diff --git a/assets/drawio/gpt1-architecture.drawio b/assets/drawio/gpt1-architecture.drawio
new file mode 100644
index 0000000..541f48f
--- /dev/null
+++ b/assets/drawio/gpt1-architecture.drawio
@@ -0,0 +1,148 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/assets/drawio/gpt1-attention.drawio b/assets/drawio/gpt1-attention.drawio
new file mode 100644
index 0000000..97063b5
--- /dev/null
+++ b/assets/drawio/gpt1-attention.drawio
@@ -0,0 +1,413 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/assets/drawio/gpt1-decoder.drawio b/assets/drawio/gpt1-decoder.drawio
new file mode 100644
index 0000000..72d7a00
--- /dev/null
+++ b/assets/drawio/gpt1-decoder.drawio
@@ -0,0 +1,148 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/assets/drawio/gpt1-embeddings.drawio b/assets/drawio/gpt1-embeddings.drawio
new file mode 100644
index 0000000..e4b7362
--- /dev/null
+++ b/assets/drawio/gpt1-embeddings.drawio
@@ -0,0 +1,148 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/assets/drawio/gpt1-forward.drawio b/assets/drawio/gpt1-forward.drawio
new file mode 100644
index 0000000..24110ee
--- /dev/null
+++ b/assets/drawio/gpt1-forward.drawio
@@ -0,0 +1,192 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/assets/models/gpt1-architecture.png b/assets/models/gpt1-architecture.png
new file mode 100644
index 0000000..14a58df
Binary files /dev/null and b/assets/models/gpt1-architecture.png differ
diff --git a/assets/models/gpt1-attention.png b/assets/models/gpt1-attention.png
new file mode 100644
index 0000000..4229b44
Binary files /dev/null and b/assets/models/gpt1-attention.png differ
diff --git a/assets/models/gpt1-decoder.png b/assets/models/gpt1-decoder.png
new file mode 100644
index 0000000..8684428
Binary files /dev/null and b/assets/models/gpt1-decoder.png differ
diff --git a/assets/models/gpt1-embeddings.png b/assets/models/gpt1-embeddings.png
new file mode 100644
index 0000000..39845ed
Binary files /dev/null and b/assets/models/gpt1-embeddings.png differ
diff --git a/assets/models/gpt1-forward.png b/assets/models/gpt1-forward.png
new file mode 100644
index 0000000..650f94d
Binary files /dev/null and b/assets/models/gpt1-forward.png differ
diff --git a/llm/src/llm/core/decoder.py b/llm/src/llm/core/gpt_decoder.py
similarity index 92%
rename from llm/src/llm/core/decoder.py
rename to llm/src/llm/core/gpt_decoder.py
index 11de653..1960b6a 100644
--- a/llm/src/llm/core/decoder.py
+++ b/llm/src/llm/core/gpt_decoder.py
@@ -4,7 +4,7 @@ from .feed_forward import FeedForward
from .multi_head_attention import MultiHeadAttention
-class Decoder(nn.Module):
+class GptDecoder(nn.Module):
"""
Decoder — базовый transformer decoder block (pre-LN), классический строительный блок современных языковых моделей.
@@ -94,7 +94,13 @@ class Decoder(nn.Module):
self._norm1 = nn.LayerNorm(emb_size)
self._norm2 = nn.LayerNorm(emb_size)
- def forward(self, x: torch.Tensor, mask: torch.Tensor = None) -> torch.Tensor:
+ def forward(
+ self,
+ x: torch.Tensor,
+ use_cache: bool = False,
+ cache: list = None,
+ attention_mask=None
+ ) -> tuple:
"""
Один прямой проход через Transformer decoder block.
@@ -117,10 +123,16 @@ class Decoder(nn.Module):
- Применяем FFN к нормализованному результату (layernorm)
- Добавляем residual-связь (ffn + предыдущий выход)
"""
+
# Self-Attention блок
- attention, _ = self._heads(x, mask, use_cache=False, cache=None)
+ attention, kv_caches = self._heads(x, attention_mask, use_cache=use_cache, cache=cache)
out = self._norm1(attention + x)
# FeedForward блок
ffn_out = self._ff(out)
- return self._norm2(ffn_out + out)
+ result = self._norm2(ffn_out + out)
+
+ if use_cache:
+ return (result, kv_caches)
+ else:
+ return (result, None)
diff --git a/llm/src/llm/models/gemma/gemma.py b/llm/src/llm/models/gemma/gemma.py
index dc41bd7..94065d1 100644
--- a/llm/src/llm/models/gemma/gemma.py
+++ b/llm/src/llm/models/gemma/gemma.py
@@ -209,14 +209,17 @@ class Gemma(BaseModel):
else:
return (logits, None)
- def generate(self,
- x: torch.Tensor,
- max_new_tokens: int,
+ def generate(
+ self,
+ x: torch.Tensor,
+ max_new_tokens: int,
do_sample: bool,
temperature: float = 1.0,
top_k: int = None,
top_p: float = None,
- use_cache: bool = True
+ use_cache: bool = True,
+ attention_mask: torch.Tensor = None,
+ **kwargs
) -> torch.Tensor:
"""
Авторегрессивная генерация токенов с использованием greedy, temperature, top-k и top-p sampling.
diff --git a/llm/src/llm/models/gpt/gpt.py b/llm/src/llm/models/gpt/gpt.py
index 69394f6..cd947c1 100644
--- a/llm/src/llm/models/gpt/gpt.py
+++ b/llm/src/llm/models/gpt/gpt.py
@@ -26,7 +26,7 @@ import torch.nn as nn
import torch.nn.functional as F
from typing import Optional, Dict
from llm.core.base_model import BaseModel
-from llm.core.decoder import Decoder
+from llm.core.gpt_decoder import GptDecoder
from llm.core.token_embeddings import TokenEmbeddings
from llm.core.positional_embeddings import PositionalEmbeddings
@@ -116,7 +116,7 @@ class GPT(BaseModel):
# head_size = emb_size // num_heads
self._decoders = nn.ModuleList(
[
- Decoder(
+ GptDecoder(
num_heads=config["num_heads"],
emb_size=config["embed_dim"],
head_size=config["embed_dim"] // config["num_heads"],
@@ -133,7 +133,9 @@ class GPT(BaseModel):
"""Возвращает максимальную длину последовательности."""
return self._max_seq_len
- def forward(self, x: torch.Tensor, attention_mask=None) -> torch.Tensor:
+ def forward(
+ self, x: torch.Tensor, attention_mask=None, use_cache: bool = True, cache: list = None
+ ) -> tuple:
"""
Прямой проход для получения логитов по последовательности токенов.
@@ -157,33 +159,60 @@ class GPT(BaseModel):
f"Длина последовательности {x.size(1)} превышает максимальную {self._max_seq_len}"
)
+ # Вычисление start_pos из кэша (если кэш передан)
+ if cache is not None:
+ seq_len = 1
+ # Безопасно извлекаем key_cache для вычисления start_pos
+ if (
+ isinstance(cache, (list, tuple))
+ and len(cache) > 0
+ and cache[0] is not None
+ and isinstance(cache[0], (list, tuple))
+ and len(cache[0]) > 0
+ and cache[0][0] is not None
+ and isinstance(cache[0][0], (tuple, list))
+ and len(cache[0][0]) > 0
+ ):
+ key_cache, _ = cache[0][0]
+ start_pos = key_cache.size(1)
+ else:
+ start_pos = 0
+ else:
+ # Без кэша работаем как раньше
+ start_pos = 0
+ seq_len = x.size(1)
+
# Эмбеддинги токенов и позиций
tok_out = self._token_embeddings(x) # [batch, seq_len, emb_size]
- pos_out = self._position_embeddings(x.size(1)) # [seq_len, emb_size]
+ pos_out = self._position_embeddings(
+ seq_len, start_pos=start_pos
+ ) # [seq_len, emb_size]
# Комбинирование
out = self._dropout(
tok_out + pos_out.unsqueeze(0)
) # [batch, seq_len, emb_size]
- # Стек декодеров
- for decoder in self._decoders:
- out = decoder(out)
+ # Стек декодеров с передачей кэша
+ new_cache = []
+ for i, decoder in enumerate(self._decoders):
+ decoder_cache = cache[i] if cache is not None else None
+ decoder_result = decoder(out, use_cache=use_cache, cache=decoder_cache)
- return self._linear(out) # [batch, seq_len, vocab_size]
+ # Извлекаем результат из кортежа
+ if use_cache:
+ out, decoder_new_cache = decoder_result
+ new_cache.append(decoder_new_cache)
+ else:
+ out = decoder_result[0]
- # def forward(self, input_ids, attention_mask=None):
- # B, T = input_ids.size()
- # pos = torch.arange(0, T, device=input_ids.device).unsqueeze(0)
- #
- # x = self.token_emb(input_ids) + self.pos_emb(pos)
- #
- # for block in self.blocks:
- # x = block(x, attention_mask)
- #
- # x = self.ln_f(x)
- # logits = self.head(x)
- # return logits
+ logits = self._linear(out) # [batch, seq_len, vocab_size]
+
+ # Возвращаем результат с учетом use_cache
+ if use_cache:
+ return (logits, new_cache)
+ else:
+ return (logits, None)
def generate(
self,
@@ -193,8 +222,9 @@ class GPT(BaseModel):
temperature: float = 1.0,
top_k: int = None,
top_p: float = None,
- attention_mask: torch.Tensor = None, # Добавляем для совместимости с HF
- **kwargs, # Игнорируем остальные параметры
+ use_cache: bool = True,
+ attention_mask: torch.Tensor = None,
+ **kwargs
) -> torch.Tensor:
"""
Авторегрессивная генерация текста с поддержкой жадного поиска (greedy), вероятностного сэмплирования с температурой,
@@ -244,12 +274,24 @@ class GPT(BaseModel):
- Holtzman et al., "The Curious Case of Neural Text Degeneration" (nucleus sampling): https://arxiv.org/abs/1904.09751
- Оригинальный GPT-2: https://cdn.openai.com/better-language-models/language-models.pdf
"""
+ cache = None
+
for _ in range(max_new_tokens):
# 1. Обрезаем вход, если последовательность слишком длинная
- x_cond = x[:, -self._max_seq_len :]
+ if use_cache and cache is not None:
+ # Используем кэш - передаем только последний токен
+ x_input = x[:, -1:] # [batch_size, 1]
+ else:
+ # Первая итерация или кэш отключен - передаем всю последовательность
+ x_input = x
# 2. Передаем последовательность в метод forward класса GPT и полуаем логиты.
- logits = self.forward(x_cond)
+ # Прямой проход с кэшем
+ logits, new_cache = self.forward(x_input, use_cache=use_cache, cache=cache)
+
+ # Обновляем кэш для следующей итерации
+ if use_cache:
+ cache = new_cache
# 3. Берем логиты для последнего токена
last_logits = logits[:, -1, :] # [batch_size, vocab_size]
diff --git a/llm/src/llm/models/gpt/gpt2.py b/llm/src/llm/models/gpt/gpt2.py
index 50c0f9a..2b173e3 100644
--- a/llm/src/llm/models/gpt/gpt2.py
+++ b/llm/src/llm/models/gpt/gpt2.py
@@ -214,6 +214,8 @@ class GPT2(BaseModel):
top_k: int = None,
top_p: float = None,
use_cache: bool = True,
+ attention_mask: torch.Tensor = None,
+ **kwargs
) -> torch.Tensor:
"""
Авторегрессивная генерация токенов с поддержкой greedy, temperature, top-k, top-p sampling и KV-кэша.
diff --git a/llm/src/llm/models/llama/llama.py b/llm/src/llm/models/llama/llama.py
index 1b98f45..dc7c53d 100644
--- a/llm/src/llm/models/llama/llama.py
+++ b/llm/src/llm/models/llama/llama.py
@@ -176,6 +176,8 @@ class Llama(BaseModel):
top_k: int = None,
top_p: float = None,
use_cache: bool = True,
+ attention_mask: torch.Tensor = None,
+ **kwargs
) -> torch.Tensor:
"""
Авторегрессивная генерация последовательностей на основе LLaMA (greedy, temperature, top-k, top-p/nucleus, поддержка KV-кэша).
diff --git a/llm/src/llm/models/mistral/mistral.py b/llm/src/llm/models/mistral/mistral.py
index 1e56eea..3547292 100644
--- a/llm/src/llm/models/mistral/mistral.py
+++ b/llm/src/llm/models/mistral/mistral.py
@@ -140,14 +140,17 @@ class Mistral(BaseModel):
else:
return (logits, None)
- def generate(self,
- x: torch.Tensor,
- max_new_tokens: int,
+ def generate(
+ self,
+ x: torch.Tensor,
+ max_new_tokens: int,
do_sample: bool,
temperature: float = 1.0,
top_k: int = None,
top_p: float = None,
- use_cache: bool = True
+ use_cache: bool = True,
+ attention_mask: torch.Tensor = None,
+ **kwargs
) -> torch.Tensor:
"""
Авторегрессивная генерация токенов с поддержкой greedy, temperature, top-k/top-p sampling
diff --git a/llm/src/llm/models/mixtral/mixtral.py b/llm/src/llm/models/mixtral/mixtral.py
index a5c6133..1d8e1c9 100644
--- a/llm/src/llm/models/mixtral/mixtral.py
+++ b/llm/src/llm/models/mixtral/mixtral.py
@@ -222,14 +222,17 @@ class Mixtral(BaseModel):
else:
return (logits, None)
- def generate(self,
- x: torch.Tensor,
- max_new_tokens: int,
+ def generate(
+ self,
+ x: torch.Tensor,
+ max_new_tokens: int,
do_sample: bool,
temperature: float = 1.0,
top_k: int = None,
top_p: float = None,
- use_cache: bool = True
+ use_cache: bool = True,
+ attention_mask: torch.Tensor = None,
+ **kwargs
) -> torch.Tensor:
"""
Авторегрессивная генерация токенов с поддержкой greedy, temperature, top-k/top-p sampling
diff --git a/llm/tests/core/test_decoder.py b/llm/tests/core/test_gpt_decoder.py
similarity index 89%
rename from llm/tests/core/test_decoder.py
rename to llm/tests/core/test_gpt_decoder.py
index 8eae46f..d1632c1 100644
--- a/llm/tests/core/test_decoder.py
+++ b/llm/tests/core/test_gpt_decoder.py
@@ -4,17 +4,17 @@ Tests for decoder block.
import pytest
import torch
-from llm.core.decoder import Decoder
+from llm.core.gpt_decoder import GptDecoder
-class TestDecoder:
+class TestGptDecoder:
"""Test cases for Decoder."""
def test_initialization(self, embed_dim, num_heads):
"""Test that Decoder can be initialized."""
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
@@ -32,7 +32,7 @@ class TestDecoder:
"""Test forward pass of Decoder."""
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
@@ -40,7 +40,7 @@ class TestDecoder:
)
# Forward pass
- output = decoder(random_embeddings)
+ output, _ = decoder(random_embeddings)
# Check output shape
assert output.shape == random_embeddings.shape
@@ -50,7 +50,7 @@ class TestDecoder:
"""Test forward pass with causal mask."""
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
@@ -62,7 +62,7 @@ class TestDecoder:
mask = torch.tril(torch.ones(seq_len, seq_len))
# Forward pass with causal mask
- output = decoder(random_embeddings, mask=mask)
+ output, _ = decoder(random_embeddings, attention_mask=mask)
# Check output shape
assert output.shape == random_embeddings.shape
@@ -71,14 +71,14 @@ class TestDecoder:
"""Test that residual connections are properly applied."""
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
max_seq_len=max_seq_len,
)
- output = decoder(random_embeddings)
+ output, _ = decoder(random_embeddings)
# With residual connections and layer norm, the output shouldn't be
# too different from input (in terms of scale/distribution)
@@ -92,14 +92,14 @@ class TestDecoder:
"""Test that layer normalization is applied."""
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
max_seq_len=max_seq_len,
)
- output = decoder(random_embeddings)
+ output, _ = decoder(random_embeddings)
# Check that output has reasonable statistics (due to layer norm)
# Mean should be close to 0, std close to 1 for each sequence position
@@ -114,7 +114,7 @@ class TestDecoder:
"""Test that gradients flow through Decoder."""
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
@@ -122,7 +122,7 @@ class TestDecoder:
)
# Forward pass
- output = decoder(random_embeddings)
+ output, _ = decoder(random_embeddings)
# Create a dummy loss and backward pass
loss = output.sum()
@@ -139,7 +139,7 @@ class TestDecoder:
"""Test that Decoder works on correct device."""
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
@@ -148,7 +148,7 @@ class TestDecoder:
inputs = random_embeddings.to(device)
# Forward pass
- output = decoder(inputs)
+ output, _ = decoder(inputs)
# Check device consistency
assert output.device == device
@@ -165,7 +165,7 @@ class TestDecoder:
for embed_dim, num_heads in test_cases:
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
@@ -174,7 +174,7 @@ class TestDecoder:
batch_size, seq_len = 2, 16
inputs = torch.randn(batch_size, seq_len, embed_dim)
- output = decoder(inputs)
+ output, _ = decoder(inputs)
assert output.shape == inputs.shape
@@ -183,7 +183,7 @@ class TestDecoder:
"""Test Decoder with different input shapes."""
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
@@ -191,7 +191,7 @@ class TestDecoder:
)
inputs = torch.randn(batch_size, seq_len, embed_dim)
- output = decoder(inputs)
+ output, _ = decoder(inputs)
assert output.shape == (batch_size, seq_len, embed_dim)
@@ -199,7 +199,7 @@ class TestDecoder:
"""Test that Decoder behaves differently in train vs eval mode."""
