File size: 5,359 Bytes
ce186be
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import torch
import torch.nn as nn
import torch.nn.functional as F
from transformers import PreTrainedModel, PretrainedConfig
from typing import Optional, Tuple

# Optional: import custom config if present
try:
    from .configuration_snowflake_core import SnowflakeCoreConfig
except ImportError:
    SnowflakeCoreConfig = PretrainedConfig

class FusedSelfAttention(nn.Module):
    def __init__(self, embed_dim, num_heads):
        super().__init__()
        self.num_heads = num_heads
        self.head_dim = embed_dim // num_heads
        assert (
            self.head_dim * num_heads == embed_dim
        ), "embed_dim must be divisible by num_heads"
        self.qkv_proj = nn.Linear(embed_dim, 3 * embed_dim)
        self.out_proj = nn.Linear(embed_dim, embed_dim)

    def forward(self, x, attn_mask=None, key_padding_mask=None):
        B, T, C = x.size()
        qkv = self.qkv_proj(x)  # [B, T, 3 * C]
        qkv = qkv.reshape(B, T, 3, self.num_heads, self.head_dim).permute(2, 0, 3, 1, 4)
        q, k, v = qkv[0], qkv[1], qkv[2]  # Each: [B, num_heads, T, head_dim]

        attn_scores = (q @ k.transpose(-2, -1)) / (self.head_dim ** 0.5)  # [B, num_heads, T, T]
        if attn_mask is not None:
            attn_scores = attn_scores + attn_mask.unsqueeze(0).unsqueeze(0).to(attn_scores.dtype)
        if key_padding_mask is not None:
            attn_scores = attn_scores.masked_fill(key_padding_mask.unsqueeze(1).unsqueeze(2), float('-inf'))
        attn_probs = F.softmax(attn_scores, dim=-1)
        attn_output = attn_probs @ v  # [B, num_heads, T, head_dim]
        attn_output = attn_output.transpose(1, 2).reshape(B, T, C)
        return self.out_proj(attn_output)

class GPTBlock(nn.Module):
    def __init__(self, embed_dim, num_heads, dropout=0.1):
        super().__init__()
        self.ln1 = nn.LayerNorm(embed_dim)
        self.attn = FusedSelfAttention(embed_dim, num_heads)
        self.dropout1 = nn.Dropout(dropout)
        self.ln2 = nn.LayerNorm(embed_dim)
        self.mlp = nn.Sequential(
            nn.Linear(embed_dim, 4 * embed_dim),
            nn.GELU(),
            nn.Dropout(dropout),
            nn.Linear(4 * embed_dim, embed_dim),
        )
        self.dropout2 = nn.Dropout(dropout)
    def forward(self, x, attn_mask=None, key_padding_mask=None):
        h = self.ln1(x)
        attn_output = self.attn(h, attn_mask=attn_mask, key_padding_mask=key_padding_mask)
        x = x + self.dropout1(attn_output)
        x = x + self.dropout2(self.mlp(self.ln2(x)))
        return x

class SnowflakeCoreG1(PreTrainedModel):
    config_class = SnowflakeCoreConfig
    supports_gradient_checkpointing = True

    def __init__(self, config):
        super().__init__(config)
        self.vocab_size = config.vocab_size
        self.embed_dim = config.embed_dim
        self.num_heads = config.num_heads
        self.num_layers = config.num_layers
        self.max_length = config.max_length
        self.ffn_dim = getattr(config, 'ffn_dim', 4 * config.embed_dim)
        self.dropout = getattr(config, 'dropout', 0.1)

        self.embed = nn.Embedding(self.vocab_size, self.embed_dim)
        self.pos_embed = nn.Embedding(self.max_length, self.embed_dim)
        self.dropout_layer = nn.Dropout(self.dropout)
        self.blocks = nn.ModuleList([
            GPTBlock(self.embed_dim, self.num_heads, self.dropout) for _ in range(self.num_layers)
        ])
        self.ln_f = nn.LayerNorm(self.embed_dim)
        self.lm_head = nn.Linear(self.embed_dim, self.vocab_size, bias=False)

        self.post_init()

    def get_input_embeddings(self):
        return self.embed

    def set_input_embeddings(self, value):
        self.embed = value

    def get_output_embeddings(self):
        return self.lm_head

    def set_output_embeddings(self, new_embeddings):
        self.lm_head = new_embeddings

    def forward(
        self,
        input_ids: torch.LongTensor = None,
        attention_mask: Optional[torch.Tensor] = None,
        labels: Optional[torch.LongTensor] = None,
        **kwargs
    ) -> Tuple:
        B, T = input_ids.size()
        pos = torch.arange(0, T, device=input_ids.device).unsqueeze(0)
        x = self.embed(input_ids) + self.pos_embed(pos)
        x = self.dropout_layer(x)
        causal_mask = torch.triu(torch.ones(T, T, device=input_ids.device), diagonal=1).bool()
        causal_mask = causal_mask.masked_fill(causal_mask, float('-inf'))
        key_padding_mask = None
        if attention_mask is not None:
            key_padding_mask = attention_mask == 0
        for block in self.blocks:
            x = block(x, attn_mask=causal_mask, key_padding_mask=key_padding_mask)
        x = self.ln_f(x)
        logits = self.lm_head(x)

        loss = None
        if labels is not None:
            shift_logits = logits[:, :-1, :].contiguous().view(-1, self.vocab_size)
            shift_labels = labels[:, 1:].contiguous().view(-1)
            loss = F.cross_entropy(shift_logits, shift_labels, ignore_index=self.config.pad_token_id)
        if loss is not None:
            return {"loss": loss, "logits": logits}
        return {"logits": logits}

    @classmethod
    def from_pretrained(cls, pretrained_model_name_or_path, *model_args, config=None, **kwargs):
        return super().from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)