig: |
| vocab_size: int = 12536 |
| hidden_size: int = 768 |
| num_hidden_layers: int = 22 |
| num_attention_heads: int = 12 |
| intermediate_size: int = 1152 |
| max_position_embeddings: int = 1024 |
| pad_token_id: int = 12531 |
| mask_token_id: int = 12535 |
| pe_theta: float = 10000.0 |
| layer_norm_eps: float = 1e-5 |
| initializer_range: float = 0.02 |
| tie_word_embeddings: bool = True |
| |
| embed_dropout: float = 0.0 |
| mlp_dropout: float = 0.0 |
| attn_out_dropout: float = 0.1 |
| attn_probs_dropout: float = 0.0 |
| |
| embed_norm: bool = True |
| skip_first_prenorm: bool = True |
| final_norm: bool = True |
| |
| init_method: str = "megatron" |
|
|
| @property |
| def head_dim(self) -> int: |
| assert self.hidden_size % self.num_attention_heads == 0 |
| return self.hidden_size // self.num_attention_heads |
|
|
|
|
| |
|
|
| class LayerNormNoBias(nn.Module): |
| """LayerNorm with weight only (no bias). 对齐 ModernBERT release。""" |
| def __init__(self, hidden_size: int, eps: float = 1e-5): |
| super().__init__() |
| self.weight = nn.Parameter(torch.ones(hidden_size)) |
| self.eps = eps |
| self.normalized_shape = (hidden_size,) |
|
|
| def forward(self, x: torch.Tensor) -> torch.Tensor: |
| return F.layer_norm(x, self.normalized_shape, self.weight, None, self.eps) |
|
|
|
|
| |
|
|
| class ScaledSinusoidalPE(nn.Module): |
| """Scaled sinusoidal positional embedding(Hua 2022 FLASH paper)。 |
| |
| 标准 sinusoidal:PE[pos, 2i]=sin(pos/θ^(2i/d)), PE[pos, 2i+1]=cos(...)。 |
| `scale_factor` 是一个 learnable 标量,初始 1/sqrt(d)。 |
| 用法:embedding 之后直接 `x = embed + pos_emb(input_ids)`,跟所有层共享。 |
| 比 RoPE 便宜:只在 embedding 层 fire 一次,attention 里 0 开销。 |
| """ |
| def __init__(self, embedding_dim: int, max_seq_length: int, theta: float = 10000.0): |
| super().__init__() |
| pe = torch.zeros(max_seq_length, embedding_dim) |
| position = torch.arange(0, max_seq_length, dtype=torch.float).unsqueeze(1) |
| div_term = torch.exp( |
| torch.arange(0, embedding_dim, 2).float() * (-math.log(theta) / embedding_dim) |
| ) |
| pe[:, 0::2] = torch.sin(position * div_term) |
| pe[:, 1::2] = torch.cos(position * div_term) |
| pe = pe.unsqueeze(0) |
| self.register_buffer("pe", pe, persistent=False) |
| self.scale_factor = nn.Parameter(torch.tensor([1.0 / embedding_dim ** 0.5])) |
|
|
| def forward(self, seq_len: int) -> torch.Tensor: |
| return self.scale_factor * self.pe[:, :seq_len, :] |
|
|
|
|
| |
|
|
| class ModernBertAttention(nn.Module): |
| def __init__(self, config: ModernBertConfig): |
| super().__init__() |
| self.num_heads = config.num_attention_heads |
| self.head_dim = config.head_dim |
| self.scale = self.head_dim ** -0.5 |
| self.attn_probs_dropout = config.attn_probs_dropout |
| |
| self.qkv = nn.Linear(config.hidden_size, 3 * config.hidden_size, bias=False) |
| self.o = nn.Linear(config.hidden_size, config.hidden_size, bias=False) |
| self.out_dropout = nn.Dropout(config.attn_out_dropout) |
|
|
| def forward(self, x: torch.Tensor, |
| block_mask=None, |
| attention_mask: Optional[torch.Tensor] = None) -> torch.Tensor: |
| """三种 attention 模式: |
| - block_mask 非空 → flex_attention(block-diag,跨 doc 隔离,训练时) |
| - block_mask 空,attention_mask 非空 → SDPA + pad mask(fine-tune) |
| - 都空 → SDPA 全可见 |
| 位置信息走 ScaledSinusoidal,已在 embedding 层加,attention 里无 cos/sin 计算。 |
| """ |
| B, L, H = x.shape |
| qkv = self.qkv(x).reshape(B, L, 3, self.num_heads, self.head_dim) |
| q, k, v = qkv.unbind(dim=2) |
| q = q.transpose(1, 2) |
| k = k.transpose(1, 2) |
