import torch
import torch.nn as nn
import torch.nn.functional as F

VOCAB = 16000
CTX = 512
N_LAYER = 16
N_HEAD = 10
N_EMBD = 640


class Block(nn.Module):
    def __init__(self):
        super().__init__()
        self.ln1 = nn.LayerNorm(N_EMBD)
        self.qkv = nn.Linear(N_EMBD, 3 * N_EMBD)
        self.proj = nn.Linear(N_EMBD, N_EMBD)
        self.ln2 = nn.LayerNorm(N_EMBD)
        self.fc = nn.Linear(N_EMBD, 4 * N_EMBD)
        self.fc2 = nn.Linear(4 * N_EMBD, N_EMBD)

    def forward(self, x):
        B, T, C = x.shape
        q, k, v = self.qkv(self.ln1(x)).split(C, dim=2)
        q = q.view(B, T, N_HEAD, C // N_HEAD).transpose(1, 2)
        k = k.view(B, T, N_HEAD, C // N_HEAD).transpose(1, 2)
        v = v.view(B, T, N_HEAD, C // N_HEAD).transpose(1, 2)
        y = F.scaled_dot_product_attention(q, k, v, is_causal=True)
        y = y.transpose(1, 2).contiguous().view(B, T, C)
        x = x + self.proj(y)
        x = x + self.fc2(F.gelu(self.fc(self.ln2(x))))
        return x


class GPT(nn.Module):
    def __init__(self):
        super().__init__()
        self.tok_emb = nn.Embedding(VOCAB, N_EMBD)
        self.pos_emb = nn.Embedding(CTX, N_EMBD)
        self.blocks = nn.ModuleList([Block() for _ in range(N_LAYER)])
        self.ln_f = nn.LayerNorm(N_EMBD)
        self.head = nn.Linear(N_EMBD, VOCAB, bias=False)
        self.head.weight = self.tok_emb.weight  # связанные веса: экономим память
        self.apply(self._init)

    def _init(self, m):
        if isinstance(m, (nn.Linear, nn.Embedding)):
            nn.init.normal_(m.weight, std=0.02)
            if isinstance(m, nn.Linear) and m.bias is not None:
                nn.init.zeros_(m.bias)

    def forward(self, idx, targets=None):
        B, T = idx.shape
        pos = torch.arange(T, device=idx.device)
        x = self.tok_emb(idx) + self.pos_emb(pos)
        for b in self.blocks:
            x = b(x)
        logits = self.head(self.ln_f(x))
        loss = None
        if targets is not None:
            loss = F.cross_entropy(logits.float().view(-1, VOCAB), targets.view(-1))
        return logits, loss


@torch.no_grad()
def sample(model, tok, prompt, device, max_new=120, temp=0.8, top_k=40):
    was_training = model.training
    model.eval()
    idx = torch.tensor([tok.encode(prompt).ids], device=device)
    for _ in range(max_new):
        with torch.autocast("cuda", dtype=torch.float16):
            logits, _ = model(idx[:, -CTX:])
        logits = logits[:, -1, :].float() / temp
        v, _ = torch.topk(logits, top_k)
        logits[logits < v[:, [-1]]] = -float("inf")
        nxt = torch.multinomial(F.softmax(logits, dim=-1), 1)
        idx = torch.cat([idx, nxt], dim=1)
    model.train(was_training)
    return tok.decode(idx[0].tolist())