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
@@ -209,11 +209,11 @@ class TestDecoder:
# Training mode
decoder.train()
- output_train = decoder(random_embeddings)
+ output_train, _ = decoder(random_embeddings)
# Evaluation mode
decoder.eval()
- output_eval = decoder(random_embeddings)
+ output_eval, _ = decoder(random_embeddings)
# Outputs should be different due to dropout
assert not torch.allclose(output_train, output_eval)
@@ -222,7 +222,7 @@ class TestDecoder:
"""Test that parameters are properly initialized."""
head_size = embed_dim // num_heads
max_seq_len = 1024
- decoder = Decoder(
+ decoder = GptDecoder(
num_heads=num_heads,
emb_size=embed_dim,
head_size=head_size,
diff --git a/llm/tests/models/test_gpt.py b/llm/tests/models/test_gpt.py
index 61f90d2..48a0101 100644
--- a/llm/tests/models/test_gpt.py
+++ b/llm/tests/models/test_gpt.py
@@ -30,7 +30,7 @@ class TestGPT:
model = GPT(gpt_config)
# Forward pass
- logits = model(random_inputs)
+ logits, _ = model(random_inputs)
# Check output shape
batch_size, seq_len = random_inputs.shape
@@ -45,7 +45,7 @@ class TestGPT:
model = GPT(gpt_config)
# Forward pass with mask
- logits = model(random_inputs, attention_mask=attention_mask)
+ logits, _ = model(random_inputs, attention_mask=attention_mask)
# Check output shape
batch_size, seq_len = random_inputs.shape
@@ -132,7 +132,7 @@ class TestGPT:
model = GPT(gpt_config)
# Forward pass
- logits = model(random_inputs)
+ logits, _ = model(random_inputs)
# Create a dummy loss and backward pass
targets = torch.randint(0, gpt_config["vocab_size"], random_inputs.shape)
@@ -157,7 +157,7 @@ class TestGPT:
inputs = random_inputs.to(device)
# Forward pass
- logits = model(inputs)
+ logits, _ = model(inputs)
# Check device consistency
assert logits.device == device
@@ -197,7 +197,7 @@ class TestGPT:
batch_size, seq_len = 2, 16
inputs = torch.randint(0, config["vocab_size"], (batch_size, seq_len))
- logits = model(inputs)
+ logits, _ = model(inputs)
expected_shape = (batch_size, seq_len, config["vocab_size"])
assert logits.shape == expected_shape
@@ -208,7 +208,7 @@ class TestGPT:
model = GPT(gpt_config)
inputs = torch.randint(0, gpt_config["vocab_size"], (batch_size, seq_len))
- logits = model(inputs)
+ logits, _ = model(inputs)
expected_shape = (batch_size, seq_len, gpt_config["vocab_size"])
assert logits.shape == expected_shape
@@ -219,11 +219,11 @@ class TestGPT:
# Training mode
model.train()
- output_train = model(random_inputs)
+ output_train, _ = model(random_inputs)
# Evaluation mode
model.eval()
- output_eval = model(random_inputs)
+ output_eval, _ = model(random_inputs)
# Outputs should be different due to dropout
assert not torch.allclose(output_train, output_eval)
@@ -271,7 +271,7 @@ class TestGPT:
"""Test that GPT output has proper distribution."""
model = GPT(gpt_config)
- logits = model(random_inputs)
+ logits, _ = model(random_inputs)
# Logits should not have extreme values
assert logits.abs().max() < 100
diff --git a/llm/tests/test_basic.py b/llm/tests/test_basic.py
index 8d18689..0565d22 100644
--- a/llm/tests/test_basic.py
+++ b/llm/tests/test_basic.py
@@ -28,7 +28,7 @@ def test_gpt_model_creation():
input_ids = torch.randint(0, config["vocab_size"], (batch_size, seq_len))
with torch.no_grad():
- logits = model(input_ids)
+ logits, _ = model(input_ids)
assert logits.shape == (batch_size, seq_len, config["vocab_size"])
print("✅ GPT model creation and forward pass test passed")
@@ -222,7 +222,7 @@ def test_gpt_with_tokenizer():
input_ids = torch.tensor([tokens])
with torch.no_grad():
- logits = model(input_ids)
+ logits, _ = model(input_ids)
assert logits.shape == (1, len(tokens), vocab_size)
print("✅ GPT with tokenizer integration test passed")
diff --git a/notebooks/gpt.ipynb b/notebooks/gpt.ipynb
index 36c7dc2..7842580 100644
--- a/notebooks/gpt.ipynb
+++ b/notebooks/gpt.ipynb
@@ -10,9 +10,23 @@
"Она заложила фундамент всех последующих поколений GPT-моделей, показав, что модель, обученная на большом корпусе текстов в режиме **предсказания следующего токена**, способна эффективно адаптироваться к различным задачам обработки естественного языка."
]
},
+ {
+ "attachments": {
+ "image.png": {
+ "image/png": "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"
+ }
+ },
+ "cell_type": "markdown",
+ "id": "52e3985a",
+ "metadata": {},
+ "source": [
+ "# Архитектура\n",
+ ""
+ ]
+ },
{
"cell_type": "code",
- "execution_count": 1,
+ "execution_count": null,
"id": "a4fba924",
"metadata": {},
"outputs": [],
@@ -32,7 +46,7 @@
},
{
"cell_type": "code",
- "execution_count": 2,
+ "execution_count": null,
"id": "1a6f2914",
"metadata": {},
"outputs": [],
@@ -177,13 +191,18 @@
]
},
{
+ "attachments": {
+ "image.png": {
+ "image/png": "iVBORw0KGgoAAAANSUhEUgAABQcAAACiCAYAAAD81HvMAAAAAXNSR0IArs4c6QAAAERlWElmTU0AKgAAAAgAAYdpAAQAAAABAAAAGgAAAAAAA6ABAAMAAAABAAEAAKACAAQAAAABAAAFB6ADAAQAAAABAAAAogAAAACNa65XAABAAElEQVR4Ae3dB1gURxsH8JciVgTsIip2xd4bKNh7TzRqrIm99y622KOfRmMvsfeusWOJ3ZjEqFFRsWIvWFCKfPMO7nnAAUc5bu/uP8+z3N5sm/3N3nH33sysFSEliUCXLl3CkuRAOEiiCyxatMgq0Xca+w5xvcRuhDUgAAEIQMA0BZLk/+rM/YMbBId+mmBrmyzVh09v85smVdRSp0pmfyO5XaozbwKf3xtV/9fRUddQZw7qQ531YimlwvWnrppGfairPlAaCLBAknw4AzURBwdnzZoFChMT6N+/PxkrOBgWcMHEtFBcCEAAAhCAQMwCVmnL8AoG/fy5/NTMTq/ePRmdzCZ5uqI53NPyATPaZ+MHs0jP3j6k528fUHBo0KObjy85p07uMG9grem91HpyqA+11oxllAvXn7rqGfWhrvpAaSCgLWDQD2faB7L0eQQHTfMKQHDQNOsNpYYABCAAAXUKGDo4OG1//6spbdNk46CgOQUEo6vNa4/OyiDhg5c33wyu/bNbdOsZKx/1YSx5HJcFcP2p6zpAfairPlAaCEQWsI2cgecQgAAEIAABCEAAAhAwNYEZ+wfezJG+oHNhl4qpTK3s8S1vIefyvKmztZVNmhkHBt0YVGuGarpPoz7UVR/xvcZMdTtcf+q6/lAf6qoPU31do9yGFbA27O6xdwhAAAIQgAAEIAABCBhWgFukWFpgUFu0iEultDnTF8w5ff+Aq9r5xppHfairPox1HRjruLj+1HX9oT7UVR/Gel3iuOoXQMtB9dcRSggBCEAAAhCAAAQgEI3AzweH/JrFIZe9JbUY1EXhlq2C3ePXfs5LT01p17nSsN90rZMUeaiPcGW11EdS1LmajoHrT13XH+pDXfWh/Vrt1q2bZ5YsWdoEBga6v3r1qqD2MjXMOzo6Xrezs/OZOHFiNzWUR98y/DponWdap9Rtgj4Fuwe+/6Q61xSpU1y3SWbt0350kyiuCA7qW8tYDwIQgAAEIAABCEBAdQLvPr7uZpfeTnXlMkaBiubwcPjzzuFx4thGCw6iPr7WvBrq42tpLGMO19/XelbD9Yf6UFd9KKXx9vZe/OjRox9SpEgR0rRpU9u8efMqi1Tz6OvrW0BM+cS9G7pmypRpoSkECddM2rn4zct3P9gkswkpXiGfbfosjqrxVAry4vHrAi/8X+ebP3Bt19SOqRZqBwnRrVhRwiMEIAABCEAAAhCAgEkJTN7be0rezCUffBl7z6TKbojC8k1Y7JIlT/+Lz5imhth/bPtEfUQUMnZ9RCyN+T/D9Rexjo19/aE+1FUfSmmmTJmyzsbG5puePXtS7969VRkY5LJywLJOnTrWs2bNIjc3twZTp07doJyDGh/XT927jqytv6lQryRVrF9KlYFBduOAZf6Srtb1O3lRJpf0DTZM36dxRXBQjVcWygQBCEAAAhCAAAQgEKtAmhQOmexs7VxiXdGCViji4m7/LvD1WGOcMuojqrox6yNqacw7B9df1Po15vWH+lBXfXBpuMVgzpw5vQYMGOCgxtaCUcXCcxo3bpwtODi47qhRoxZEt44x87nFoGOmtF7ujUo7qLG1YHQ2buXyZgsJCam7csJ26YrgYHRSyIcABCAAAQhAAAIQULXAp5DAyhnsERuMXEm21nZGuWMz6iNyTYQ/N1Z96C6N+ebi+tNdt8a6/lAf6qoPLg13JW7UqFFm3SVTd27Dhg3tnz592pXHSlRbSbkrcaFyeUzStVC5vPbvX3/oymMlYsxBtV1ZKA8EIAABCEAAAhCAgF4CHz69zc9d55C+CrDHh6CAfF9zkm4O9RHV2pj1EbU05p2D6y9q/Rrz+kN9qKs+hg8fPrdUqVKBolQpo5ZM/Tnc0lFMYQEBAa1EaX3UUuIV3tvmZsuT2WRduaVj+qyOYR8Dg1qh5aBariqUAwIQgAAEIAABCEAAAhCAAAQgAAEIJLLA58+fa+TPn98kA4MKRe3ata2CgoI8ledqeAwTrumdnUzaNV/JXFahwZ89ERxUwxWFMkAAAhCAAAQgAAEIQAACEIAABCAAAQMIvHr1qqApjTOoi4DL//r16wK6lhkrL/D9p4KmNM6gLicu/8f3HwsgOKhLB3kQgAAEIAABCEAAAhCAAAQgAAEIQAACELAAAQQHLaCScYoQgAAEIAABCEAAAhCAAAQgAAEIQAACENAlgOCgLhXkQQACEIAABCAAAQhAAAIQgAAEIAABCEDAAgQQHLSASsYpQgACEIAABCAAAQhAAAIQgAAEIAABCEBAl4CtrkzkGV7g1dtPdPHGc8MfCEfQW4Dr5BvP3HqvjxUhAAEIQAACEIAABCAAAQhAAAIQgICpCyA4aKQadLJPTmevvaBCuTMbqQQ4rLZA2tQp6ML1hwgOaqNgHgIQgAAEIAABCEAAAhCAAAQgAAGzF0Bw0IhVHPD+I5UvmsOIJcChtQUOnrmu/RTzEIAABCAAAQhAAAIQgAAEIAABCEDA7AUw5qDZVzFOEAIQgAAEIAABCEAAAhCAAAQgAAEIQAACugUQHNTtglwIQAACEIAABCAAAQhAAAIQgAAEIAABCJi9AIKDZl/FOEEIQAACEIAABCAAAQhAAAIQgAAEIAABCOgWQHBQtwtyIQABCEAAAhCAAAQgAAEIQAACEIAABCBg9gIIDpp9FeMEIQABCEAAAhCAAAQgAAEIQAACEIAABCCgWwDBQd0uyIUABCAAAQhAAAIQgAAEIAABCEAAAhCAgNkLIDho9lWME4QABCAAAQhAAAIQgAAEIAABCEAAAhCAgG4BBAd1uyAXAhCAAAQgAAEIQAACEIAABCAAAQhAAAJmL4DgoNlXMU4QAhCAAAQgAAEIQAACEIAABCAAAQhAAAK6BRAc1O2CXAhAAAIQgAAEIAABCEAAAhCAAAQgAAEImL0AgoNmX8U4QQhAAAIQgAAEIAABCEAAAhCAAAQgAAEI6BZAcFC3C3IhAAEIQAACEIAABCAAAQhAAAIQgAAEIGD2AggOmn0V4wQhAAEIQAACEIAABCAAAQhAAAIQgAAEIKBbAMFB3S7IVZFASHCwikqDokAAAhCAAAQgAAEIQAACEIAABCAAAfMRQHDQfOqSqpfPT/mz2Oqc/rp4VueZzpw0kqaNH6ZzWUIzWzWqorMs40f01XvXj/0fkFv2lPTp00e9t8GKEIAABCAAAQhAAAIQgAAEIAABCEAAAvoJ2Oq3GtYyBYENu0/Q59BQWVT3Ejlo0syFVLV6XfncKX0GnacQFhZGn8M+61yWGJld+wyjlm07R9hVGvu0EZ7jCQQgAAEIQAACEIAABCAAAQhAAAIQgIBxBNBy0DjuBjlqhoyZKVMWZznxAZzSZdA8P35kP9WrUoxK5nWiHh2aEbfIi5yuX71M39avTCd9DspFWzespIbVSsoWiXOmj6PQkBDZgq95nQq0ae0yuT9urbhpzdLIu9I8T5cuPbnkyBVhcnRKr9nPysVz5DG4lSEft3+3NlSxiLN8/PgxULOfhXOmyvxald3owJ5tmnzMQAACEIAABCAAAQhAAAIQgAAEIAABCMRfAMHB+NuZzJa+N65S9/ZNyd2zJq3YuJ9CRJCvX9fWxK0GlXTn1nXiAB23NOT1jh/5nYb17UzftulME6b/SpvXLqdf/zeZPn/+TJf/ukCL582gEeNnUoXKnjRyYNdou/3+I9bdvW19hOn1qxea/WxYtYR69B9JyWyTUadWdSl1GnuaMGMB7dm+gY7s36UUj44d3ifzS5WpSL06f0MP7/tplmEGAhCAAAQgAAEIQAACEIAABCAAAQhAIH4C6FYcPzeT2mrH5rVUonR5Gczjgg/znkZ13IuQ/6P78jz8bt2g1o29qFnL9tRzwCiZt3bFAvqmdSf6vnMv+bzPkLG06Jfp1LnHQPl8/LT5MjBYskx4K0IO1uXOW1Au0/5z5uRR4v1rpwJuxURLQleZNXrSbKroUY0+ffxIZ08do+He02WAsFqtBuR3+yaVKldJrjdwxCSqVKU6cf7+PVvp1Ikjsnza+8U8BCAAAQhAAAIQgAAEIAABCEAAAhCAQNwEEByMm5dJrn3f7xaVFC3ulOTsklPOvnzxXD4eObBbPmq3xrstAnqcz92HlcSt+pSU1dlFzip5QZ+ClEURHrv0Gkwdu/WPkMdPAgM/yDxnlxzyMXmKFJQ3v5sMDHJG8uQpKPTL+In8nIOQnKytrWWgk1sfIkEAAhCAAAQgAAEIQAACEIAABCAAAQgkTCDabsXlypVrmiVLln9dXV19xSG4/6lJTLlz5/bNnDnzPy4uLrMTRmM+W+cvVIQePbynOaF7d27J+fwFC8vHGnUa0e6jf8lgoBIodHR0oi69htDfdwLkdOLSXVq19bBmHzY2Npr5hMzou583b15pDvPflX+oWMlymueYgQAEIAABCEAAAhCAAASiFfAU3+su58mTh78EGOQ7Xfny5VfZ29tPjrYEWAABCEAAAqoWiBIcdHNzWyBKHCaCNmvWrVtXePny5Xl4bDpTmZYuXZpn/fr1RVu1atWaz6NYsWKLVV0DSVC4qjXqyZt9XDz7hzzarq3rZPdcO7vk8nmOXHmJA4gdu/Yj76G9ZKu+SlVq0IF92+nZE3/68P4djRvem5bOnxnn0j72f0h8oxPtSbuFor47XLVkHgUFfaIt61fQi+dPZetBfbfFehCAAAQgAAEIQAACELBEgRIlSnA3oKPie13BJUuW5DbUd7q6deu2ffv27TDxXXKhJTrjnCEAAQiYukCEbsXOzs5XxC8+OY4ePUqenp4pTfHkRLllscVjxunTp1P//v3rpU+ffqM4p29N8XwSo8xuRUqQpwgQfte4quy2myJFSlq4aodm19ZW4THiHgNG0kZx5+El4mYjnURX4D/Pn6IaFQrI9YqWKENzl27UbBN5xsrKKnKWfL584WziSTvVbtCMps1doZ2lc157n/9cOkdFcqSW642dPEd2O9a5ETIhAAEIQAACEIAABCAAARbw/uuvvzp++W4X4XtfYvOMHTuWqlatSl5eXl3y58+/48aNG3sT+xjYHwQgAAEIGE5A80+CA4Nt2rTJPG3atDSGO1zS73nWrFnOw4YNqylaQm45dOhQ86QvgXGOeONxiObAHGSbtWCNvNnH+/dvKUfOPGRjG171g0b9pFnPwcGJLvl+7b67ctMBevTgrmyx55o7PynBOu1988aRnys7XL/zuDKr81F7u7oNWxBPSvrf4vXKrGb/fAMVLmOq1GZ1iWrOEzMQgAAEIAABCEAAAhBILAEPD4+CIlj3WTSaiNJbLLGOob0fbqTBgUjxnXK6yEdwUBsH8xCAAARULiAjRNyVWDQFTy8Cg+lVXt54FW/KlCmOYhyMmnnz5p3n6+vbM147MYONMmVxjtNZcDAwW3bXOG1jyJWzOmc35O6xbwhAAAIQgAAEIAABCJiNwIkTJ1oePx7zj/WGONkU3E0JCQIQgAAETEpABgevXr3a9cqVKyZV8LgWdurUqfbil7MeYjuLDQ7G1QzrQwACEIAABCAAAQhAAAIQ0FeAWw/evn07l77rYz0IQAACEFCHgHWmTJmmDxky5GtfUnWUK9FLwf+oKlWqFFSrVq1Wib5z7BACEIAABCAAAQhAAAIQgAAEIAABCEAAAiYoYC2S6FFc18kEyx7nIk+aNMnu8uXLo+O8ITaAAAQgAAEIQAACEIAABCAAAQhAAAIQgIAZCliLMSFSmOF5RXtKKVOmTB7tQiyAAAQgAAEIQAACEIAABCAAAQhAAAIQgIAFCVj7+fnl4S63lpC+jIGRxxLOFecIAQiYh8CbD2HmcSI4i3gLfP4c702xIQQgAAEIQAACEIAABCAAgVgF5A1JYl0LK0AAAhCIp8CSQx/pb7/geG5t2ZulTmlHDUtbU+WCySwbwsLP/nNoGC2a6EthYQgUR3cp2CW3oc7D8NtfdD7IhwAEIAABCEAAAhCAQEwCCA7GpINlEIBAggUcUltTfQ83ypM9fYL3ZWk7OHruujjlAEs7bZyvDgErK6J6Hb10LEFWSFAIHd10ChAQgAAEIAABCEAAAhCAQDwFrOO5HTaDAAQgoLfAuw+f9F4XK0IAAhCIi0CwCA4iQQACEIAABCAAAQhAAALxF0BwMP522BICEIAABCAAAQhAAAIQgAAEIAABCEAAAiYtgOCgSVcfCg8BCEAAAhCAAAQgAAEIQAACEIAABCAAgfgLIDgYfztsCQEIQAACEIAABCAAAQhAAAIQgAAEIAABkxZAcNCkqw+FhwAEIAABCEAAAhCAAAQgAAEIQAACEIBA/AUQHIy/HbaEAAQgAAEIQAACEIAABCAAAQhAAAIQgIBJCyA4aNLVh8JDAAIQgAAEIAABCEAAAhCAAAQgAAEIQCD+AggOxt8OW0IAAhCAgBkIBIeEmMFZ4BQgAAEIQAACEIAABCAAAQjET8A2fpthq8QSmLP2RGLtCvuBAASMKPD582cKE5ONLd5WuRrSu9agly9fG7FG9D/0+ycnKRnqTX+weKwZEhpCtjZ4beiis0pbRle20fOcHB3o5b3DRi8HCgABCEAAAhCAAAQgYHgBfFI3vHG0R5jatVy0y7AAAhCIKjCwx/e0a+s6+m3zQarg7qVZ4c/zp6lVQw+q27AF/W/xek2+vjNvA95Q6fzp6fz1Z+Tg4KTvZhHW27BqMZ08dpDmLdscId9Sn6RMkZxevHhB6dKlUzWBnZ2dqsuXFIVr3rYRXbh0PsqhVi5YQ54e1aLkxzXj+s3/qFYTL7p7xT+um1rM+mFhYao619evX5NboQKqKhMKAwEIQAACEIAABCBgOAEEBw1niz1DAAKJLMCt8zjt3701QnDw0L4dMv9zWPhy+SQ+f1T2BT0+p6CmbV6+fKn64KCavIxZlh87dKP233WMUIQMGTJGeI4nliPw6tUryzlZnCkEIAABCEAAAhCAAGHMQVwEEICASQmUKF2etm9eTaFfxonjgOG2jb8R5yvp9asXNHpwd6pYxJlaNapCm9YsVRbRwb3b6ftm1alkXica0rsDvXsboFnGM7y/yd6DiVsphgQH07Onj6lfl+/kvjq2rENXL/8l1+f15s4YT7UquxHn//3nuQj7wRMImJKAk6MTZXfJEWFKmSKlPIVN2zdQnabVyaN2Bfr5l+nEXYQ5PXv+lHoO7EqlPIpS2x9b0ZVr/8p8/rP3wG6q/00tatSyLm3btUWTjxkIQAACEIAABCAAAQhAQH0CCA6qr05QIghAIAaBshWqUAoRtDh/9qRc69+/L9LHj4GiJeHX7o9zZ0ygm9ev0C9LN9F37brSSBHAePr4Eb16+Zx6dmpBrTt2l91/r135O0LgkLv2TR03hPbt2ESDRv0kxw/s1q4JvXnzimbOX0VFipWmJjXLUMCb17Rv5yZatmCW2H8Xqii6Xm7dsDKGUmMRBNQtwIG9HXu3aaYz50/JAvucOEKDRvajVi1ai6D5NNoouvXPWzSH+LXSuWd7eiO65M+ZNo+KuhWjei1qUoAItj989IC69/+RypYqT+1bd6KN2+Le1V/dWigdBCAAAQhAAAIQgAAEzEsA3YrNqz5xNhAwewErKytq1Lw1HdizjSpU9pQtAZu0aEvJkiXTnHvNuo2p3Q+9KGu27JQyVWqZf9v3OuXIlUfOv3j2hKpWq0MLV+2gTyKwqKRZU8bQzi1racehC5TVOTv9+8+fdPmvC3T0vC9ly+4qg4Cb1y2n0yJgcuzwPvq27Q/UsWs/uTlaDiqKeDRFAZ+TR+maCKgrqVJ5d6pQthKtWr+SWorXW4c2neWiAb0G069Lf6FqVWrQ3//+RX8cOEcu4nVWuYIHbRCBwz/OnKC3795Sofxu5D18gtzm3fu3NGbSSGXXeIQABCAAAQhAAAIQgAAEVCaA4KDKKgTFgQAEYheoWa8J/dimIY0YN4N2bF5DU/63lC6e+0OzoZW1NfXq/A1dv3qZsmR10eQ7Z8tBw8ZOo/Ej+srJs0Y9GjF+pmb5upUL5Ty3DOR0/+5t+ehVNq98VP68fPGMDopxDifPXqJkUakyFenPC6c1zzEDAVMS6Pljb+r5Y58oRb7t50uHfA7QBhE0V1Lq1Gno3oO78mnlWhFvrPXi5Qtxc5NzVK5MBWV1KlakhGYeMxCAAAQgAAEIQAACEICA+gTQrVh9dYISQQACsQiULB0eeFi+aDa9ffuGylb00GwRGhpKg0V3R+5+zHcfPvbnHUqdxl4u57EIq9dpRP/ee0+/bTkkxxvkcQOVtFW0gmrY7DsaPagb8X7s7R3kolOXH9LfdwLktHX/WardoBkVcCtKT/wfKpvSnds3NfOYgYC5CPDdu7t37kX/Xbwtp7NHLtGG5VvEayOtPMWLx//RLNu18XeqW7M+uebIRf6iG7+S7t7zU2bxCAEIQAACEIAABCAAAQioUADBQRVWCooEAQjELGBja0vclXj6hOFUt2EL0aXYTrNBSEiwJmDIwT2+Wcl70c2RbyDy8ME9MWZgWRnU4y7JFdy96K0YM01J2XO40uDRk+nK5Uu0cfUSKlyspFy0Y9NqshXHPPuHDzWrXZ5ePH9Gtes3o+0bV9HdO75ifMOrdPj3ncpu8AgBsxFwF4H3/Yf30lPRFf/Dh/c0euJwWrTiVypSqKg8x607N5OtjS2dPvsHNfy2Dr0Q43pWq1qD/hBjgp4WrXlfvnpJm3dsNBsPnAgEIAABSxDYsGEDnThxwhJOFecIAQhAAAJfBNCtGJcCBCBgMgLWorswjznIqVb9prRGBCn4UUnWVtaUPHkK6tZnKPUVd0/lFoOlylakKmJ8Qb4RySXfV1TFqzZVL5+f0mfIRPYOjjRl9tc7GYudy27IA0dOomkThskWgrMXrqV+XVvT1PFD5WEGjphI+Qq4kaO4u+umNcuoZsWCMp+7KHP5kCBgTgJdOnSni5cuUJU6FeVpFRddhBeI7vTpnNLRLzMWUC/RynbSl9a3Q/oNp/x5C8hAfJVKValVxxZym1ri9YcEAQhAAAKmIzB37lxyd3cnD4+vPTNMp/QoKQQgAAEIxEcAwcH4qGEbCEDAKAJ8x2Al8R2CbzwOUZ5S70FjNPNd+wyjVuIuwsHBwZQhY2Z5Z9UXz5/K5f9bvJ6GiPHSgoOCyDV3Ps022vvq2nso8cSpXuNvyUt0lbwvukZmzupM3M2SU8bMWWnXkT/p3t1blCmzs6brslyIPxAwIYEtq6Nv9ZpWdB9eu3SjvANxkHjN5HLNrQnQNxQ3/qnuWZPuixa5WcTrwSFteDd8DpJzAPHBw/siWJ+cMopAPBIEIAABCJiOwMmTJ02nsCgpBCAAAQgkigCCg4nCiJ1AAAJqE3BwTKcpErc25CChkrK55FRm9XrkOx7nL1g4yrrcvTlXngJR8pEBAXMS4NcP35FYV0qVMhUVyBfeejby8ui2ibwenkMAAhCAAAQgAAEIQAACxhVAHzjj+uPoEIAABCAAAQhAAAIQgAAEIAABCEAAAhAwmgCCg0ajx4EhAAEIQAACEIAABCAAAQhAAAIQgAAEIGBcAQQHjeuPo0MAAhYmEBrydZxENZx6cLC6yqMGE5Qh6QVCQtV5HeL1kfTXgiUcMTQk1BJO02TOEfVhMlVllgXF9aeuakV9qKs+TLk0r16/ojcBb+J8CvHZJs4HiWYDBAejgUE2BCCQdAK3bl6j/FlsqVWjKlEO2vm7enLZ9auXoyzTzljwv8k0uFd7mXVw73a653dLzs8Vd1Id2qeT9qo65/kOxhtWLY6w7MrlS/LYr14+j5CfkCdlCmak/678k5BdJHjby1d8qXOvyZQpT21Kk7WKfOTnnJ8UKXPmzJQnTx76+PGj5nD//fefvNFFQECAJg8zCRdg45yFs+qc3r17m/ADJNIeilUsRNeuX0mkvSVsN8Z+fXDp8RpJWB3GtnWjKo2pcJYimql+5QY0b8Z8CkmiH0sqFqxE169cj62YFrMc9WExVa3KE8X1p65qQX0Ytz4aN25MBw4ciFKICxcuEC9Dil3g5Onj1PbHVlSishsVq1iQmrdtRIePHYx1Qw4K8roVa5Qhvgngxq3rKPBjYKzbJeYKCA4mpib2BQH9BDzFat76rar+tcZNXkQ8JSSFhYXJzf88d4qe+D/U7IqDcieORv0HpVlBa4b38fnzZ5kzZ8Y44sAep6bffk9d+wyR87H9UcoReb3o8iOvZ8znPicv6lUP67ccojJV21G+giXp0qW/6dOnT/IxX6EyMn/95ti99T1WTB63b9+mGTNmaFYxBWNNYRN5JjFeQ9EVKYzCX1vzf15EJ/efjTClTp0mus3MLl/fazYxXh/aePoeV3sbZR6vESnhKf56y7lE/BMaGkp9h/elE/8ep91/7KYGzerT+hXraczAsYl4FLPclac4K+/EPjPUR7xFPcWW3vHe2vQ29BRF9k7sYuP6i7eop9jSO95bR7Mh6iMamNizPcUq3rGvFvsauj6TFyhQgH766afYNza/NTzFKXnre1oczGvzQ0vyqFiFrl24RacPX6SypcpTpx7tYm1F+N+Na3Th0nn68/hl+hD4gQaPHkAB8Wh5qG9Zda2H4KAuFeTpLXD+/HnavXu33utjRSngI/5yEzf+1u4tJpNP3iI42LH7OL2CUzGdbInS5enQ/p2aVY4e3EOcp6Q/z5+mbu2aKE+Jg4m9f/hW85xnZk8dS9zKcPKYQXTq+GE6ffIo7d+9NcI6q5b+QhfOnoyQp8+Tf8QbduvGnlSxiDMN6d2BXr96odls4ZwpVK9KMaolfiWaPHYQ8YcbTjf+u0LtWtSU+XOmj6P3Bmqt5elemlas2U1WactEWw/cIqp91zF0+PBhGjZ8JF25coWWLFlC2bJlo2HDhsn89t28Y21BqM+xNDDRzHTu3JlGjx5Nt26Ft/CMvNrOnTupcOHClDZtWmrSpAk9ePBArrJ8+XIaM2YM8fZdu3alhQsX0sCBA6ldu3aUPXt2Ob9582biDzE87dmzJ/KuVfd87PAuxK+hmOouoYXOnCkLZXfJEWHiuxD73r5JLTs0J7dy+ahRy7ryQwkf68q1f6lbvx9o4bL5VKdpdfqxd0c6cOR3WYwde7dRk9YNZGCZM4Z7D6b9h/bJZfMWz6EajaqSV/3KNEG8Fvl1wBPve+O29XLZnbu36frN/+i7Tt/I9X7+ZTq9f/9Obm+oP/pcs4n1+tA+B32Oq72+9jxeI1LDR/w1yP9L+7RpKF2GdOKO867UfWB3mvrLFNqxcQc9uBf+A9XlS5epXeP25FGkCg3vPVy837+WBeIuZ9zKkPO5hcvaZWvD80M/04JZC6layepy2dSx00Tr6E9y2c3/blKnFp2JWyj+Mn2e+D/wXubzn+iOs239Npo77Rca3X80eQ8ep1nfyDM+4vioDwuoDyNfZ9EdHtcf3g/ktRHd+ybenw3z/szoDx8+pHXr1kl//oy+dOlSmjhxomxNOGTIEHr9Ovx/5MuXL2ns2LEyf8CAAXTz5k25Df85efIk9erVSy7jbZ89eyaX7d27V+5v6tSpERoOaDY07oyPOLze//eePH0sS1usSHFKlTIVOWdxpn49BlLf7gPEZ4LwVoAHj+6Xn4f5szd/vvZ/4k+83cjxQ+W2XfuJ7zh9O8v59l3bkP/jR/Jz9DLRw40/k3PrwuOnjlHvwd2plEdR+aj0xvpXfAfmz++c37HH93Tu4lm5n59mTpDrcUMa/lzeY0AXWrB0nlym/QfBQW0NzMdZwM/Pj/7+++84b4cNqOMXg7HiMUxM3l+em9wDBzY4rVizSwY4EhIkrNOgOe3ZtkFj8PuuLcR5Snr39g1dvfyX8lT8mvI6wnNe0LhFa8qeMzd927YzFRJvzP4P79OdWzc02/DMH8cO0d07UbvQ+hzaS4t+maaZtqxbodnu+bMn1KJuRbHPEjRnyQbxRfEl9RBBFU43RCBlxaI5NGDERJo0cyHt3LKWjuzfJZd1b9+UkiVLRv2HjadTJw7LPEP9Wb6ALydxMUUTaJr960Ya5z2G3N3d5XpXr16l48ePy3n+w/njxo2j2b9u0uRFN8PHcs3hHO2xottOyW/QoAF9++231LdvXyVL88jl4q4LtWrVooMHD1KIGKexVatWxL9kPnr0iCZMmCA/pHAeP//5558pV65cMmjI8z179pTnwefTp08fzX7VPOOtvI7Wxhzgje85HBNBcg7qKdNN8ZrgLgttxa+bKVKkoJUL1lAVd09q17U1vRAtdt9/eE/7RHB+/ZY11KFNJ3J0dCLeB6fDPofo0t8X6d9r/8h9rN20mlxz5iL+xXPpb4tpSL/hNGXcTNomgvKHRMtfbr34979/iWtlNDWs25gypMtAP/TqQMlsbWlQn2H0x5kT8T2tOG2XlK8P7YLFdlztdbXn8RrRaCTJ/8vSFUvLA97xvU0vnr2gVnW/o4JFCtKsJT/Tm1dvqE+H8PeqLWu3iNfLSpo4eyL92OdHmjTiJxlQ3LZuKy2bt0zk/SC32b9zPy3+X3ir+l7te5NtMlvqM6w3nTlxRnNiMR3n6eNntODnBfTk8VOq16SuZhsVzKA+LLA+VHDdKUXA9WeB1x/en5XLP9ZHg70+3r17R5cvX5YFeP78Oa1cuZJcXFyof//+dP36ddq+fbv8nM6NDd6+fSsbAPCP9PxDJ28bHBxM06dPp+rVqxMHATmIuH79+gj742AhL1dh0tvVJVt2yuGSUwzf1EF+7uUf1j8FfaIBvQYT/1DPn7/5M3CVylVpjfg+yWNu9xrYlRzSOlLrb9oS9+rpJwKJPbuEf3/p33OQzOPP0fx5u3e3fmRrm4y+F92WU6dKLT5vT6edYjgtDjhy6j24GzmJz+wrFqymPLny0ohxQ2S9tGrWWq63Zecm8dl+LfmIz/TNGrWIQm0bJQcZEIBAUgj4iIP4iclVTJw4qsPTODF5i8mkEgc2OCDFiYOEnPzu+YvgUVZy8+wgn+vzx6tWffEmN4Seil9IkqdMSRysGzF+pszTZ3teJ1eeApQqdWrKk78QOYkgRFzSbfGGzYEoJQW8eaXM0m7R6ilLVhcaNXGWHBsvQ8ZMVLtyYXomfumxsbWhecu3UPFS5eix/wNyFi20/rv6j+i6W5jui1ZSm/edkmXJ6ZqXmtQso9lnYs9wKyUO2PndeyR3zXUiJ1E/HMTdte8YXfppToyH/f7776lkSe7uOzzG9fhY2inysbSXRTc/c+ZM2dpv27ZtVLBgQc1qq1evpgoVKtCsWbNkHnc/LlSoEN2/f18+z5QpE/GvjNbW1uTj40OVK1eWwUBeOGrUKOrRo4cMJubLl4+WLVsmP5BwgFbNSWk96Hc3vO5WiCChUncjB/2Y4KKv2bhKfPBIq9lP9x9600PRhZ9/rTy867j84FFWXL/LVi2hkyJYlzWzs1x3+a8i8JcjF6USH0BmzZsu83gslSJuRenPvy6StZU1pRevs/x5C9AtEXBfPHcZlShaih6L/bo4u9C1G1epuldNud04ETz/pklL4paD9x7cpR3r91I6p3Ry//VE61pDp6R8fWifS2zH1V438jxeI1LER/z1E5OrmDgZ5P9l8uTJ5c4/fQwSP1LtpcxZM9PwicPk+336jBmogWj19/zpc9qydqtoKd+BqtYIHyP3w/sP9PiRP21avVl8WG9L33X8Tu6nx6AeNF+0MGz0TSN6cPcBrd+3TvwfcKIcrjmoRc1v5DoxHYdX4JaNC9b8Kt/r5Abq+OMjiuEnJlcxcUJ9hDsY66+POLCfmFzFxMkg9RG+a1X89RGl8BOTq5g4GeR88X4QjqvHXx+xjp+YXMXECfUR7mCsvz7iwH5ichUTJ4PUB++4WLFi1KFDB56lGzdu0L179+QjjyG+ceNGypIlC5UuXVp+Xr948aL4blFS9u6pUqWKbGWYI4f4riTWVZKTkxNNmzZNbf/vlOL5iBk/MbmKiVO0rrY2trRt7S7asmMTbRONXJavXiI3+KFdFxo5eKzMK1m8NI0ZOl7mjxJ51Rp40EvRG61QATdKkTwF8fLXb8JbYpYoVlI2MuGV+XN05QoesufOmfOnaPQQb/n5vYZnLfK7e0fur3/PwVStSnX52aVwwSK0eMUC+UN+7lx5aGj/ETT2p1FyvYmjxNjzGTPLee0/cQoO8gDyPAaOrnT69Gn5ZS7ysuHDh8umi1zZSOYhwMGT5s2by5Ph8co+fPhA586dk8/5y3zu3LnN40S/nkXY11mDzylvNrKLocGPZsADKEFC73TFqXA5F72OlCFjFqrg7kWHRau7VOKXE+5SnEUEGKJL2oG86NZR8rlLcs+O4dfti+dP6eypYzRj4gh5nMNnw1sWduraj1qJN28l8biFTWuWlU/v+d2Wgb8CWSMGmV69fEH2IuiycM5UatXQg1KnsZfre9aoR+dFkMU1T35NkLKACKjENbmXqxjXTaKsrwTuOEDGgTW+2QGnwMBA+eve/v3hvzYdOnSIihYtKv5pB+h9/Tk62It/YG81x1SOlS6doyYvuhn+xZEDf926daNdu3ZpVuOuxhUrfj3vnDlzymVK94MaNWpE+PCQN29ezbaOjo7yAwtn2NnZyXxuPh9TcDB1ZnfN9mqZUYKEiueAVmOoLnnGu3iL5iyjMiXDr2VlJxwwzCeuT+2xBwuJoDr/IszBQc7nwCCniuUqiV8jb8sWgPy8fetOdPDIfvlrZA2vWvJDCAcQuVtxszaNNPusVrUGry5TyWKl5OO5C2cot2tuGRjkjILimAlNHwI/6X3NRj6WYmyI10fkY2k/V46rnRd53pivEX/RYo27uhsoJcb/Vfn/0u+yP1HESzteRX71MvzHoHwF89LZk2fF+LdPqEjWiO/ZvM4d3zvUqWdHzTFatg8f2oLzu/b7+v/DJUc2uY8LZy6K/wOuMjDIG+V3K6DZ9r7f/WiPwytVrFIxwnudZkP9ZxLDWd+jyfrQd+XY1kN9xCYU6/JErY9Yj6Z7hSS//vB+oLsivuQmeX3EWJo4LMT7QRywdK+a6O8HPCSRklKLRhncG4V783DinkHaibscpxSNPjiIyOMWcuwgY8aMMoCorMeBRP7RPw4pKa/n6IoVxZVjIylFd+KunXrI6Zn4zrl+81qaMXcqVRCfpe/e96PSIvinpGyi4QmnF+L7ZGwp25fvxPzjhfbnd34e+jl8OCsOKlapW0n2AuIWjNqpY9sfaOqsn8Rn/KzUtGH4d2Lt5Twfp+DgqVOnZKCPN+QLYvHixVSvXj1+Shky6G6hw93AlJsEyBXN94+3ODWezD7Z2NjIpsJ8otxqh8cCa9u2rTxvJehgZghWBjofXW9q48SxvMUUFhZwwUCH1X+3XvW7yhZnkVuJRd4D31CBv+gqiVuvuebMKrd9ZhPxy5WyTnSPdRu2oL07NsogW73GEf+58DbBwUGaTf0f3tPMxzaTJ39Bmjl/lVxt5k8jqUx5d6pava7e/4jSOjhS0RJlaPW2I3IfIaJ5/E0xnmCuPPlo+oTh5CfGbjsqBp7NJt6I+4qm3pyyi8CKn2iNGCoC6jaiCyW3KoxrOnnuNFUuGDEgGdM+dH2Z55ad3DKN70785MkTWrNmjdzFpk2byNfXl/hHHE4chOMxRRwd09LTW+EBQ7kgmj/chVwJAiureHqUlsdq+8MYJSvGx969e9OiRYtkl2BlRQ5Q/vXX1+7jXEZORYoUod9/Dx/3TlmXH22FrXbisfTikt4/OSnGBUkRl00irKvcdOLonoUR8uP6JLq645aDS6eEG8R1nzGtnzd3PnokWuly8JTf1/n/9TXRNXjEoNEi73OETTNmyESF8rvRouW/krsYZJlbGXI34SDxemzx5XW6ZOUC0SrwDp06eJ74AwyPZ6KdlHrhsQ9vi2A7d6XgX1i5lWFCU6qUyUmf98zojA3x+tA+p5iOq2uZ9rbGeo1kzZKJHt3Yq12URJn/cr5xe5GGD78R+fjy/6Vr0ay6/pdGXjfW57+LbsCcXHJmp7TiR48iJYrQym0rZF6wuIuxrxg3kIN8BQsXFK3bw8dJ4oWH9hyirC6ilXwxNxnokxuIP3637lJlr8rEQUK/W37i/4B4nYlW5k/8HyurxHgczUoJm4mrs75H02Uu60PsQNcyfferWQ/1oaHQZ0aXeaLWhz6F0LFOkl9/eD/QUQtfs5K8PsShdV2bX0uk5xzeD/SECl9Nl3mivx/oCuSlSRN+o7sdO3bIYCAXx8/PTwYBuSHZb7/9JuNH3LOH1+FGCQlIhrqeoyuSXq7bd2+hqbMn058nLsv98Odn7gq8ZuNvdO/+PSqYrxBd+e9fzTH87t2R8wXyFaSLf53X5OuasbG20ZWtyePA4+iJw+ln0UusQZ1GdPvuLTlGobLC2o2r5Y/33Gvod9E7r27N+soizWOcwrMc+HF2dpYT74EDgspz/sKma/B4zZHEzD///CNbgyi3x16xYgUVL16cuEUiD1zJrYB4MMVy5crJQSl5f7yMB8xXEvdR5/7rPPA8D2Sp6246yrpJ/DhWHM9HTN5iMuvEX/DKlBF3NhUTBxM48q88518OkPQS8BZrhbcXDl+d37T5Tc47/Klp/VUCgxwU5OAQj7HFgZLYgoq6zrJ67YayVd+RA7uJ57VT5izZiFv9cYs8vpPxupVfA5La63HA4Y0YE1A7OTg4USXRzJqnjGLMh/yFish5bqmoT+Jg4uW/LtC/Yqw1fg2sFC2kenZqIbtVcmCwSPEyMjDIrQ2Pi3Ef+EeR0uUryyDnut8WilZ6H2jdioQFj2IrJwdpuSWfkjgoyAETDnxwaljXk1avWkHc8o4n7qrLP/Qoz/nmH6t+WyHXU/YR3SMfSzswyPV+dO/CONc7t+5bsGABKa0X+Xj8oxP/n+CBizmtXbuWGjZsSEo3H5lpZn9iqztDnG7pEuG/XPLYIxyoU8YULF60pM7DcSvA3b/vpHKiRW8uMa4nd33wOXFEtirkDTgwWKxwcRkY5AGReTwTDjxGTnzXNm6VuGbDb8R3dVu1fmXkVQzyPDbjxHx9aJ9AbMfVXlfXPF4j8v9iov+/fCpaRt4RQbsb127Km4pMHDaRxkwdIwLl1uK9uzT9+9e/4v1efIAX7/erF6+iPp36yvf7GvWq086NO+nh/Yfke92X+nbuR2ns01D1utVp+wZxQ5O7D+RQBntF12R3L3cqVb6U+D+Qmjb8toE+Bn4Ud0X+Oq5uTMfRdS2oJM9blAP1oZLKEMUwSH1s2LCBTpw4oZ6z/FoSg5wv3g++AsdxDvURRzADr54o9fH48WN5w0DuycMTNxzQJ+XPn1+uxp/p+UfnS5cuUZcuXejVq1fk7y+Gm3J1Je7t8+bNG3nDQBNqQKa3aynRkITH7uabh/A4i3yO/FmZA3KFChQiL/E9lG8mcv7P8F6XO/ZsI+4WrPR2UpyV4OubONyt+NXr8B4Q7hU9wr+rrlkmd/c57LNssThefMaZ7D1N3hxl5Phhmq7LyjH50Vb7SXznlcHj+/UTXfLEAPE8WDw/av9T4YEqeXD4wYMHy0Hm9+3bRx07dqQ5c+bIL6g8zxfRoEGDiO+AyxfN3LlzZZ/1H3/8UbZM4xtf8GDzHGnmKDRfbDxgPQcTVZDGiTJwgLCqVlm8tebNdlZpDWK2J2iYE+NrhRNfN948Y6qJv/hyUlqMxScgqFxD/JhJ3NWplGh2/Ul0eeUbiyh3duJj5BXdDyuLAEWbJl4y6FZRBPaui5uBcOJtlTdSd6/aNGZID3IQ45mFL4v6O4hyTLlCLH94XXfPmtSxW395bF49vfglaJa4iQO3CGwnxm/juyZvE78KpUufkZp+247m/TyRatdvRgOGT6DxI/rKic+Lux3H5dixFC3CYiVIq7QUjLBQPOnX/RsqU7UduXt4am5KopjxuhyMG+s9ni4c+y3yplGe85iSnBJS78pOvby8qHXr1jIIyHk8Lkn9+vXJw8OD7O3t5a+PSrdj7XpWto/8GNk38vPI66vheWx1F98yWsnfHHRvzQMaDxPjj/BgxZNmjJd3DJ4+4WfZmk+XWeUK7rLbcOmSZeQOq7p70g3f65RBXPOcuLsC3yFts2j5m94pvWxROGfBLKpdvY5crpSFu+8O6TuMxkwaKSfu7iy7NovXmSFTbMaJ+frQPo/Yjqu9bnTzFv4aSfT/l/x5c9H/FsuJzYuXLk5DxV0ClS7ClTwrUftu7al9kw6ySnjsvxkLpsuWfw2aN6Bdm3dTrbK1ZdCvW/+ulDN3TqorAoBbxU1JapcPv9655WGDZvXlkAZ9h/eRNy7hm5eULFdSbsevsZiOw8utrQ37mpAnF/c/qI+4mxlyi0SvDy4sfwfj7238f1hlKdHPF+8HCaph1EeC+BJ940SpD74TMU9K4kZA/Dk9psT/sxwcHMjb21tO8+fPl6tzvIZvGsif53mMcf58z6lp06ayJxPf+Tj8/13U72pyRXX80duVu/tOE+PlDxkzkMZNCe9Fxd14+WZ9lURDE27YVl18l23xfWP52Zd/aF82P/x7FzvYfRkjPa19WjkcUM3GnnT899OxKvC2xcVNM/mH/HJe4T/yd273oxwXfOiYQXJMwwplK1Gjuk3kmIU8HiLfwZjLqp34U4coY5h2nl7zXACu4CZNmtCIESPo6NGjxM1FOfHgktwi5e7du8QXBgcPz549Sy1btpTBQF6nUaNGcuwrpVUgDxrPrQI5wsytz44cOUL8YZjvdsOtWa5duybvkMN92A8fPkw8mCXvl1utZc2alXepV+Jyi2SoT1uRIcd9KZS3eGGEKYPr61VQrKQKAb4Dk+jymNjXi/eXk1MedZ2ryXQrjq3r8abTQWSVwoWKF3DWdZ7xyuObgDiKwF+yZHbRbv/m9UvRmsNB/ugQ7UrxWPDyxTN6/uwp5RQDuyYXb+hKChJ3ouIbqWTL7ir/yfF6Do7p5PG5leNb8csPBzu/vAcpm8X4ePTcdSqcJUCvbsVKkFZpJRjdjtdvPULtu4yicd6j6ft2HWXLQf5FkFsMcmBw5cLx1Kp5jeg2l/l8LO5Ky8eKLhjsUrA++Rw7KX8hjHFnMSzksUv4fwC3II/cdTiGzeK0iH+pe/3gqFG7FcdWdyHBYbJbcd0OnnE6N31XDngbQP7i2s2RPSelTJFS3810rsdjzjwRr0++Wxtf6zyGiqPoks9fviKnl6J179t3AfKubnF5XUTeT+C7j3RyxznqPCxP5EWa57EZKysmxutD2Rc/6nNc7mYbn89hvH9DvUbu3LlDlStVUEO3Ym8+T5G85V8df8bv6hLWvKxh7kr+8sUrcefi55QjV44oLZd5TEIOGnKwW0ncHf/hvQeUzC6ZGC83S4T3ex4z613AO9Ft2SVCPm8b03GUfcf1ccv5OTSmYdJ/fkF96K4pY9WHKE28vufpPgv9cw30fcv7SwmUxygFwvUXhURmGOv6Q32oqz6SKh7BPUG5pSD3MuWgoJL4sw5/buGblfDnwvfv38vP93HtGWSg7+dKMSM/en/JUB4jL6f5A9eG1e/kFSGfe8bcu3+X0oieMspYgdor8GdlPv8cOXLKH+W1l2nPc8tBh7QO2lmxzvOQQfzZO5UY+5DHQOS7JXOwMba0Z9nRxGk5GNvg8UqLD27tpyRuScj52lFp7YuHuw1zUvL4xJo1a0bt27eXt7jmfG5tOHnyZGWXcXmMHMSLaVufmBbGskyJMseyGhZbmIC3OZ1vQsdZi48FdwuOLXFgzhCJWwbyFDnZ2SUXY0vl0mRrr8N3TY7rnZM1O9JzJragoLKbVs2qUeECv9HsXzeJFnrF5c1HeIzBhnWryhaDRQvnVVaN9pGPlRRvbjxshSUkfevOUBb8gUGfDw36HJ+DrTymoJLSp0uvzEZ55DsV85QUSV/jxHh9aJ+PvsfV3iYu8xbwGvGOi0dir5suvZN4v3fSuVu+m3HkxF2SOZCoK/GdinnSlWI6jq71jZjnbcRjy7pAfUSoAaPWR4SSJM0To55vTK9TvB8kzQWgfRTUh7aGnPeOkmOkjBQpUsjWgpEPzz8aRL6RSeR1VPjcOz5l4h/beRzB6FJmPb7L8rZxDQzyNs6iF56SOPAal+CrrbJhQh5jGjyeB7xv3Lix7GrMt7zmgCCPHZUuXToaOnSoZhB67ovO/duVpKuVAd/4YsqUKcQt8HgAy169eskB6rnbcRxTXFqCecZh35HX5TFZ/MTkI6ak+C4tDoMEAQhAIGYBDgAu/WW4WGm4vBupPjcfiXmPWAoB8xHA68N86hJnAgEIQAACEIAABCCgn0CidO6ObfB4HniSA4jcBLRHjx7y9tU8AD53S+Ymp+/evaOePXvSjBkzYiz19u3bqUOHDnKd5s2byxuTcLNVAycfsX99Jk+x3jExceKg4F9iaiom7vTtIyYkCEAAAqoSULo8Ko+qKhwKAwEjCyivC+XRyMXB4SEAAQhAAAIQgAAEIGAwgUQJDmoPHs/jA/L4gaNGjdIUWhnwfvTo0fJGI9OmTaOBAweSi4uLHI+K74LM/c95zMHoEjdD5ZucXLlyRbY6dHR0lHd/adOmTXSbJHU+twysKiYEBZNaHseDAATiJaC5UcKXm8rEayfYCAJmKoDXh5lWLE4LAhCAAAQgAAEIQCCKQLy7FWsPns2Bu3Xr1tHMmTOjDB7P3YCV5OTkRAEBAcpT2TWYb1rC4wnyra+/DF4bZWBu7WP5+vrS7du3iffFg1mqJHmLcviIadyXR/GABAEIQEC9ApFbQ/FzQ4+Ppl4NlAwCEQXw+ojogWcQgAAEIAABCEAAAuYtkCgtBxUiHhi7QIECet9VkoOBrq6uchslMKjsK7pHHlCR74SsosAgF9VHTF5fHsUDEgQgkNgCL54/pfxZbKlbuyYRdv3wwV3at2uzJu/g3u10z++W5nlcZ7T3Fxj4QR7z7h3fuO5G9esrraKUgkZ+ruTj0bQF7t73o5yFs9LAEX0jnMj6zWuoUcu6EfJ0Pdl/aB/xPhI7+Zw4QnWaVk/s3Sba/iK/HiI/T7QDYUcQgAAEIAABCEAAAhBQgUCiBgdVcD7GKoKPsQ6M40LAUgR+372FUqexpyMHdtPLF880p3396mWaMXGE5vmcGePoyuVLmudxndHeX/LkKWjN9qOUSeuuT3HdnxrXj9wqSiljdPnKcjyarsDmHRvp7IUzmhMI08zFPDNr3gz6V7zGLClF9zqILt+SbHCuEIAABCAAAQhAAALmKYDgoHnWK84KAmYnsHnNMho8ajKlz5CJuHUgpwf37tCk0QPo/t3b1PuHb2n21LHEwb3JYwbRqeOHKTg4SAYOq5fPT01rlqU92zfI7T59+kjN61SgTWuXUb0qxYiXb1qzNMr+QkKCadZkMVbq65cUGhpK82dNoiolXaliEWeaPHYQffwYKPc3ckAXWrN8Pn3frLpcNl600uL11Zq4FZSjgz2VKJo/QhHROioCh1k9adm8NQ0bO1C8JoJ1ntcm8drglnwetSvQz79Mp5DQEJoxdypdu3GVJojX1e7fd8qWhg8fPZDbT5zmLfY3SM4HBQVRk9YN6P6De+R7+ya17NCc3Mrlk+tfuHRernPl2r/Urd8PtHDZ/CgtBgPeBlDnnu1p/pK5OsuW1Jl4fSS1OI4HAQhAAAIQgAAEIGBsAQQHjV0DOD4EIBCrwK2b12RrwDoNm1Ozlu1p26ZVcpv0GTPTt206yYBht77DqXGL1pQ9Z276tm1nKlSkOP1vqrfscjx49GRq/2Mf6t+tDZ09dUzezOjyXxdosWgVNWL8TKpQ2ZNGDuxK9mkdIuyPA3wXzp6kj4GBtGXdcrl+1z5DaM6SDbRv52Za8L/wMVVv+f5H44b3oRatO1HXPkNp9bJ5crtYT8wIK3DrJw4Kbls3gy79sVaWICzgAnkP7yLn0TrKe3C6ugAAG49JREFUCJWSBIfs3aUvBYrrePnqJVGOxl18B43sR63E62ey9zTauHUdzVs0h5o1aEE5XHLSd9+0pUrl3enpsyd0UbxueBzgtaJb8joxcbDx8tW/6YbvdcqQPiO1/aElpUiRglYuWENV3D2pXdfW9OLlc3r/4T3tO7iH1m9ZQx3Ea1ZJH0TX/Y7d28pg5A/tuirZRnvE68No9DgwBCAAAQhAAAIQgIARBRAcNCI+Dg0BCOgnsHvbBqpWqwGlE8GHmvWa0J/nTtFDMQ5aypSpKF/BIpQqdRoqXLQk5cpTQMynpjz5C5GDYzpa9Ms06jN4LNVp0JyafPs9NRFBjgN7tmkOOn7afHL3rCkDhJzJ4xpq70+zopjZIIIqHGBs07EHla3gQb0Hjaat61dqVuHjNG7RhjqIIEzREmXonkrHKazqUVoGBT3dS2vKzjN8MxIOEiKZp0CqVKlp/KifaNKM8fTI/2GEk1wlrmNuWdihTWdyr1iFBvQaTNtFN/7cufJQarFd3tz5KJ1TOqouXiscHPQTLXZTiC736dNloCv//UvnL56j2tXr0NmLZ8j/iT/Nn7mQypYqR4N6D5XHOXnmhOZ4y39dLYKQbeRzDhh26dNJBhgXzF5CdnZ2mvWMNYPXh7HkcVwIQAACEIAABCAAAWMK2Brz4Dg2BCAAgdgEuPXe+t8WyS683P03WHRh5PT7ri3UucfAaDd//vSxXDa4V3viSUn1Gn2jzFJWZxc5z2MZcgr6FL5v+STSn9uiZVT3fl/HNnTJkYse+4d3seRVnV1yaLZwFIGUoKBPmudqmokcFIxcNtyxOLKI+TyvVa0OVatag8aLbsJV3fkeWuHptp8vHfI5QBu2hLck5dzUIuAeOXlUqkqz580ULU9Liu09ydrKmv4UwcLT509Rk/rNiLsc58uTP8K2hUSg/vnz55Q1s7PMdxWvGyXdEzcT4omDjKEhIUq2UR/x+jAqPw4OAQhAAAIQgAAEIGAkAQQHjQSPw0IAAvoJcLdebtG3bP0+zZ3Qd4luj1tEa6eYgoPcRZjTgt+2U0WPanL+8aP7ZG1tI+f5j43N13lNZjQzRYqVoidawcA7t26Qh1ctzdo2Nng71WBgRrUCY4eNp6p1K9EH0WpPSQ4OTtS9cy/q22OAzHrz5g09E6+5yKl8mYpyDMKDR/cTBwr59cN3M+ZuyVPHzZR3NX70+JEcb5OXcffjazeu0QjRyjY09HPk3cmg4P5th6nht3Vo9q8/0yjR+hYJAhCAAAQgAAEIQAACEEh6AXQrTnpzHBECEIiDwE4xrlntBs1k998KorUTT607didfcaOEG+ImB9bW1vTh/TtNyyNbEaR78+olpRTdIUuULk+8Pbc25G7I7VvUpjMnj8Z49Mj7U1auUbcxbd3wm7z5Cd/oZPe29SI4WFtZjEcImIQAt9zjbsPH/vDRlNe9ogftP7xXjinIQcPRE4fTohW/yuU2tjb0+s1rOe/k6ETFi5SgPft3UanipWXXYW5xyK0Fs2TOQqVLhHdVXy9aIPINTY59ea0VFy0NdaVM4uZCGcXE3Z0Xr1ggA4+61kMeBCAAAQgkrcCGDRvoxImvQ0Ik7dFxNAhAAAIQMIYAgoPGUMcxIQABvQQCxc0K+I7C9Zu0jLC+mwhQZMnqQvv3bCWe51S3ajH56C4CdmOG9JA3IpkwfQH9ef40lSmQQd6VuGjJMtS8VQe5nq4/VlZWUfbH63F+vcbfyq7C3LW5cPZUxHc8btjsO127QR4EVC3QRQTXs2bOqiljlw7dRXDPmarUqUilPIrSE9Elf/iAUXJ51cpeNGLcEBkQ5AyvKtVl92AOCHKgkbsE81iEnGxtk9Gw/iPk+sUqFqL24gZA3qKlIgfs+TUUXeLuztwScdT4YdGtgnwIQAACEEhCgblz59KePXuS8Ig4FAQgAAEIGFsA/eCMXQM4PgQgEK0A33DkxuOoY5FxoOH4JT/Ndif/uifHJOSMgSMm0g+ie2QaewfZ7fHoeV+6K8ZU45uWcECRk22yZFH2q30cZX+Rj7/94AV6IG7GkEzcOCGrc3ZNwGP9zuNyv8qfpev2KrMW/ZguXTqLPn81nHzO7K5094p/hKKkTJGSzhz5U5OX1j4trV26UY4ZGCRa2eZyza25tof0G04cTLT/Mi5n/56DiCcl/XnisjIrH9t915GaiBsA+YvuxTmy5yQ+Fie+QcnVczflPP/xFF39eVLS6sXrlVk8qkDAyclJBaVAESAAAWMJnDx50liHxnEhAAEIQMBIAggOGgmeD7tyzA4jHh2H1iXQfnxjXdnIU7mAja0tKTcV4aLynYqVxMty5y2oPNXrMfL+lI14HLWcufIqT/EYg0Dgx0+UPn36GNZQxyIbGzSg55rggLtLtuw6K8XRwVFnfnSZHGzkCUl/gZhaVuq/l8Rd08kxfNzWxN0r9gYBCEAAAhCAAAQgoEYBBAeNXCv1O329Y6SRi2Lxh9+zLOax6CweCAAQiIPAC79DcVjbuKsGq+ROucZVwNGNJRAWcMFYh8ZxIQABCEAAAhCAAAQgIAXQZAIXAgQgAAEIWLRAMtG6FAkCEIAABCAAAQhAAAIQgIClCiA4aKk1j/OGAAQgAAEIQAACEIAABCAAAQhAAAIQsHgBBAct/hIAAAQgAAEIQAACEIAABCAAAQhAAAIQgIClCiA4aKk1j/OGAAQgAAEIQAACEIAABCAAAQhAAAIQsHgBBAct/hIAAAQgAAEIQAACEIAABCAAAQhAAAIQgIClCiA4aKk1j/OGAAQgAAEIQAACEIAABCAAAQhAAAIQsHgBBAct/hIAAAQgAAEIQAACEIAABCAAAQhAAAIQgIClCiA4aKk1j/OGQBIKpEmVPAmPhkNBAAKWJJDMztaSThfnCgEIQAACEIAABCAAgUQXwCfqRCfFDiEAAW2BN+8/0/ErV7WzMK+nQOqUdlQ4C37D0ZPLrFf7/DmM9iw7atbnmJCTs7HF6yQhftgWAhCAAAQgAAEIQMCyBRActOz6x9lDwOACP9RIIY7BE1J8BN58CIvPZtjGjASsbayo+9j8ZnRGOBUIQAACEIAABCAAAQhAQE0C+KldTbWBskAAAhCIJOCQyipSDp5amoA1/lNbWpXjfCEAAQhAAAIQgAAEIJCkAvjKkaTcOBgEIAABCEAAAhCAAAQgAAEIQAACEIAABNQjgOCgeuoCJYEABCAAAQhAAAIQgAAEIAABCEAAAhCAQJIKIDiYpNw4WGIJhIaGEk9IEIAABCAAAQhAAAIQgAAEIAABCEAAAvEXQHAw/nZG3fKPMycoZ+GsEabOPdvT6XN/xKtcgR8D5b787t2hoKAg2rh1HXGedn68dhxpI13lVs7jydPHkdaO/umYSSNo7sLZ0a+AJRCAAAQgAIEECnh6lE7gHrA5BCAAAeMJ5M6d+46Pj4/xCoAjQwACEICAyQjgbsUmU1URCxoWFn4H0wvH/iZrMVr9hw8faI4IlvUc2I3O+VwiW5u4VW1yu+S0ceU2ypwpC30I/ECDRw+gqu5elDFDJk1+xBLE75lS7pP7z0bZQYb0GaPkIQMCEICA2gU83UuT14mLai8myhdHAZ+TqNM4kmF1CEBAZQLv3r17m9RFGjduHHl4eGw4ceJEUh8ax4MABCAAgQQIoOVgAvDUsGn6dBmIp+wuOajT9z/Qi5fP6eatG7LL7dwFs6lCtVJUyqMoTZg6lj5+/CiLvP/QPmrZoTm5lctH/Yf3Jv7cEBISQjPmTKHXb15T176d5Xrtu7ahJ8+eaPI58+DR/VSjUVW57Y+9O5L/E3+57pqNq2jm3GnUb1gvebz23dqIsryQy3T94fJGnmxsbIj3M3GatywXl53n9x7YTV71K8vpyLFDmt3d8L1O37RrIsvSZ0hPehPwRrMMMxCAAASSUoBbmCGYlJTihj/WMRHw5cAvEgQgAAFTFXB0dBzu5eVFSdV6kI/j7e1Nvr6+j0zVDOWGAAQgYKkCCA6aeM0rXX8fPnpAazb8JgOF+fMUoI3b1tOvy+ZRjx9706+zFtOe/bvol8X/o5evXlKXvp2o3XcdaNH/ltK1/67Sui1rKfRzKJ27eFYEEAOpZ5c+UqV/z0GUOlVqTT4HHX/o1YGqVK5Ka5ZsoJDQEOo1sCtxa0DuEjxnwSzKlTM3TRw9mS5f+YdWb1gZre6OvdtIezpz/pRcl/ezeOVCyp4tB/XtMVDOj5ownPr3HExlSpWnsT+N0uyTz8mjUlWaNHoKnTx9nKbN/kmzDDMQgAAEIAABCEAAAhCwZIEbN27sdXNzW8QBQm7RZ6jEQUGe+Di5cuWa6+/vP8BQx8J+IQABCEDAMAJx63tqmDJgrwkQcCubV7N1ofxuNHX8DOIWeOs2rZYtCdt911Eu7ycCbbPnz6TWLdrK589fPCNPj2q0bP5vmhaFyo6KFS4uZ0sUK0nJkiVTsmnbri1UsnhpGjN0vMwbNXgsVWvgQf6PH8nn5UqXp77dwz8L/Hv1Mt26c0uzbeSZRct/jZBVVgT+KpStJPPKlCxLA3oNlvPT/zdFBjIb1WsiA488FmJISLBcxmXp062/nH8rWj9OmTWJJo2ZKp/jDwQgAIGkFBg7vAuNm7yIPPcsTMrD4lgGFPAW9RkWcMGAR8CuIQABCBhe4OrVq13t7e1fHj9+PK+VlVULQxwxb968d0QX5vdVq1Ydf+zYsU2GOAb2CQEIQAAChhVAcNCwvgbf+7F9p0j8o5djDGZzdtEcz/eOL/Xq2lfznLvwchdg56zZiIN6YyaNlFO1qjVEsE+/XxLv3vej0iIgp6RsWcOPp3QfzpndVVlE4kMIffwUqHkeeWbPpgORszTPXXPk0sw7pE1LHPTkpAQqQ0M/y+fuFT3kI/9xK1iY3r9/J7tTc3AUCQIQ0E8AN1zQzym2tbj7Kb+TctdidEWNTUv9yznQ6y0CvkgQgAAEzEHg7du3w48cOWKwUxHdiOW+Hz/W/+aCBisMdgwBCEAAAvESsBZ3sbrFzcAtIfF58vma07nmcMlJHJTTDgzy+RV1K0aPv4wHyM9v+92mqpU96dXrV1SzWm26eekurV++WY43OGveTF4l1lQwXyF66P9Qsx7f2ZhTgXwF5aN1IgXlbGwjxaxF8FNXeiW6SCvpzt3bxC0XERhURBLnkYMcPO4WknkKYIy8xK1Xbj3oVa9r4u4Ue0tyARkYFMFBrk8kCEAAAhCAAAQgAAEIWIKA9fv37z9Ywokq5yju6ht+Vw4lw0wfa1evQ5u3b6R7D+5ScHAw7RRj/FV1r0Y8NmG9FrXo8VN/qliuspzevguIoMB3P+YU+QYfXlWq0/FTx+j8n+fk8h17tlENz1pkZ2cnn8flz7UbVynypNwwRd/9bNu9lThAyee0fvMa4vIhJa5AVdxkIXFBVbY33HAhcSuEg+nc2swqbRnZxThx9469GVqAg+Ve9bvK1p9H96J7uKG9sX8IQAACEIAABCAAAfUI2BYvXvynYcOGLTlz5kxq9RTLMCXZt2/fq9SpUx81zN6Tdq/clTim1LBuY9q4dT151K4gVytepAQ1adBU3rDE091L5vNdjrnb7oxJszW74v2mtU9LPO5fzcaedPz303IZ5xcuVISqi27ILb5vTKlTp6EUyVPIMQs1G0easbaKer8bpdx1mkYN5P2+7XCkPYQ/jXyqyj7y5spLVeuGj1PIXY9bNP5W5/bIhAAEIJBUAtzaTBl/kIOEStdUDrSju3FS1YJ+x9FuOcutBX1EK2muL7QY1M8Pa0EAAhCAAAQgAAEImI+A7YEDB9aL01nHXW49PT3N58x0nMm0adOcRHZvHYtMLqtyBQ+6e8U/2nJnzJCJ9mw+QPcf3iO7ZHaUNYuzHJuQN5j/8yLZ2i4oOEje5EPZifb+tqzeKVsOOqR1iHCcudN/lWMWihanlCNHTjnWIW+v3EBE2Vf3zr2U2QiPsZVbGV9Q2ejonj+UWSqYv5CmLMqNRz59+kTPnj+V3aqVoKFmA8wkWICDGRhHLcGMqt0BbrhguKqJHGDi4JMXuugbDjwee1bG2+T3ufCgLiGAGw9HbAIBCEAAAhCAAAQgYPoCcnC3YsWKLRk6dGjLs2fP2pv+Kek+g379+t13c3PbK+7YpXsFM8zl8fe0b+6hfYqRxyjUXqbMc2BQV8qcKYuubKPkJU+enFyyZTfKsS3loEorKNyF1bxqHDdcMHx9agcIxxr+cDgCBCAAAQhAAAIQgAAEIACBeAnIfp///PPPj/dFEt2LX8drLyrfaMSIEe/FOZ4TgcFuKi8qigcB1QkoXSE5mIRkHgIyMIgbLphHZeIsIAABCEAAAhCAAAQgAAEIJFBAMyicv79/4dWrV/v379//UQL3qZrNuat02bJl34jxFH8/evRoC9UUDAWBgIkJHN0TPjg/D9aPIKGJVZ5WcXHDBS0MzEIAAhCAAAQgAAEIQAACEICAFJDdihWLhw8fuu3evXve7NmzewwfPvxtrVq1ZDdjUxqLkAOCnERrwcDTp0+nFF2J11+4cAEtBqUK/kAg/gKyi6Robcbj1CmJb7LASWldqOTjUR0CuOGCOuoBpYAABCAAAQhAAAIQgAAEIKBmgQjBQS6or69vT/HQc9WqVbNXrFhRM2XKlMlv376dR80noV223Llz3/rw4cPHChUqTBX5qyxpjEFtB8xDwBAC4YP2d9G0HjwmbrAgW6PhRguG4E7wPnHDhQQTYgcQgAAEIAABCEAAAhCAAATMXiBKcFA54wcPHvRT5k3pUQQyZXG3b99uSsVGWSFgUgK40YJJVRcKCwEIQAACEIAABCAAAQhAAAIQiFZAM+ZgtGtgAQQgAAEIQAACEIAABCAAAQhAAAIQgAAEIGCWAggOmmW14qQgAAEIQAACEIAABCAAAQhAAAIQgAAEIBC7AIKDsRthDQhAAAIQgAAEIAABCEAAAhCAAAQgAAEImKUAgoNmWa04KQhAAAIQgAAEIAABCEAAAhCAAAQgAAEIxC6A4GDsRlgDAhCAAAQgAAEIQAACEIAABCAAAQhAAAJmKYDgoFlWK04KAhCAAAQgAAEIQAACEIAABCAAAQhAAAKxCyA4GLsR1oAABCAAAQhAAAIQgAAEIAABCEAAAhCAgFkKIDholtWKk4IABCAAAQhAAAIQgAAEIAABCEAAAhCAQOwCCA7GboQ1IAABCEAAAhCAAAQgAAEIQAACEIAABCBglgIIDpplteKkIAABCEAAAhCAAAQgAAEIQAACEIAABCAQu4Bt7KtgDUMJpEqTgm5e8jPU7rHfOAikFHWBBAEIQAACEIAABCAAAQhAAAIQgAAELE0AwUEj1fi71x8of1lXIx0dh40qEEZ5S+aImo0cCEAAAhCAAAQgAAEIQAACEIAABCBgxgIIDhqpctM4pqLiXgWMdHQcFgIQgAAEIAABCEAAAhCAAAQgAAEIQAACRBhzEFcBBCAAAQhAAAIQgAAEIAABCEAAAhCAAAQsVADBQQuteJw2BCAAAQhAAAIQgAAEIAABCEAAAhCAAAQQHMQ1AAEIQAACEIAABCAAAQhAAAIQgAAEIAABCxVAcNBCKx6nDQEIQAACEIAABCAAAQhAAAIQgAAEIAABBAdxDUAAAhCAAAQgAAEIQAACEIAABCAAATMVcHR0vObr62vSZ8fld3Jy+k9NJ5EidfJrLx6/VlOR4lwWLn/K1Mn/Q3AwznTYAAIQgAAEIAABCEAAAhCAAAQgAAEImIaAnZ3d8b17934yjdLqLuWNGzcCU6ZMeVL3UuPk2iSzOX7j4m2Tdn3x6FWgXfJkJxEcNM41hKNCAAIQgAAEIAABCCRQIFUy+xvP3j5M4F7Ma3P2SJXc/oYxzgr1EVXdmPURtTTmnYPrL2r9GvP6Q32oqz6eP3++/s6dO8lNufXgwYMHU3p7e/8YVdZ4OR/eBK5/+eRNclNuPXjzL7+UbUY2+hHBQeNdRzgyBCAAAQhAAAIQgEACBELCggMTsLlZbvr87QNKmSzNOWOcHOojqrox6yNqacw7B9df1Po15vWH+lBXfSxYsMAne/bsy3ft2vU2asnUn7N169Z7zs7OS9RW0u4zvvNxSG+//No5X5N0vXL25j2HdGmkK4KDaru6UB4IQAACEIAABCAAAb0ErKytBv/74KRJfiDX6wTjsVJwSNDTgMCXRmlOifqIWmHGrI+opTHvHFx/UevXmNcf6kNd9cGlGT16dCdra+udO3bsMMr/iKgisedwS8fZs2e/f/jw4Sm1tRpUSt9mRMNONtZWO6+e8zUZV27p+Meui+8Dnr07xa0G+VwQHFRqFI8QgAAEIAABCEAAAiYlMLT2/w4GBX96ga7F4dV27dFZuvnkUqbh9eYOM0ZFoj4iqhu7PiKWxvyf4fqLWMfGvv5QH+qqD6U0I0aMaHvr1q0D/fv3p927dweotZsxl2vu3Lkh8+bNo+Dg4HXDhg37TjkHNT62Gtqg7Uv/1wf2LDtK/124HaDWbsZcrtN7/gw5s/cShYZ+XtdqaD2Nq60aYVEmCEAAAhCAAAQgAAEI6CPgkCbdWNF68BevQi3t9VnfnNcJCgl6kNIuzVZjniPq46u+Gurja2ksYw7X39d6VsP1h/pQV30opeEWhGK+U6ZMmRZcvHixyuvXrwspy9TyyHclFq0cD4lpi2gx6KOWcsVUDm5BKJZ3Wjlh+4KHtx5X+fj+k+pc+a7EVsLVyspqy/cjG0VwtYrp5LAs8QS6dOkSNmvWrMTbIfaUJAL8i8qiRYuM8ToJCwu4kCTniINAAAIQgAAEkkrAKm0ZPlSi/1+dvn/A1RzpC2Yr4lIpbVKdi9qOc+XB6Q93X1y7N7j2z0b/MoL6IFJTfajtWjV0eXD9qev6Q32oqz4M/frD/k1XAN2KTbfuUHIIQAACEIAABCAAASEgAmJu91/eeHL14ZkgSwThQNS9l9cfqiEwyP6oD3XVh6W9JnD9qev6Q32oqz4s7f0A56u/ALoV62+FNSEAAQhAAAIQgAAEVCowqNaM/NxCJfhzUCo7m+Q5CzmXV2lJE69YPNbi5XsnAz6GvH+klsCgcnaoD+O34FTqwhIfcf2p6/pDfairPizxPQHnHLsAgoOxG2ENCEAAAhCAAAQgAAETEOAWKjMPDP7l/ac3PYNDgx4ls7FzzmDvQhnts5lA6fUronLzlcv3T74PDvn0zD6l04je1Ses02/rpF0L9ZG03jhaRAFcfxE9jP0M9WHsGsDxIRCzQKKP+RLz4Sx3KY85aLlnb9pnbqwxB01bDaWHAAQgAAEIRCuQJJ8/p+3rNz1lcvtMn4IDK34ICsgXbWlMbEGq5PY3QkNDPtpa240cWHv6blMpPurDVGrKPMuJ609d9Yr6UFd9oDQQgAAEIAABCEAAAhCAAAQgAAEIQAACEIAABCBgwQL/B6xLUrHY8Fr7AAAAAElFTkSuQmCC"