| v = v.transpose(1, 2) |
|
|
| if block_mask is not None: |
| |
| out = _flex_attention(q, k, v, block_mask=block_mask) |
| else: |
| sdpa_mask = None |
| if attention_mask is not None: |
| sdpa_mask = attention_mask[:, None, None, :].to(torch.bool) |
| out = F.scaled_dot_product_attention( |
| q, k, v, |
| attn_mask=sdpa_mask, |
| dropout_p=self.attn_probs_dropout if self.training else 0.0, |
| is_causal=False, |
| ) |
| out = out.transpose(1, 2).reshape(B, L, H) |
| return self.out_dropout(self.o(out)) |
|
|
|
|
| |
|
|
| class GeGLU(nn.Module): |
| """Linear(H, 2*I) → split → GELU(gate) * up → Linear(I, H). |
| 全无 bias。 |
| """ |
| def __init__(self, config: ModernBertConfig): |
| super().__init__() |
| I = config.intermediate_size |
| self.w_in = nn.Linear(config.hidden_size, 2 * I, bias=False) |
| self.w_out = nn.Linear(I, config.hidden_size, bias=False) |
| self.dropout = nn.Dropout(config.mlp_dropout) |
|
|
| def forward(self, x: torch.Tensor) -> torch.Tensor: |
| gate, up = self.w_in(x).chunk(2, dim=-1) |
| return self.dropout(self.w_out(F.gelu(gate) * up)) |
|
|
|
|
| |
|
|
| class ModernBertLayer(nn.Module): |
| def __init__(self, config: ModernBertConfig, is_first: bool = False): |
| super().__init__() |
| |
| |
| self.skip_norm1 = is_first and config.skip_first_prenorm |
| self.norm1 = nn.Identity() if self.skip_norm1 else LayerNormNoBias(config.hidden_size, eps=config.layer_norm_eps) |
| self.attn = ModernBertAttention(config) |
| self.norm2 = LayerNormNoBias(config.hidden_size, eps=config.layer_norm_eps) |
| self.mlp = GeGLU(config) |
|
|
| def forward(self, x, block_mask=None, attention_mask=None): |
| x = x + self.attn(self.norm1(x), block_mask, attention_mask) |
| x = x + self.mlp(self.norm2(x)) |
| return x |
|
|
|
|
| |
|
|
| class ModernBertModel(nn.Module): |
| def __init__(self, config: ModernBertConfig): |
| super().__init__() |
| self.config = config |
| self.embed = nn.Embedding(config.vocab_size, config.hidden_size, |
| padding_idx=config.pad_token_id) |
| |
| self.pos_emb = ScaledSinusoidalPE( |
| embedding_dim=config.hidden_size, |
| max_seq_length=config.max_position_embeddings, |
| theta=config.pe_theta, |
| ) |
| self.embed_norm = (LayerNormNoBias(config.hidden_size, eps=config.layer_norm_eps) |
| if config.embed_norm else nn.Identity()) |
| self.embed_dropout = nn.Dropout(config.embed_dropout) |
| self.layers = nn.ModuleList( |
| [ModernBertLayer(config, is_first=(i == 0)) |
| for i in range(config.num_hidden_layers)] |
| ) |
| self.final_norm = (LayerNormNoBias(config.hidden_size, eps=config.layer_norm_eps) |
| if config.final_norm else nn.Identity()) |
|
|
| |
| self.apply(self._init_weights) |
| if config.init_method == "megatron": |
| self._megatron_residual_init() |
|
|
| def _init_weights(self, m): |
| if isinstance(m, nn.Linear): |
| nn.init.normal_(m.weight, std=self.config.initializer_range) |
| if m.bias is not None: |
| nn.init.zeros_(m.bias) |
| elif isinstance(m, nn.Embedding): |
| nn.init.normal_(m.weight, std=self.config.initializer_range) |
| if m.padding_idx is not None: |
| with torch.no_grad(): |
| m.weight[m.padding_idx].zero_() |
| elif isinstance(m, LayerNormNoBias): |
| nn.init.ones_(m.weight) |
|
|
| def _megatron_residual_init(self): |
| """对每个 residual 路径的输出 W 缩 1/sqrt(2*L)。 |
| 防止深层网络早期 forward variance 爆炸。 |
| residual outputs: attn.o, mlp.w_out。 |
| """ |
| L = self.config.num_hidden_layers |
| scale = (2.0 * L) ** -0.5 |
| for layer in self.layers: |
