model_class.py

import torch
import torch.nn as nn
from transformers import PreTrainedModel, PretrainedConfig


class MultiheadAttention(nn.Module):
    def __init__(self, d_in, d_out, context_length, dropout, num_heads, qkv_bias=False):
        super().__init__()

        self.d_out = d_out
        self.num_heads = num_heads
        self.head_dim = d_out // num_heads

        #step 3
        self.W_query = nn.Linear(d_in, d_out, bias=qkv_bias)
        self.W_key = nn.Linear(d_in, d_out, bias=qkv_bias)
        self.W_value = nn.Linear(d_in, d_out, bias=qkv_bias)

        self.out_proj = nn.Linear(d_out, d_out)
        self.dropout = nn.Dropout(dropout)
        self.register_buffer("mask",torch.triu(torch.ones(context_length, context_length), diagonal=1))


    def forward(self, x):
        b, num_tokens, d_in = x.shape

        #step 4
        keys = self.W_key(x)
        queries = self.W_query(x)
        values = self.W_value(x)

        #step 5
        keys = keys.view(b, num_tokens, self.num_heads, self.head_dim)
        queries = queries.view(b, num_tokens, self.num_heads, self.head_dim)
        values = values.view(b, num_tokens, self.num_heads, self.head_dim)

        #step 6
        keys = keys.transpose(1,2)
        queries = queries.transpose(1,2)
        values = values.transpose(1,2)

        #step 7
        attn_scores = queries @ keys.transpose(2,3)

        #step 8
        mask_bool = self.mask.bool()[:num_tokens, :num_tokens]
        attn_scores.masked_fill_(mask_bool, -torch.inf)

        attn_weights = torch.softmax(attn_scores / keys.shape[-1]**0.5, dim=-1)
        attn_weights = self.dropout(attn_weights)

        #step 9 - 11
        ctx_vec = (attn_weights @ values).transpose(1, 2)

        #step 12
        ctx_vec = ctx_vec.contiguous().view(b, num_tokens, self.d_out)
        ctx_vec = self.out_proj(ctx_vec)

        return ctx_vec

#==========================================================================


class LayerNorm(nn.Module):
    def __init__(self, emb_dim):
        super().__init__()
        self.eps = 1e-5
        self.scale = nn.Parameter(torch.ones(emb_dim))
        self.shift = nn.Parameter(torch.zeros(emb_dim))

    def forward(self, x):
        mean = x.mean(dim=-1, keepdim=True)
        var = x.var(dim=-1, keepdim=True, unbiased=False)
        norm_x = (x - mean) / torch.sqrt(var + self.eps)
        return self.scale * norm_x + self.shift

#==========================================================================


class GeLU(nn.Module):
    def __init__(self):
        super().__init__()

    def forward(self, x):
        return 0.5 * x * (1 + torch.tanh(torch.sqrt(torch.tensor(2.0/torch.pi)) * (x + 0.044715 * torch.pow(x,3))))

#==========================================================================


class FeedForward(nn.Module):
    def __init__(self, cfg):
        super().__init__()
        self.layers = nn.Sequential(
            nn.Linear(cfg.emb_dim, 4*cfg.emb_dim),
            GeLU(),
            nn.Linear(4*cfg.emb_dim, cfg.emb_dim)
        )
    def forward(self, x):
        return self.layers(x)

#==========================================================================

class TransformerBlock(nn.Module):
    def __init__(self, cfg):
        super().__init__()
        self.att = MultiheadAttention(
            d_in = cfg.emb_dim,
            d_out = cfg.emb_dim,
            context_length = cfg.context_length,
            dropout = cfg.drop_rate,
            num_heads = cfg.n_heads,
            qkv_bias = cfg.qkv_bias
        )
        self.ff = FeedForward(cfg)
        self.norm1 = LayerNorm(cfg.emb_dim)
        self.norm2 = LayerNorm(cfg.emb_dim)
        self.drop_shortcut = nn.Dropout(cfg.drop_rate)

    def forward(self, x):
        shortcut = x
        x = self.norm1(x)
        x = self.att(x)
        x = self.drop_shortcut(x)
        x = x + shortcut

        shortcut = x
        x = self.norm2(x)
        x = self.ff(x)
        x = self.drop_shortcut(x)
        x = x + shortcut

        return x

#=======================================================================

class TicketGPTConfig(PretrainedConfig):
    model_type = "ticket_gpt"  # Unique identifier for the AutoClass
    def __init__(self, classes=8, context_length=1024, drop_rate=0.1, emb_dim=768, n_heads=12, n_layers=12, qkv_bias=True, vocab_size=50257, **kwargs):
        super().__init__(**kwargs)
        self.classes = classes
        self.context_length = context_length
        self.drop_rate = drop_rate
        self.emb_dim = emb_dim
        self.n_heads = n_heads
        self.n_layers = n_layers
        self.qkv_bias = qkv_bias
        self.vocab_size = vocab_size

class TicketGPT(
    PreTrainedModel, 
):
    config_class = TicketGPTConfig
    def __init__(self, config):
        super().__init__(config)
        self.tok_emb = nn.Embedding(config.vocab_size, config.emb_dim)
        self.pos_emb = nn.Embedding(config.context_length, config.emb_dim)
        self.drop_emb = nn.Dropout(config.drop_rate)

        self.trf_blocks = nn.Sequential(
            *[TransformerBlock(config) for _ in range(config.n_layers)]
        )

        self.final_norm = LayerNorm(config.emb_dim)
        self.out_head = nn.Linear(config.emb_dim, config.classes, bias=True)

    def forward(self, x):
        batch_size, seq_len = x.shape
        tok_embeddings = self.tok_emb(x) #[2,4,768]
        pos_embeddings = self.pos_emb(torch.arange(seq_len, device=x.device)) #[2,4,768]
        x = tok_embeddings + pos_embeddings #[2,4,768]
        x = self.drop_emb(x)
        x = self.trf_blocks(x)
        x = self.final_norm(x)
        logits = self.out_head(x) #[2,4,50257]
        return logits

    def predict(self, text, tokenizer, max_length=1024, pad_token_id=50256):
        lookup = {
            0:"Hardware",
            1:"HR Support",
            2:"Access",
            3:"Miscellaneous",
            4:"Storage",
            5:"Purchase",
            6:"Internal Project",
            7:"Administrative rights"
        }
        
        current_device = next(self.parameters()).device
        self.eval()
    
        # Prepare inputs to the model
        input_ids = tokenizer.encode(text)
        supported_context_length = self.config.context_length
    
        # Truncate sequences if they too long
        input_ids = input_ids[:min(max_length, supported_context_length)]
    
        # Pad sequences to the longest sequence
        input_ids += [pad_token_id] * (max_length - len(input_ids))
        input_tensor = torch.tensor(input_ids, device=current_device).unsqueeze(0) # add batch dimension
    
        # Model inference
        with torch.no_grad():
            logits = self(input_tensor)[:, -1, :]  # Logits of the last output token
        predicted_label = torch.argmax(logits, dim=-1).item()
    
        # Return the classified result
        return lookup[predicted_label]