+ }
+ },
"cell_type": "markdown",
"id": "ef121b7b",
"metadata": {},
"source": [
"# Архитектура GPT-1: Принципы работы и ключевые компоненты\n",
"\n",
- "\n",
+ "\n",
"\n",
"Модель **GPT-1 (Generative Pretrained Transformer)** — это первая версия архитектуры семейства GPT, основанная на **декодере трансформера**. \n",
"Она была представлена исследователями **OpenAI** в 2018 году и стала основой для всех последующих моделей, включая GPT-2, GPT-3 и GPT-4. \n",
@@ -193,14 +212,19 @@
]
},
{
+ "attachments": {
+ "image.png": {
+ "image/png": "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"
+ }
+ },
"cell_type": "markdown",
"id": "47d11c5c",
"metadata": {},
"source": [
"## 1. Эмбеддинги (Embeddings)\n",
"\n",
+ "\n",
"\n",
- "\n",
"\n",
"Перед тем как текст подается в трансформер, его необходимо преобразовать в числовое представление. \n",
"Это делается с помощью **эмбеддингов** — плотных векторов, которые кодируют смысл и структуру слов.\n",
@@ -232,7 +256,7 @@
},
{
"cell_type": "code",
- "execution_count": 3,
+ "execution_count": null,
"id": "1464a012",
"metadata": {},
"outputs": [],
@@ -274,7 +298,7 @@
},
{
"cell_type": "code",
- "execution_count": 4,
+ "execution_count": null,
"id": "94ddd50d",
"metadata": {},
"outputs": [],
@@ -297,14 +321,19 @@
]
},
{
+ "attachments": {
+ "image.png": {
+ "image/png": "iVBORw0KGgoAAAANSUhEUgAABQgAAAH1CAYAAAC6IxO6AAAAAXNSR0IArs4c6QAAAERlWElmTU0AKgAAAAgAAYdpAAQAAAABAAAAGgAAAAAAA6ABAAMAAAABAAEAAKACAAQAAAABAAAFCKADAAQAAAABAAAB9QAAAAAKhl38AABAAElEQVR4AeydB1wUxxfHH0WsKNhFVOyKvTcQsPeeaNRYE3vvWLHFHo1GYy+x966xY4m9/WPUqKjYe+8C8p83uOdxHHDAHbfH/ebz2dvZ2Slvvm/3bu/tmxkbQogXAh06dAiNl4bQiNEJzJ0718bolUZfIa6X6BkhBwiAAAiAgGUSiJff1Sm7+tcJCvk02t4+UbL3n97ksUxUEaVOlsjxamKHZMdffXh6e2jtP4ZFzKHOFOhDnXqxFqlw/alL09CHuvQBaUBAIRAvD2hKY9a8ZwPh1KlTrRmBRfa9d+/eZC4DYejr0xbJDEKDAAiAAAiAQGQEbFKW5FMmff5cdHRKuxdvHw1LZJc4daGsHim5wXSOmXmXIMKTN/fo6Zu7FBTy+f61h+dckidONbNvtUnd1No56EOtmrEOuXD9qUvP0Ie69AFpQECXgEkf0HQbs+ZjGAgtU/swEFqm3iA1CIAACICAOgmY2kA4cVfvS0ntU2Rmw2BCMgpGps3L909IQ+Hd59de9a/+q3tk+cyVDn2YizzaZQK4/tR1HUAf6tIHpAEBfQTs9SUiDQRAAARAAARAAARAAAQsicDkXX2vZU2Tz6WAa7lkliR3XGTN71KGi7vY2tilmLy739V+1SarZig19KEufcTlOrPEsrj+1HX9QR/q0ocl3tOQOX4I2MZPM2gFBEAABEAABEAABEAABExDgD1TrM04qE2yoGv5lNnS5Ms2aVefS9rp5opDH+rSh7muA3O1i+tPXdcf9KEufZjrvkS7lkEAHoSWoSdICQIgAAIgAAIgAAIgoIfAr3sG/JExVXZHa/Ic1IOB3DOXdXj4MtBlwdHxrdqXH/SnvjzxkQZ9hFFWiz7iQ+dqagPXn7quP+hDXfrQvlc7derknTFjxhYfPnzwePHiRT7tc2qIOzk5XXFwcPAfM2ZMJzXIY6gMf/Rb6Z3SOXmLz5+CPD68+6Q6rkmSJ7lil8jWv/WwBnq5wkBoqKaRDwRAAARAAARAAARAQHUE3n582ckhjYPq5DKHQIWyeqY6e3PfSNG22QyE0Mc3zatBH9+ksY4Yrr9velbD9Qd9qEsfijR+fn7z7t+//1OSJEmCGzZsaJ8rVy7llGr2AQEBecWWW6zl0DF9+vRzLMFQuHzslnmvnr/9yS6RXXCRsrnt02R0Ug1PRZBnD1/mffbgZe5ZfVd0TO6UbI6uoRBDjBVS2IMACIAACIAACIAACFgUgXE7uo/PlaHY3a9z8VmU7KYQlhdmcUiUOM3v/sMbmqL+6OqEPsITMrc+wkuT8I9w/YXXsbmvP+hDXfpQpBk/fvxKOzu777p27Urdu3dXpXGQZWWjZY0aNWynTp1K7u7udSZMmLBa6YMa96sm7FhJtrbfla1VjMrVLq5K4yBzY6NlnmJutrXb+VB61zR1Vk/aGY4rDIRqvLogEwiAAAiAAAiAAAiAQLQEUiRJld7B3sE12oxWlKGgq4fj2w8vR5ijy9BHROrm1EdEaRJ2Cq6/iPo15/UHfahLHywNew5my5bNp0+fPqnU6DUYkVhYSv369TMHBQXVHDp06OzI8pgznT0HndKn9PGoVyKVGr0GI2PjXjpX5uDg4JpLRm/ScIWBMDJaSAcBEAABEAABEAABEFA1gU/BHyqkdYR9UFdJ9rYOZlnJGfrQ1UTYsbn0oV+ahJuK60+/bs11/UEf6tIHS8PDiuvVq5dBv2TqTq1bt67j48ePO/LciWqTlIcV5y+d0yK55i+dy/Hdy/cdee5E5goDodquLsgDAiAAAiAAAiAAAiBgEIH3n97k4WF0CN8IMI/3n1/n/pYSfzHoIyJrc+ojojQJOwXXX0T9mvP6gz7UpQ9fX98ZVatW/RBRKstIYY9HsYWmTZu2mZokXuy3cUbuom4Wy5U9HtNkcgpNliqp5AoDoZquLsgCAiAAAiAAAiAAAiAAAiAAAiAAAiAAAkYk8OXLlyp58uRJasQq472q6tWr23z+/Nk73huOosFQwTWNi7NFc81dLLtNSNAXyRUGwiiUjVMgAAIgAAIgAAIgAAIgAAIgAAIgAAIgYMkEXrx4kc+S5h3Ux5rlf/nyZV5958yV9uHdp3yWNO+gPk4s/8d3HyVXGAj1EUIaCIAACIAACIAACIAACIAACIAACIAACIAACFgJARgIrUTR6CYIgAAIgAAIgAAIgAAIgAAIgAAIgAAIgAAI6CMAA6E+KkgDARAAARAAARAAARAAARAAARAAARAAARAAASshAAOhlSga3QQBEAABEAABEAABEAABEAABEAABEAABEAABfQTs9SUizfQEXrz5RGeuPjV9Q2jBYAKsk++8cxicHxlBAARAAARAAARAAARAAARAAARAAARAICEQgIHQTFp0dkxMJy4/o/w5MphJAjSrTSBl8iR0+so9GAi1oSAOAiAAAiAAAiAAAiAAAiAAAiAAAiBgFQRgIDSjml+/+0hlCmU1owRoWpvAnuNXtA8RBwEQAAEQAAEQAAEQAAEQAAEQAAEQAAGrIIA5CK1CzegkCIAACIAACIAACIAACIAACIAACIAACIAACOgnAAOhfi5IBQEQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAAGrIAADoVWoGZ0EARAAARAAARAAARAAARAAARAAARAAARAAAf0EYCDUzwWpIAACIAACIAACIAACIAACIAACIAACIAACIGAVBGAgtAo1o5MgAAIgAAIgAAIgAAIgAAIgAAIgAAIgAAIgoJ8ADIT6uSAVBEAABEAABEAABEAABEAABEAABEAABEAABKyCAAyEVqFmdBIEQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAAE9BOAgVA/F6SCAAiAAAiAAAiAAAiAAAiAAAiAAAiAAAiAgFUQgIHQKtSMToIACIAACIAACIAACIAACIAACIAACIAACICAfgIwEOrnglQQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAAEQsAoCMBBahZrRSRAAARAAARAAARAAARAAARAAARAAARAAARDQTwAGQv1ckAoCIAACIAACIAACIAACIAACIAACIAACIAACVkEABkKrUDM6CQIgAAIgAAIgAAIgAAIgAAIgAAIgAAIgAAL6CcBAqJ8LUlVEIDgoSEXSQBQQAAEQAAEQAAEQAAEQAAEQAAEQAAEQSFgEYCBMQPqsXCYP5clor3c7f+aE3p5OGTuEJo4apPdcXBOb1auoV5ZRg3saXPXDB3fJPUtS+vTpo8FlkBEEQAAEQAAEQAAEQAAEQAAEQAAEQAAEQMBwAvaGZ0VOtRNYve0wfQkJkWJ6FM1KY6fMIa/KNeWxc5q0esUPDQ2lL6Ff9J4zRmLHHoOoacv24apK4Zgy3DEOQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAAEzEcAHoTmY2/0ltOmy0DpM7rIjSt3Tp1Wc3xo/y6qVbEwFcvlTF3aNCL2zNMNVy5doO9rV6Aj/nvkqQ2rl1DdSsWIPROnTxpJIcHB0pOvcY2ytHbFQlkfn1u7fIFuVZrj1KnTkGvW7OE2J+c0mnqWzJsu22BvQ263d6cWVK6gi9x//PhBU8+c6RNkerUK7rR7+0ZNOiIgAAIgAAIgAAIgAAIgAAIgAAIgAAIgAAJxIwADYdz4WUTpgKuXqHPrhuThXZUWr9lFwcLQ16tjc2LvQSXcvH6F2EjHHoec79D+v2hQz/b0fYv2NHrSH7RuxSL647dx9OXLF7pw/jTNmzmZBo+aQmUreNOQvh0jHQL8j8i7beOqcNvLF8809axeOp+69B5CiewTUbtmNSl5CkcaPXk2bd+0mvbv2qqIRwf37ZTpxUuWo27tv6N7dwI15xABARAAARAAARAAARAAARAAARAAARAAARCIPQEMMY49O4spuXndCipaoow06LHQg/wmUg2PgvTg/h3Zh8DrV6l5fR9q1LQ1de0zVKatWDybvmvejn5s300e9xgwgub+Ponad+krj0dNnCWNg8VKhnkTssEuR6588pz2x/EjB4jr1w553QsLj0I3mTRs7DQq51mJPn38SCeOHiRfv0nSSFipWh0KvHGNipcuL/P1HTyWylesTJy+a/sGOnp4v5RPu17EQQAEQAAEQAAEQAAEQAAEQAAEQAAEQAAEYk4ABsKYM7O4EncCr1Mx4XmnBBfXbDL6/NlTud+/e5vca3vl3RBGPU7nocRKYO8+JWRycZVRJe3zp8/KqXD7Dt36U9tOvcOl8cGHD+9lmotrVrlPnCQJ5crjLo2DnJA4cRIK+TqfIh+zIZKDra2tNHayFyICCIAACIAACIAACIAACIAACIAACIAACIBA3AlEOsS4dOnSDTNmzPivm5tbgGiGx6JaxJYjR46ADBky/OPq6jot7ngSRg158hek+/duazpz++Z1Gc+Tr4DcV6lRj7YdOC8Ngoqx0MnJmTp0G0D/u/labofP3aKlG/Zp6rCzs9PE4xIxtJ5Xr15omvnv4j9UuFhpzTEiIAACIAACIAACIAACIAACkRLwFv/rLuTMmZP/BJjkP12ZMmWWOjo6jotUApwAARAAARBQPYEIBkJ3d/fZQupQYbhZvnLlygKLFi3KyXPVWcq2YMGCnKtWrSrUrFmz5tyPwoULz1O9FkwsoFeVWnIBkDMn/pYtbd2wUg7VdXBILI+zZs9FbERs27EX+Q3sJr37ylesQrt3bqInjx7Q+3dvaaRvd1owa0qMJX344B7x4ifam7anoqEVLp0/kz5//kTrVy2mZ08fSy9CQ8siHwiAAAiAAAiAAAiAAAhYI4GiRYvycKAD4n9dvvnz5+cw1X+6mjVrtnzz5s0g8V9yjjVyRp9BAARAICEQCDfE2MXF5aJ485P1wIED5O3tndQSOyjklmKLfbpJkyZR7969a6VJk2aN6NP3ltgfY8jsXrAoeQsj4Q/1veQQ3iRJktKcpZs1VdvahNmJu/QZQmvEisTzxQIk7cSw4LOnjlKVsnllvkJFS9KMBWs0ZXQjNjY2uknyeNGcacSbdqhepxFNnLFYO0lvXLvOf86dpIJZk8t8I8ZNl0OQ9RZCIgiAAAiAAAiAAAiAAAiAABPwO3/+fNuv/+3C/e8zNp4RI0aQl5cX+fj4dMiTJ8/mq1ev7jB2G6gPBEAABEDAtAQ0PxRsHGzRokWGiRMnpjBtk/Fb+9SpU10GDRpUVXhErt+7d2/j+G3dfK1dfRisaZwNbVNnL5cLgLx794ayZstJdvZhqu839BdNvlSpnOlcwLehvEvW7qb7d29Jzz23HHlIMdhp182FdY+VCldtOaRE9e61y9Ws24R4U8Jv81YpUU39vKgKy5gseYK6RDX9RAQEQAAEQAAEQAAEQAAEjEXA09MznzDYfRGOExFGjRmrDe162FGDjZHiP+UkkQ4DoTYcxEEABEDAAghIKxEPKxZu4WmEcTCNBcgcYxHHjx/vJObFqJorV66ZAQEBXWNcQQIpkD6jS4x6wgbBzFncYlTGlJkzuWQxZfWoGwRAAARAAARAAARAAAQSDIHDhw83PXQo6hf2puhsEh6uhAACIAACIGBxBKSB8NKlSx0vXrxoccLHROAJEyY4ijdoXUQZqzUQxoQX8oIACIAACIAACIAACIAACIBATAiwF+GNGzeyx6QM8oIACIAACKiDgG369OknDRgw4Nu4UnXIZXQp+MeqfPnyn6tVq9bM6JWjQhAAARAAARAAARAAARAAARAAARAAARAAARCwUAK2IojRxTWdLVT+GIk9duxYhwsXLgyLUSFkBgEQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAIEETMBWzBGRJAH3L0LXkiZNmjhCIhJAAARAAARAAARAAARAAARAAARAAARAAARAwEoJ2AYGBubk4bfWEL7OiZHTGvqKPoIACCQQAs9eJZCOoBuxJfDlS2xLohwIgAAIgAAIgAAIgAAIgAAIGEZALlJiWFbkAgEQAIGYEwgdMYdsDp2NeUGUoJDUacmubQ2iOp6gYcUEvoSE0twxARQaGmrFFKLuukNiO2o/CO//oqaEsyAAAiAAAiAAAiAAAiAQOQEYCCNngzMgAALGIJDWmR4MmUBvy3sbozarqiPj72PJ0ap6jM5GRsDGhqhWW5/ITlt1evDnYDqw9qhVM0DnQQAEQAAEQAAEQAAEQCCuBGzjWgHKgwAIgEB0BOyfPo4uC86DAAiAQKwIBAkDIQIIgAAIgAAIgAAIgAAIgEDcCMBAGDd+KA0CIAACIAACIAACIAACIAACIAACIAACIAACFk0ABkKLVh+EBwEQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAIG4EYCBMG78UBoEQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAAELJoADIQWrT4IDwIgAAIgAAIgAAIgAAIgAAIgAAIgAAIgAAJxIwADYdz4oTQIgAAIgAAIgAAIgAAIgAAIgAAIgAAIgAAIWDQBGAgtWn0QHgRAAARAAARAAARAAARAAARAAARAAARAAATiRgAGwrjxQ2kQAAEQAAELJxAUHGzhPYD4IAACIAACIAACIAACIAACIBA3AvZxK47ScSUwfcXhuFaB8iAAAiog8OXLFwoVm509vlZZHWncqtDz5y9VoJnoRXj36Aglgt6iBxWHHMEhwWRvh3tDH0KblCX1JZs9zdkpFT2/vc/sckAAEAABEAABEAABEACB+CGAp/X44ay3lQkdS+tNRyIIgIB+An27/EhbN6ykP9ftobIePppMZ08do2Z1Palm3Sb027xVmnRDI29ev6ISedLQqStPKFUqZ0OLhcu3euk8OnJwD81cuC5curUeJE2SmJ49e0apU6