| with torch.no_grad(): |
| layer.attn.o.weight.mul_(scale) |
| layer.mlp.w_out.weight.mul_(scale) |
|
|
| def forward(self, input_ids: torch.Tensor, |
| seg_ids: Optional[torch.Tensor] = None, |
| attention_mask: Optional[torch.Tensor] = None) -> torch.Tensor: |
| """seg_ids: [B, L] int32/uint8,同 doc 同 id;非空时走 flex_attention 跨 doc 隔离。 |
| attention_mask: [B, L] 0/1,只在 seg_ids=None 时使用(fine-tune 路径)。 |
| 位置信息:ScaledSinusoidal 加在 embedding 后,attention 内部无位置计算。 |
| """ |
| B, L = input_ids.shape |
| x = self.embed(input_ids) |
| x = x + self.pos_emb(L).to(x.dtype) |
| x = self.embed_norm(x) |
| x = self.embed_dropout(x) |
|
|
| block_mask = self._build_block_mask(seg_ids, B, L) if seg_ids is not None else None |
|
|
| for layer in self.layers: |
| x = layer(x, block_mask, attention_mask) |
| x = self.final_norm(x) |
| return x |
|
|
| def _build_block_mask(self, seg_ids: torch.Tensor, B: int, L: int): |
| """seg_ids: [B, L] 用 flex_attention 构造 doc-internal mask。 |
| mask_mod 闭包捕获 seg_ids,在 batch/query/kv 索引下查 doc 是否一致。 |
| """ |
| seg_ids_long = seg_ids.to(torch.int32) |
|
|
| def mask_mod(b, h, q_idx, kv_idx): |
| return seg_ids_long[b, q_idx] == seg_ids_long[b, kv_idx] |
|
|
| |
| return create_block_mask(mask_mod, B=B, H=None, Q_LEN=L, KV_LEN=L, |
| device=seg_ids.device) |
|
|
|
|
| |
|
|
| class ModernBertForMLM(nn.Module): |
| """MLM head 简化版(Cramming Section 4.2 推荐): |
| - 无 nonlinear head(去 Dense + LN + GeLU)— "without ill effect" |
| - 无 decoder bias(去 head_bias)— "drop the decoder bias" |
| - 仅 tied embedding projection:logits = h @ embed.weight.T |
| - final LayerNorm 已经在 bert.final_norm 提供,这里不需重复 |
| 省参数 ~0.6M,forward 略快。""" |
| def __init__(self, config: ModernBertConfig): |
| super().__init__() |
| self.config = config |
| self.bert = ModernBertModel(config) |
|
|
| def get_input_embeddings(self): |
| return self.bert.embed |
|
|
| def forward(self, input_ids, seg_ids=None, attention_mask=None, labels=None): |
| h = self.bert(input_ids, seg_ids=seg_ids, |
| attention_mask=attention_mask) |
| |
| logits = F.linear(h, self.bert.embed.weight) |
| loss = None |
| if labels is not None: |
| loss = F.cross_entropy( |
| logits.view(-1, self.config.vocab_size), |
| labels.view(-1), |
| ignore_index=-100, |
| ) |
| return {"logits": logits, "loss": loss} |
|
|
| def num_parameters(self): |
| return sum(p.numel() for p in self.parameters() if p.requires_grad) |
|
|
|
|
| |
|
|
| if __name__ == "__main__": |
| print("=== Test 1: default config (22L/768H release-aligned) ===") |
| cfg = ModernBertConfig() |
| print(f"Config: V={cfg.vocab_size} H={cfg.hidden_size} L={cfg.num_hidden_layers} " |
| f"head={cfg.num_attention_heads} d={cfg.head_dim} I={cfg.intermediate_size} " |
| f"pe_theta={cfg.pe_theta} ln_eps={cfg.layer_norm_eps}") |
| print(f"embed_norm={cfg.embed_norm} skip_first_prenorm={cfg.skip_first_prenorm} " |
| f"final_norm={cfg.final_norm} init={cfg.init_method}") |
| print(f"dropout: embed={cfg.embed_dropout} mlp={cfg.mlp_dropout} " |
| f"attn_out={cfg.attn_out_dropout} attn_probs={cfg.attn_probs_dropout}") |
| model = ModernBertForMLM(cfg) |
| n = model.num_parameters() |
| print(f"params: {n:,} = {n / 1e6:.1f}M") |
|
|
| |
| B, L = 2, 128 |
| ids = torch.randint(0, cfg.vocab_size, (B, L)) |
| mask = torch.ones(B, L, dtype=torch.long) |
| labels = torch.randint(0, cfg.vocab_size, (B, L)) |
| out = model(ids, attention_mask=mask, labels=labels) |
| print(f"logits: {out['logits'].shape} loss: {out['loss'].item():.4f}") |
