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

class NeoraConfig(PretrainedConfig):
    model_type = "neora"
    def __init__(
        self,
        vocab_size=24000,
        d_model=1280,
        n_layers=24,
        n_heads=20,
        ffn_dim=5120,
        max_seq_len=2048,
        **kwargs
    ):
        super().__init__(**kwargs)
        self.vocab_size = vocab_size
        self.d_model = d_model
        self.n_layers = n_layers
        self.n_heads = n_heads
        self.ffn_dim = ffn_dim
        self.max_seq_len = max_seq_len

class TransformerBlock(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.norm1 = nn.LayerNorm(config.d_model)
        self.q_proj = nn.Linear(config.d_model, config.d_model)
        self.k_proj = nn.Linear(config.d_model, config.d_model)
        self.v_proj = nn.Linear(config.d_model, config.d_model)
        self.o_proj = nn.Linear(config.d_model, config.d_model)
        self.norm2 = nn.LayerNorm(config.d_model)
        self.fc1 = nn.Linear(config.d_model, config.ffn_dim)
        self.fc2 = nn.Linear(config.ffn_dim, config.d_model)
        self.n_heads = config.n_heads

    def forward(self, x):
        residual = x
        x = self.norm1(x)
        q, k, v = self.q_proj(x), self.k_proj(x), self.v_proj(x)
        batch, seq, dim = q.shape
        head_dim = dim // self.n_heads
        q = q.view(batch, seq, self.n_heads, head_dim).transpose(1, 2)
        k = k.view(batch, seq, self.n_heads, head_dim).transpose(1, 2)
        v = v.view(batch, seq, self.n_heads, head_dim).transpose(1, 2)
        x = F.scaled_dot_product_attention(q, k, v, is_causal=True)
        x = x.transpose(1, 2).contiguous().view(batch, seq, dim)
        x = self.o_proj(x)
        x = residual + x
        residual = x
        x = self.norm2(x)
        x = self.fc2(F.gelu(self.fc1(x)))
        return residual + x

class NeoraModel(PreTrainedModel):
    config_class = NeoraConfig

    def __init__(self, config):
        super().__init__(config)
        self.token_embedding = nn.Embedding(config.vocab_size, config.d_model)
        self.position_embedding = nn.Embedding(config.max_seq_len, config.d_model)
        self.blocks = nn.ModuleList([TransformerBlock(config) for _ in range(config.n_layers)])
        self.final_norm = nn.LayerNorm(config.d_model)
        self.lm_head = nn.Linear(config.d_model, config.vocab_size, bias=False)
        self.lm_head.weight = self.token_embedding.weight

    def forward(self, input_ids):
        batch, seq = input_ids.shape
        positions = torch.arange(seq, device=input_ids.device)
        x = self.token_embedding(input_ids) + self.position_embedding(positions)
        for block in self.blocks:
            x = block(x)
        return self.lm_head(self.final_norm(x))