dWNQIHBwdVyxcfwjVuWY9OnzsVoakls5eTt2elCOkxTbhy7T+q1sCHbl18ENOiVpM/NDRUVX19+fIluefPqyqZIAwIgAAIgAAIgAAIgIBpCcBAaFq+qB0EQMCIBNhLj8OubRvCGQj37tws07+Ehp2XB7H5UNmf9Nh0QU1lnj9/rnoDoZp4mVOWn9t0otY/tA0nQtq06cId48B6CLx48cJ6OoueggAIgAAIgAAIgAAISAKYgxAXAgiAgEURKFqiDG1at4xCvs4bx0bDjWv+JE5XwssXz2hY/85UrqALNatXkdYuX6Ccoj07NtGPjSpTsVzONKB7G3r75rXmHEe4vnF+/Ym9FYODgujJ44fUq8MPsq62TWvQpQvnZX7ON2PyKKpWwZ04/X9nT4arBwcgYEkEnJ2cKYtr1nBb0iRJZRfWblpNNRpWJs/qZenX3ycRDxfm8OTpY+ratyMV9yxELX9uRhcv/yvT+WPH7m1U+7tqVK9pTdq4db0mHREQAAEQAAEQAAEQAAEQAAF1EoCBUJ16gVQgAAKREChVtiIlEYaLUyeOyBz//u8Mffz4QXgUfhsKOWPyaLp25SL9vmAt/dCqIw0RRozHD+/Ti+dPqWu7JtS8bWc5FPjyxf+FMx7yML8JIwfQzs1rqd/QX+R8gp1aNaBXr17QlFlLqWDhEtSgakl6/eol7dyylhbOnirq70DlxDDMDauXRCIxkkFA/QTYuLd5x0bNdvzUUSm0/+H91G9IL2rWpLkwnE+kNWKI/8y504nvlfZdW9MrMTx/+sSZVMi9MNVqUpVeC4P7vft3qXPvn6lU8TLUunk7WrMx5sP+1U8MEoIACIAACIAACIAACIBAwiKAIcYJS5/oDQgkeAI2NjZUr3Fz2r19I5Wt4C09Ahs0aUmJEiXS9L1qzfrU6qdulClzFkqaLLlMvxFwhbJmzynjz548Iq9KNWjO0s30SRgXlTB1/HDasn4Fbd57mjK5ZKF//zlLF86fpgOnAihzFjdpCFy3chEdE0aTg/t20vctf6K2HXvJ4vAgVChib4kE/I8coMvCqK6E8mU8qGyp8rR01RJqKu63Ni3ay1N9uvWnPxb8TpUqVqH//Xue/t59klzFfVahrCetFsbDv48fpjdv31D+PO7k5ztalnn77g0NHztEqRp7EAABEAABEAABEAABEAABFRKAgVCFSoFIIAACUROoWqsB/dyiLg0eOZk2r1tO439bQGdO/q0pZGNrS93af0dXLl2gjJlcNekumbPSoBETadTgnnLzrlKLBo+aojm/cskcGWcPQQ53bt2Qe59SueRe+Xj+7AntEfMejps2X0mi4iXL0dnTxzTHiICAJRHo+nN36vpzjwgi3wgMoL3+u2m1MJwrIXnyFHT77i15WKFa+MW2nj1/JhY8OUmlS5ZVslPhgkU1cURAAARAAARAAARAAARAAATUSQBDjNWpF0gFAiAQBYFiJcKMD4vmTqM3b15RqXKemtwhISHUXwx95KHIvCrxwbM3KXkKR3me5yasXKMe/Xv7Hf25fq+cf5DnEVTCBuENVbfRDzSsXyfiehwdU8lTRy/co//dfC23DbtOUPU6jSiveyF69OCeUpRu3rimiSMCAgmFAK/q3bl9N/rvzA25ndh/jlYvWi/ujZSyi2cO/aM5t3XNX1Szam1yy5qdHogh/Uq4dTtQiWIPAiAAAiAAAiAAAiAAAiCgUgIwEKpUMRALBEAgcgJ29vbEw4onjfalmnWbiOHFDprMwcFBGqMhG/h4AZN3YsgjLypy7+5tMYdgKWnY4+HJZT186I2YQ00JWbK6Uf9h4+jihXO0Ztl8KlC4mDy1ee0yshdtnvjbnxpVL0PPnj6h6rUb0aY1S+nWzQAx3+El2vfXFqUa7EEgwRDwEMb3Xft20GMxLP/9+3c0bIwvzV38BxXMX0j2ccOWdWRvZ0/HTvxNdb+vQc/EPJ+VvKrQ32KO0GPCq/f5i+e0bvOaBMMDHQEBEAABayCwevVqOnz4sDV0FX0EARAAARDQIoAhxlowEAUBEFA3AVsxdJjnIORQrXZDWi4MFbxXgq2NLSVOnIQ69RhIPcWqquw5WLxUOaoo5hvkxUnOBbygij7VqXKZPJQmbXpyTOVE46d9W+FYVC6HJPcdMpYmjh4kPQWnzVlBvTo2pwmjBspm+g4eQ7nzupOTWPV17fKFVLVcPpnOw5VZPgQQSEgEOrTpTGfOnaaKNcrJbhURw4Vni6H1qZ1T0++TZ1M34W079qsX7oBevpQnV15pjK9Y3ouatW0iy1QT9x8CCIAACICA5RCYMWMGeXh4kKfntxEaliM9JAUBEAABEIgtARgIY0sO5UAABOKdAK8krAReOfjqw2DlkLr3G66Jd+wxiJqJ1YWDgoIobboMcsXVZ08fy/O/zVtFA8T8aUGfP5NbjtyaMtp1dew+kHjjUKv+9+Qjhk3eEcMkM2RyIR5yySFdhky0df9Zun3rOqXP4KIZxixP4gMELIjA+mWRe7+mFEOJVyxYI1cm/izumexuOTRG+rpiMaDK3lXpjvDMzSjuh1Qpw4bks6GcjYh3790RBvvElE4Y4xFAAARAAAQsh8CRI0csR1hICgIgAAIgYDQCMBAaDSUqAgEQUBOBVE6pNeKw1yEbCpWQ2TWbEjVozysh58lXIEJeHuqcPWfeCOlIAIGERIDvH16pWF9IljQZ5c0d5kWrez6yMrr5cAwCIAACIAACIAACIAACIGB+AhgPZ34dQAIQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAAEQMBsBGAjNhh4NgwAIgAAIgAAIgAAIgAAIgAAIgAAIgAAIgID5CcBAaH4dQAIQAAErIhAS/G3eRDV0OyhIXfKogQlkiH8CwSHqvA5xf8T/tWANLYYEh1hDNy2mj9CHxagqQQqK609daoU+1KUPS5bmxcsX9Or1qxh3ITZlYtxIFAVgIIwCDk6BAAjED4Hr1y5Tnoz21KxexQgNtv+hljx35dKFCOe0E2b/No76d2stk/bs2ES3A6/L+AyxwurAHu20s+qN88rGq5fOC3fu4oVzsu0Xz5+GS4/LQcl86ei/i//EpYo4l71wMYDadxtH6XNWpxSZKso9H3N6fIQMGTJQzpw56ePHj5rm/vvvP7n4xevXrzVpiMSdADPOViCT3u3t2zdxb8BINRQul58uX7lopNriVo257w+WHvdI3HQYXel6FetTgYwFNVvtCnVo5uRZFBxPL0zK5StPVy5eiU5MqzkPfViNqlXZUVx/6lIL9GFefdSvX592794dQYjTp08Tn0OInsCRY4eo5c/NqGgFdypcLh81blmP9h3cE21BNgxy3nJVShIvDLhmw0r68PFDtOWMnQEGQmMTRX0gED0Bb5HFL/pslpFj5Li5xFtcQmhoqCx+9uRRevTgnqYqNswdPhDxR0qTQSvCdXz58kWmTJ88kti4x6Hh9z9Sxx4DZDy6D0UO3XyRpevmM+ex/5EzBulh1fq9VNKrFeXOV4zOnfsfffr0Se5z5y8p01eti563oW1FxePGjRs0efJkTRZLYKwR1sgRY9xDkYkUSmH31qxf59KRXSfCbcmTp4isWIJLN/SaNcb9oQ3P0Ha1yyhx3COShLf49JMxI36EhIRQT9+edPjfQ7Tt721Up1FtWrV4FQ3vO8KIrSTIqrxFr/yM3TPoI9ZEvUVJv1iXtryC3kJkP2OLjesv1kS9RUm/WJeOpCD0EQmY6JO9RRa/6LNFn0PfM3nevHnpl19+ib5wwsvhLbrkZ2i32KDX4qem5FmuIl0+fZ2O7TtDpYqXoXZdWkXrTfjf1ct0+twpOnvoAr3/8J76D+tDr2PhgWiorJHlg4EwMjJIN4jAqVOnaNu2bQblRSYNAX8RY1c3/ufuJzaLD37CQNi280iDDFRRdbZoiTK0d9cWTZYDe7YTpynh7Klj1KlVA+WQ2KDY/afvNcccmTZhBLG34bjh/ejooX107MgB2rVtQ7g8Sxf8TqdPHAmXZsjBP+JLu3l9bypX0IUGdG9DL1880xSbM3081apYmKqJt0XjRvQjfsDhcPW/i9SqSVWZPn3SSHpnIq8tb48StHj5NrJJWTJSPbBnVOuOw2nfvn00yHcIXbx4kebPn0+ZM2emQYMGyfTWnfyi9SQ0pC0NmEgi7du3p2HDhtH162GenrrZtmzZQgUKFKCUKVNSgwYN6O7duzLLokWLaPjw4cTlO3bsSHPmzKG+fftSq1atKEuWLDK+bt064gcZ3rZv365bteqOR/h2IL6HotJdXIXOkD4jZXHNGm7j1YkDblyjpm0ak3vp3FSvaU35YMJtXbz8L3Xq9RPNWTiLajSsTD93b0u79/8lxdi8YyM1aF5HGpc5wdevP+3au1OemzlvOlWp50U+tSvQaHEv8n3AG9e9ZuMqee7mrRt05dp/9EO772S+X3+fRO/evZXlTfVhyDVrrPtDuw+GtKudXzuOe0TS8BefJvm9dEyZglKnTS1Wonejzn0704Tfx9PmNZvp7u2wl1QXzl2gVvVbk2fBiuTb3Vd837+UAvHwM/Y25HT2dFmxcEVYesgXmj11DlUqVlmemzBiovCS/iTPXfvvGrVr0p7YU/H3STPF78A7mc4fkbWzcdVGmjHxdxrWexj59R+pyW/miL9oH/qwAn2Y+TqLrHlcf/g+kNdGZN+b+H42zfczQ7937x6tXLlS8udn9AULFtCYMWOkV+GAAQPo5cuw38jnz5/TiBEjZHqfPn3o2rVrsgx/HDlyhLp16ybPcdknT57Iczt27JD1TZgwIZzzgKageSP+onmDf/cePX4opS1csAglS5qMXDK6UK8ufaln5z7imSDMG3DPgV3yeZifvfn5+sGjB8TlhowaKMt27CX+4/RsL+OtO7agBw/vy+fohWKkGz+Ts5fhoaMHqXv/zlTcs5DcK6Oy/hX/gfn5ndPbdvmRTp45Iev5ZcpomY+dafi5vEufDjR7wUx5TvcDBkJdIjiOEYHAwED63//+F6MyyCwJtP3KYYTYh4rN7+uxxe3YuMFh8fKt0sgRF0NhjTqNafvG1RoGf21dT5ymhLdvXtGlC+eVQ/FW5WW4Yz5Rv0lzypItB33fsj3lF1/OD+7doZvXr2rKcOTvg3vp1s2Iw2n99+6gub9P1GzrVy7WlHv65BE1qVlO1FmUps9fLf4sPqcuwrDC4aowpiyeO536DB5DY6fMoS3rV9D+XVvluc6tG1KiRImo96BRdPTwPplmqo9Fs/lyEhdTJMamaX+soZF+w8nDw0Pmu3TpEh06dEjG+YPTR44cSdP+WKtJiyzCbblldYm0rcjKKel16tSh77//nnr27KkkafYsFw9jqFatGu3Zs4eCxbyNzZo1I36jef/+fRo9erR8UOE0Pv71118pe/bs0nDI8a5du8p+cH969OihqVfNET/lPloRtZE3tn04KAzlbNhTtmvinuDhCy3FW84kSZLQktnLqaKHN7Xq2JyeCc/dd+/f0U5hoF+1fjm1adGOnJycievgsM9/L5373xn69/I/so4Va5eRW7bsxG8+F/w5jwb08qXxI6fQRmGY3ys8gNmL8X//nhfXyjCqW7M+pU2dln7q1oYS2dtTvx6D6O/jh2PbrRiVi8/7Q1uw6NrVzqsdxz2ioREvv5clypWQDd4MuEHPnjyjZjV/oHwF89HU+b/SqxevqEebsO+q9SvWi/tlCY2ZNoZ+7vEzjR38izQqbly5gRbOXCjSfpJldm3ZRfN+C/Ou79a6O9knsqceg7rT8cPHNR2Lqp3HD5/Q7F9n06OHj6lWg5qaMiqIQB9WqA8VXHeKCLj+rPD6w/ezcvlHuzfZ/fH27Vu6cOGCFODp06e0ZMkScnV1pd69e9OVK1do06ZN8jmdHQ7evHkjnQD4RT2/7OSyQUFBNGnSJKpcuTKxIZANiatWrQpXHxsM+bwKg8FcXTNnoayu2cRUTm3kcy+/XP/0+RP16daf+GU9P3/zM3DFCl60XPyf5Dm4u/XtSKlSOlHz71oSj+7pJYyJXTuE/X/p3bWfTOPnaH7e7t6pF9nbJ6IfxRDm5MmSi+ftSbRFTK3FRkcO3ft3ImfxzL549jLKmT0XDR45QOqlWaPmMt/6LWvFs/0K8hfP9I3qNdGL2l5vKhJBAARMTcBfNBAoNjexcWDLDm8jxeYnNosKbNxgoxQHNhRyCLz9QBiQMtHwFE7y2JAPn2q1xRfdAHos3pQkTpqU2GA3eNQUmWZIec6TPWdeSpY8OeXMk5+chSEiJuGG+NJmY5QSXr96oURpm/B+ypjJlYaOmSrnykubLj1Vr1CAnog3Pnb2djRz0XoqUrw0PXxwl1yEp9Z/l/4Rw3gL0B3hLbVu51EpSza3XNSgaklNncaOsLcSG+0Cb9+XVbNO5Cb0w4bcrTsP0rlfpkfZ7I8//kjFivHQX98o83Fb2kG3Le1zkcWnTJkivf42btxI+fLl02RbtmwZlS1blqZOnSrTeChy/vz56c6dO/I4ffr0xG8bbW1tyd/fnypUqCANgnxy6NCh1KVLF2lQzJ07Ny1cuFA+lLCRVs1B8SIMvBWmu8XCUKjobki/n+Ms+vI1S8XDR0pNPZ1/6k73xHB+fmu5b+sh+fBRSly/C5fOpyPCYJcpg4vMu+gPYfzLmp2SiYeQqTMnyTSeW6WgeyE6e/4M2drYUhpxn+XJlZeuC6P7vBkLqWih4vRQ1Ovq4kqXr16iyj5VZbmRwoD+XYOmxB6Et+/eos2rdlBq59Sy/lrCy9bUIT7vD+2+RNeudl7dOO4RScRffAaKzU1sHEzye5k4cWJZ+aePn8WLqh2UIVMG8h0zSH7fp0mXluoI77+nj5/S+hUbhMd8G/KqEjZn7vt37+nh/Qe0dtk68cDekn5o+4Osp0u/LjRLeBrW+64e3b11l1btXCl+B5wpq1tWalL1O5knqnY4A3s4zl7+h/yukwXU8eEvxAgUm5vYOEAfYRzM9ekvGg4Um5vYOJhEH2FVq+LTX0gRKDY3sXEwSX/xfRAG14BPf5EnUGxuYuMAfYRxMNenv2g4UGxuYuNgEn1wxYULF6Y2bdpwlK5evUq3b9+We55TfM2aNZQxY0YqUaKEfF4/c+aM+G9RTI7yqVixovQ2zJpV/FcSeZXg7OxMEydOVNvvnSKev4gEis1NbBwi5WpvZ08bV2yl9ZvX0kbh6LJo2XxZ4KdWHWhI/xEyrViREjR84CiZPlSkVarjSc/FqLT8ed0pSeIkxOdfvgrzyCxauJh0NOHM/BxdoaynHMFz/NRRGjbATz6/V/GuRoG3bsr6enftT5UqVpbPLgXyFaR5i2fLl/k5suekgb0H04hfhsp8Y4aKuejTZZBx3Y8YGQh5UnmeE0dfOHbsmPxDp3vO19dXujGywhESBgE2oDRu3Fh2hucve//+PZ08eVIe8x/6HDlyJIyOfutF6LeoyWPKF44cbmjy1kzYgGIo9PrpByqQrqhBLaVNl5HKevjQPuF9l0y8QeHhxRmFkSGyoG3MiyyPks7Dk7u2Dbtunz19TCeOHqTJYwbLdvadCPMwbNexFzUTX+BK4HkMG1YtJQ9vB96Qxr+8mcIbml48f0aOwvAyZ/oEalbXk5KncJT5vavUolPC0OKWM4/GUJlXGFViGmya945pkQj5FeMdG8nYuMYLIHD48OGDfMu3a1fYW6e9e/dSoUKFxA/3a4OvP6dUjuJH7I2mTaWt1KmjNwzzm0c2/nXq1Im2bt2qqYOHHZcrV05znC1bNhlXhiJUqVIl3ANErly5NHmdnJzkQwsnODg4yHR2pY/KQJg8g4emvFoiiqFQ4dmn2XCqSd6xFm/u9IVUsljYtaxUwkbD3OL61J6LML8wrPObYTYQcjobBzmUK11evJW8IT0B+bh183a0Z/8u+Vayik81+SDCRkQeYtyoRT1NnZW8qnB2GYoVLi73J08fpxxuOaRxkBPyiTbjGt5/+GTwNavblsLYFPeHblvax0q72mm6cXPeI3xf87B3EwVj/K7K38vACw+Iwl/asRL5xfOwF0K58+WiE0dOiPlwH1HBTOG/sznPzYCb1K5rW00bTVuHTXPB6R17ffv9cM2aWdZx+vgZ8TvgJo2DXCiPe15N2TuBdyJthzOVq1gu3HedpqDhEWNwNrQ1qQ9DM0eXD/qIjlC0542qj2hb058h3q8/fB/oV8TX1HjXR5TSxOAkvg9iAEt/VqN/H/D0REpILhwzeFQKj+rhwCOEtAMPP04qHD/YkMjzGLLtIF26dNKIqORjYyK/+I9BiM/rOTKxInBl20hSMbS4Y7sucnsi/nOuWreCJs+YQGXFs/StO4FUQhgAlZBZOJ9weCb+T0YXMn/9T8wvMLSf3/k45EvY1FZsWKxYs7wcDcSejNqhbcufaMLUX8QzfiZqWDfsP7H2eSUeIwPh0aNHpbGPC/NFMW/ePKpVq5asK21a/Z46PCRMWThAaTSB7v1Ev3hL8MHOzk66DXNH2XuH5wZr2bKl7LdieEhgEGxM1B99X2wjRVt+YgsNfX3aRM0aXq1P7Y7S80zXW0y3Bl5kgf/sKoG92NyyZZJlvc5dp6fKCQP2Nes2oR2b10hDW6364X9guHhQ0GdNLQ/u3dbEo4vkzJOPpsxaKrNN+WUIlSzjQV6Vaxr8Y5QylRMVKlqSlm3cL+sIFq7y18T8gtlz5qZJo30pUMzldkBMRptZfBn3FG7fHLII40qg8EoMEUZ1OzGckr0LYxpCVwgvOvFmydCg7w89e3iyhxqvWvzo0SNavny5rG7t2rUUEBBA/CKHAxvieI4RJ6eU9Ph6mNFQnojkg4eTK4ZgJYu3ZwnZVsufhitJUe67d+9Oc+fOlcODlYxspDx//rxyKGXkg4IFC9Jff4XNg6c5KSL2gq124Ln1YhLePToi