|
|
| |
| out["loss"].backward() |
| |
| no_grad = [n for n, p in model.named_parameters() if p.requires_grad and p.grad is None] |
| if no_grad: |
| print(f"ERROR: 缺 grad 的参数: {no_grad}") |
| else: |
| print("backward OK (所有参数都有 grad)") |
|
|
| |
| assert model.bert.layers[0].skip_norm1, "layer[0].skip_norm1 应为 True" |
| assert not model.bert.layers[1].skip_norm1, "layer[1].skip_norm1 应为 False" |
| assert isinstance(model.bert.layers[0].norm1, nn.Identity), "layer[0].norm1 应为 Identity" |
| assert isinstance(model.bert.layers[1].norm1, LayerNormNoBias), "layer[1].norm1 应为 LayerNormNoBias" |
| print("skip_first_prenorm 配置正确") |
|
|
| |
| assert isinstance(model.bert.embed_norm, LayerNormNoBias), "embed_norm 应为 LayerNormNoBias" |
| assert isinstance(model.bert.final_norm, LayerNormNoBias), "final_norm 应为 LayerNormNoBias" |
| print("embed_norm + final_norm 配置正确") |
|
|
| |
| L_layers = cfg.num_hidden_layers |
| expected_scale = (2.0 * L_layers) ** -0.5 |
| |
| o_std = model.bert.layers[5].attn.o.weight.std().item() |
| expected_std = cfg.initializer_range * expected_scale |
| |
| assert 0.5 * expected_std < o_std < 1.5 * expected_std, \ |
| f"Megatron init: attn.o.weight.std={o_std:.6f}, expected≈{expected_std:.6f}" |
| print(f"Megatron init OK: attn.o.weight.std={o_std:.6f} ≈ {expected_std:.6f}") |
|
|
| |
| for name, p in model.named_parameters(): |
| if "bias" in name and name != "head_bias": |
| print(f"ERROR: 不该有的 bias: {name}") |
| print("Linear/Norm no-bias 配置正确") |
|
|
| print("\n=== Test 1b: seg_ids → flex_attention 路径(cross-doc 隔离)===") |
| if torch.cuda.is_available(): |
| |
| model_cuda = ModernBertForMLM(cfg).cuda().to(torch.bfloat16) |
| ids_c = ids.cuda() |
| labels_c = labels.cuda() |
| |
| seg_ids = torch.zeros(B, L, dtype=torch.int32, device="cuda") |
| seg_ids[:, L // 2:] = 1 |
| out_flex = model_cuda(ids_c, seg_ids=seg_ids, labels=labels_c) |
| print(f" flex_attention forward OK loss={out_flex['loss'].item():.4f} " |
| f"logits={out_flex['logits'].shape} dtype={out_flex['logits'].dtype}") |
| out_flex["loss"].backward() |
| no_grad = [n for n, p in model_cuda.named_parameters() if p.requires_grad and p.grad is None] |
| if no_grad: |
| print(f" ERROR: 缺 grad: {no_grad[:5]}") |
| else: |
| print(f" flex_attention backward OK") |
|
|
| |
| |
| with torch.no_grad(): |
| seg_v1 = torch.zeros(B, L, dtype=torch.int32, device="cuda") |
| seg_v2 = torch.zeros(B, L, dtype=torch.int32, device="cuda") |
| seg_v2[:, :L // 2] = 0 |
| seg_v2[:, L // 2:] = 1 |
| h1 = model_cuda.bert(ids_c, seg_ids=seg_v1) |
| h2 = model_cuda.bert(ids_c, seg_ids=seg_v2) |
| |
| front_diff = (h1[:, :L // 2] - h2[:, :L // 2]).abs().mean().item() |
| print(f" 前半 token 输出差异(应 > 0,验证隔离生效):{front_diff:.6f}") |
| assert front_diff > 1e-3, f"隔离未生效:front_diff={front_diff}" |
| print(f" 跨 doc 隔离工作正常 ✓") |
| else: |
| print(" 跳过(无 CUDA)") |
|
|
| print("\n=== Test 2: legacy v1 config (12L/1024H) — 验证向后兼容 ===") |
| cfg_v1 = ModernBertConfig( |
| hidden_size=1024, num_hidden_layers=12, num_attention_heads=16, |
| intermediate_size=2752, |
| embed_norm=False, skip_first_prenorm=False, init_method="normal", |
| ) |
| model_v1 = ModernBertForMLM(cfg_v1) |
| n_v1 = model_v1.num_parameters() |
| print(f"v1-style params: {n_v1 / 1e6:.1f}M") |
| out_v1 = model_v1(ids, attention_mask=mask, labels=labels) |
| print(f"v1 forward OK loss={out_v1['loss'].item():.4f}") |
|
|
| print("\n=== All smoke tests passed ===") |
|
|