5glJEpMi4fIqC1Ec2D4nXHpMDyLTHXsQLhgfENPqos2fK0duui+8ddmAyt/r/Ht9WQwTHtxvmEj7Eq58urTpKX8ed5q76A/yEBMvs7chDxn+LO7HJl/v0/lLZgvvwJt0dM8p4ocYnt9EOyh64bkQbwiDOw+r4Det7G0Y15AsaWIy5DszMsamuD+0+xRVu/rOaZc11z3Cxv/3jw5ri2KU+Nf+xuwmDZuKQ7d9+XvpViiTvt9S3bzRHv8lhgRzcM2WhVKKvhcsWpCWbFws04LE6sYBYh5BNvTlK5BPeLmHzZvEJ/du30uZXDORe2F3aeyTBcRH4PVbVMGnArGhMPB6oPgdEPeZ8DZ/9OChkiXKdjSZ4haJKWdDW9PHXOpDVKDvnKH1avJBHxoUhkT0MTeqPgwRQk+eeL/+8H2gRwvfkuJdH6JpfdfmN4kMjOH7wEBQYdn0MTf694E+Y16KFCmkBJs3b5YGQT4IDAyUhkB2Jvvzzz+l/YhH+HAedkyIQzDV9RyZSAZx3bRtPU2YNo7OHr4g6+HnZx4WvHzNn3T7zm3Klzs/XfzvX00bgbdvynje3PnozPlTmnR9ETtbO33JmjQ2Pg4b40u/itFidWrUoxu3rss5C5UMK9Ysky/wefTQX2KUXs2qtZVT4fYxMtOy8cfFxUVuXAsbBZVj/tOmb0J57db++ecf6RWiLJ29ePFiKlKkCLFnIk9myd5APMFi6dKl5USVXB+f40n0lcBj1nk8O09Gz5Nb6ltlR8kbz/sRoj1/sfmJLUEH/pNXsqRY8VRsbFDgNwDKMb9BQDCIgJ/IFeY7HJadv7j5i84v7NCyPhXjIBsG2UDEc26xsSQ6w6K+XlauXld69+3fvY04rh0yZMxM7P3Hnnm8wvHKJd+Mktr52OjwSswRqB1SpXKm8sLlmrd0Yg6IPPkLyjh7LBoS2KB44fxp+lfMvcb3wBLhKdW1XRM5xJKNgwWLlJTGQfY6PCTmgeAXIyXKVJCGzpV/zhHeeu9p5eK4GZCik5MNtfynXglsGGSjCRs/ONSt6U3Lli4m9sDjjYft8sse5ZgXBFn652KZT6kjsj23pW0cZL0f2DEnxnpnL7/Zs2eT4sXI7fGLJ/6d4MmMOaxYsYLq1q1LypAfmZjAPqLTnSm6W6Jo2BtMnouEjXXKHINFChXT2xx7A277awuVFp692cU8nzwMwv/wfuldyAXYOFi4QBFpHORJknl+EzY+6gZezY29E5ev/pN4tbelq5boZjHJcXSMjXl/aHcguna18+qL4x6Rv4tG/718LOb2uykMd1cvX5MLjYwZNIaGTxgujOW24ru7BP17/l/xfS8e4sX3/bJ5S6lHu57y+75Krcq0Zc0WunfnHgVcCaCe7XtRCscUVLlmZdq0WixycuuunNZghxim7OHjQcXLFBe/A8lp9Z+r6eOHj2K15G/z7EbVjr5rQSVpfkIO6EMlyhBimEQfq1evpsOHjf+CwAjYTNJffB/EWjPQR6zRmaSgUfTx8OFDuYggj+jhjZ0HDAl58uSR2fiZnl88nzt3jjp06EAvXrygBw/E1FNubsSjfl69eiUXEbQgJzKDuRYXziQ8lzcvKMLzLnIf+VmZjXL58+YnH/E/lBcYOXU2bPTl5u0biYcIK6OeFM6KAfZVDFYxfvEybCSERznPsP+qyxfK6r6EfpGei6PEM844v4lywZQhowZphjErbSp7eyUSl70yoXyvXmJ4npg0nieQ5732DwtPXskTxvfv319OPL9z505q27YtTZ8+Xf5J5ThfSP369SNeGZcvnBkzZsgx7D///LP0UOPFMHgCerY4szWaLziexJ4NiioII4UMbCT00pLFTyueYKOKV0iC7aBpOsbXCge+bvw4YqmB//xyUDzHYmMUVK4h3qcXqz0VFy7Yn8TwV15sRFnxidvIJYYiVhBGihYNfKThrZww7l0RC4Rw4LLKl6mHT3UaPqALpRLzm4Wdi/guRGlTZojmg/N6eFeltp16y7Y5exrxRmiqWNiBPQNbifnceDXljeLtUOo06ajh961o5q9jqHrtRtTHdzSNGtxTbtwvHoIck7ajES3cacVQq3gMhjspDnp1/o5KerUiD09vzUIlCjPOywa5EX6j6PTBP3WLRjjmOSY5xEXvSqU+Pj7UvHlzaQjkNJ6npHbt2uTp6UmOjo7yLaQyBFlbz0p53b0uX91j3fxqOI5Od7GV0Ua+d9Bfmic5HiTmI+EJjMdOHiVXEp40+lfp1aePWYWyHnIIcYliJWWFXh7edDXgCqUV1zwHHrrAK6etEx7AaZzTSM/C6bOnUvXKNeR5RRYeyjug5yAaPnaI3HjosxzmLO4zU4boGBvz/tDuR3TtaueNLG7l94jRfy/5eXPub/PkxsyLlChCA8Xqgcpw4fLe5al1p9bUukEbqRKeC3Dy7EnSA7BO4zq0dd02qlaqujT8derdkbLlyEY1hRFwg1iopHqZsOudPRDrNKotpzfo6dtDLmbCC5oUK11MluN7LKp2+LytrWnvCdm5mH9AHzFnZsoSRtcHC8v/wfh/G/8OqywYvb/4PoiThqGPOOEzemGj6INXKOZNCewIxM/pUQX+zUqVKhX5+fnJbdasWTI722t4IUF+nuc5x/n5nkPDhg3liCZeETns9y7ifzWZUR0fBnPlob8Txfz5A4b3pZHjw0ZT8ZBeXsCvvHA2Yee2yuK/bJMf68tnX37ZvnBW2P8u5uDwdc70lI4p5dRAVet706G/jkVLgcsWEQtp8sv80j5hL/rbt/pZzhM+cHg/Ocdh2VLlqV7NBnIOQ54fkVc2Zll1Az95CDlDddOjPWYhWMkNGjSgwYMH04EDB4hdRznwhJPsmXLr1i3ii4MNiCdOnKCmTZtKgyDnqVevnpwLS/EO5Ink2TuQLc3shbZ//37iB2JeBYe9Wi5fvixXzuEx7fv27SOe4JLrZe+1TJkycZUGBZZbBFM9cemCHPlVKD9xc4QqE+4bJCgyqYIAr8wkhj8a+3rx+9o5Za+vrxYzxDi6YcihM9bQUzEH4ct6EYcL6+u4IWm8MIiTMP4lSuQQafZXL58Lr45U8sVDpJliceL5syf09MljyiYme00svtSV8FmsUMWLq2TO4iZ/6DhfKqfUsn32dnwj3gCxwfPrd5BSLMp9xt/HkqO7WFjCgCHGiqFW8RaMrOJVG/ZT6w5DaaTfMPqxVVvpQchvBtlzkI2DS+aMomaNq0RWXKZzWzysltuKzCDsmq82+R88It8URllZFCd5LhP+DWBPct1hxFEUi9EpfmP38u4Bsw4xjk53wUGhcohxzTbeMeqboZlfv3lND8S1mzVLNkqaJKmhxfTm4zloHon7k1dx42ud51RxEsPz+Q+YbnguvHzfvH0tV3uLyX2hW8+Htx/pyOaT1H5QTt1TmuPoGCsZjXF/KHXx3pB2echtbJ7DuH5T3SM3b96kCuXL0v2rO7gZo4YYDjH2+9q4so8gy6itHUIblwpb7S/CyTgmPH/2Qqxo/JSyZs8awYOZ5yhkwyEbvJXAQ/Pv3b5LiRwSiflzM4b7vuc5tN6+fiuGMLuGS+eyUbWj1B3T/fpT02l43fh/foE+9GvKXPoQ0sTqf57+XhieaqL/W35fJVD2EQTC9RcBiUww1/UHfahLH/Flj+ARoewxyKNN2TCoBH7W4ecWXsCEnwvfvXsnn+9jOkLIRP/PFTF1935fE5S97nma1XdFaO12PuHSeYTM7Tu3KIUYMaPMHaidgZ+Vuf9Zs2aTL+a1z2nH2YMwVcpU2knRxnn6IH72TibmQuQ5EXkVZTY4Rhe2LzxAXaY0tzGKB2F0E8ornh/s9acE9ijkdG3rtPYFxEOIOShp3LlGjRpR69at5fLXnM5eh+PGjVOqjMle15AXVVn/qE5Gc06xNkeTDaetjIBfQupvXOddiw0LHiIcXWDjnCkCewjyphscHBKLuaaya5K18/BqyjFdUVlTkYGR6AyDSjXNGlWiAnn/pGl/rBWeekXkgiQ852Ddml7Sc7BQgVxK1kj33FZ8fLnxFBbWEAzVnalY8EODIQ8OhrTPBleeY1AJaVKnUaIR9ryCMW/xEQxlbIz7Q7s/hrarXSYmcSu4R/xiwsPYeVOncRbf9856q+VVjnUDD09mY6K+wCsY86YvRNWOvvxmTPMzY9tSF9BHOA2YVR/hJImfA7P2N6r7FN8H8XMBaLcCfWjTkHG/CClmSkiSJIn0GtRtnl8c6C5uoptHhcd+sZGJX7jzvIKRhQwG/JflsjE1DnIZFzEaTwlsfI2pAdYoBsKoJpTnSfDr168vhx3zcthsFOS5pFKnTk0DBw7UTEzPY9N5vLsS9Hkb8GIY48ePJ/bE40ktu3XrJiet5yHIMQwx8QjzjkHdunl5jpZAsfmLLT7+T4tmEEAABEAgagJsBFzwu6/I5CtXKTVkQZKoa8RZEEg4BHB/JBxdoicgAAIgAAIgAAIgAAKGEzDKYO/oJpTnySjZiMjuoF26dJFLW/Ok+DxEmd1P3759S127dqXJkydHKfmmTZuoTZs2Mk/jxo3lYiXswmri4C/qN2TzFvkOio0DGwbPi62h2HgQuL/YEEAABEBAVQSU4Y/KXlXCQRgQMDMB5b5Q9mYWB82DAAiAAAiAAAiAAAiAgEkJGMVAqD2hPM8XyPMJDh06VCO4Mgn+sGHD5OIjEydOpL59+5Krq6ucn4pXR+bx6DwHYWSBXVJ54ZOLFy9K70MnJye5KkyLFi0iKxLf6ewh6CU2GAbjmzzaAwEQiBUBzeIJXxeaiVUlKAQCCZQA7o8Eqlh0CwRAAARAAARAAARAQC+BWA8x1p5Qm413K1eupClTpkSYUJ6HBCvB2dmZXr9+rRzKYcK8kAnPL8jLYn+d0DbCZN3abQUEBNCNGzeI6+IJLlUS/IQc/mIb+XUvdgggAAIgoF4Cul5RfGzq+dLUSwOSgUB4Arg/wvPAEQiAAAiAAAiAAAiAQMInYBQPQgUTT5adN29eg1ebZIOgm5ubLKMYB5W6ItvzJIu8QrKKjIMsqr/YfL7uxQ4BBEDA2ASePX1MeTLaU6dWDcJVfe/uLdq5dZ0mbc+OTXQ78LrmOKYR7fo+fHgv27x1MyCm1ag+v+IdpQiqe6ykY2/ZBG7dCaRsBTJR38E9w3Vk1brlVK9pzXBp+g527d1JXIexg//h/VSjYWVjV2u0+nTvB91jozWEikAABEAABEAABEAABEBAJQSMaiBUSZ/MIYa/ORpFmyBgTQT+2raekqdwpP27t9HzZ080Xb9y6QJNHjNYczx98ki6eOGc5jimEe36EidOQss3HaD0WqtBxbQ+NebX9Y5SZIwsXTmPveUSWLd5DZ04fVzTgVBNLOrI1JmT6V9xj1lTiOw+iCzdmtigryAAAiAAAiAAAiAAAgmXAAyECVe36BkIJCgC65YvpP5Dx1GatOmJvQQ53L19k8YO60N3bt2g7j99T9MmjCA28I0b3o+OHtpHQUGfpfGwcpk81LBqKdq+abUs9+nTR2pcoyytXbGQalUsTHx+7fIFEeoLDg6iqePE3Kkvn1NISAjNmjqWKhZzo3IFXWjciH708eMHWd+QPh1o+aJZ9GOjyvLcKOGtxfnVGtgbyimVIxUtlCeciPCSCocjQR00bdycBo3oK+6JIL39WivuDfbo86xeln79fRIFhwTT5BkT6PLVSzRa3Ffb/toiPQ7v3b8ry4+Z6Cfq6yfjnz9/pgbN69Cdu7cp4MY1atqmMbmXzi3znz53Sua5ePlf6tTrJ5qzcFYEz8HXb15T+66tadb8GXpli+9E3B/xTRztgQAIgAAIgAAIgAAIqIEADIRq0AJkAAEQiJLA9WuXpVdgjbqNqVHT1rRx7VKZP026DPR9i3bSaNippy/Vb9KcsmTLQd+3bE/5Cxah3yb4yeHH/YeNo9Y/96DenVrQiaMH5QJHF86fpnnCO2rwqClUtoI3DenbkRxTpgpXHxv5Tp84Qh8/fKD1KxfJ/B17DKDp81fTzi3raPZvYXOsXg/4j0b69qAmzdtRxx4DadnCmbJclJ0y00n2gmLD4MaVk+nc3yukFKGvT5OfbwcZh5eUmRRj4ma7d+hJH8R1vGjZ/Agt8XDffkN6UTNx/4zzm0hrKfiHdwAAQABJREFUNqykmXOnU6M6TSirazb64buWVL6MBz1+8ojOiPuG5wVeIYYorxQbGxwvXPofXQ24QmnTpKOWPzWlJEmS0JLZy6mihze16ticnj1/Su/ev6Ode7bTqvXLqY24Z5XwXgzjb9u5pTRI/tSqo5Jstj3uD7OhR8MgAAIgAAIgAAIgAAJmJgADoZkVgOZBAASiJ7Bt42qqVK0OpRYGiKq1GtDZk0fpnpgXLWnSZJQ7X0FKljwFFShUjLLnzCviySlnnvyUyik1zf19IvXoP4Jq1GlMDb7/kRoIQ8fu7Rs1DY6aOIs8vKtKIyEn8jyH2vVpMorIamFYYSNji7ZdqFRZT+rebxhtWLVEk4Xbqd+kBbURhphCRUvSbZXOW+jlWUIaBr09Smhk5wgvUMKGQoSESSBZsuQ0augvNHbyKLr/4F64Ti4V1zF7GLZp0Z48ylWkPt360yYxpD9H9pyUXJTLlSM3pXZOTZXFvcIGwkDhuZtEDL9PkzotXfzvXzp15iRVr1yDTpw5Tg8ePaBZU+ZQqeKlqV/3gbKdI8cPa9pb9McyYYhsIY/ZaNihRztpZJw9bT45ODho8pkrgvvDXOTRLgiAAAiAAAiAAAiAgLkJ2JtbALQPAiAAAlERYC++VX/OlcN5eShwkBjOyOGvreupfZe+kRZ9+vihPNe/W2viTQm16n2nRCmTi6uM89yGHD5/CqtbHuh83BAeUp17fZvr0DVrdnr4IGy4JWd1cc2qKeEkjCmfP3/SHKspomsY1JUNKxnrEkk4x9Uq1aBKXlVolBgy7OXB62qFhRuBAbTXfzetXh/mUcqpyYXRXTd4lveiaTOnCA/UYqK8N9na2NJZYTA8duooNajdiHj4ce6cecKVzS+M9U+fPqVMGVxkupu4b5RwWywwxBsbGkOCg5Vks+5xf5gVPxoHARAAARAAARAAARAwIwEYCM0IH02DAAhET4CH+LJn38JVOzUrpG8VQyDXC6+nqAyEPFyYw+w/N1E5z0oy/vD+HbK1tZNx/rCz+xbXJEYSKVi4OD3SMgjevH6VPH2qaXLb2eHrVAMDEdUSGDFoFHnVLE/vhfeeElKlcqbO7btRzy59ZNKrV6/oibjndEOZkuXknIR7DuwiNhby/cOrHPMQ5Qkjp8jVju8/vC/n3+RzPBT58tXLNFh424aEfNGtThoGd23cR3W/r0HT/viVhgovXAQQAAEQAAEQAAEQAAEQAAHzEMAQY/NwR6sgAAIGEtgi5jmrXqeRHApcVng98da8bWcKEIsnXBULH9ja2tL7d281Hkj2wlD36sVzSiqGRhYtUYa4PHsd8pDk1k2q0/EjB6JsWbc+JXOVmvVpw+o/5YIovPjJto2rhIGwunIaexCwCALswcdDiA/+7a+R16OcJ+3at0POMciGw2FjfGnu4j/keTt7O3r56qWMOzs5U5GCRWn7rq1UvEgJOYyYPQ/ZazBjhoxUomjYsPVVwhORFzk5+PVeKyI8DvWF9GLBoXRi46HP8xbPlsZHffmQBgIgAAIgEL8EVq9eTYcPf5seIn5bR2sgAAIgAALmIgADobnIo10QAIFoCXwQCxjwSsO1GzQNl9ddGCkyZnKlXds3EMc51PQqLPcewmg3fEAXuTjJ6Emz6eypY1Qyb1q5WnGhYiWpcbM2Mp++Dxsbmwj1cT5Or1X/ezlsmIc5F8iSjHgl5LqNftBXDdJAQNUEOggDe6YMmTQydmjTWRj4XKhijXJU3LMQPRLD8337DJXnvSr40OCRA6RRkBN8KlaWQ4XZKMjGRh4ezHMTcrC3T0SDeg+W+QuXy0+txaJAfsJjkY32fA9FFnjoM3skDh01KLIsSAcBEAABEIhHAjNmzKDt27fHY4toCgRAAARAQA0EMCZODVqADCAAAnoJ8CIkVx9GnJuMjQ2HzgVqyhw5f1vOUcgJfQePoZ/EUMkUjqnkEMgDpwLolphjjRcyYaMiB/tEiSLUq92OUp9u+5v2nKa7YoGGRGIxhUwuWTRGj1VbDsl6lY8FK3coUavep06d2qr7r4bOZ8viRrcuPggnStIkSen4/rOatJSOKWnFgjVyDsHPwts2u1sOzbU9oJcvsUHR8es8nb279iPelHD28AUlKvetfmhLDcSiQA/EUOOsWbIRt8WBFy25dPKajPOHtxj2z5sSls1bpUSxVwEBZ2dnFUgBEUAABMxF4MiRI+ZqGu2CAAiAAAiYkQAMhGaEv2T4ZjO2jqb1EWg9qr6+ZKSpnICdvT0pC42wqLyCsRL4XI5c+ZRDg/a69SmFeF61bNlzKYfYR0Hgw8dPlCZNmihyqOOUnR0c6VkTbHR3zZxFr1KcUjnpTY8skQ2OvCEYTiAqD0vDazFuTmensHlcjVsragMBEAABEAABEAABEFArARgIzayZ2u2+rSRpZlGsvvntC6Oem87qAQEACMSAwLPAvTHIbd6sQSpZQde8FNC6uQiEvj5trqbRLgiAAAiAAAiAAAiAAAhoCMB1QoMCERAAARAAAWskkEh4mSKAAAiAAAiAAAiAAAiAAAiAgDUTgIHQmrWPvoMACIAACIAACIAACIAACIAACIAACIAACFg9ARgIrf4SAAAQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAAFrJgADoTVrH30HARAAARAAARAAARAAARAAARAAARAAARCwegIwEFr9JQAAIAACIAACIAACIAACIAACIAACIAACIAAC1kwABkJr1j76DgIgAAIgAAIgAAIgAAIgAAIgAAIgAAIgYPUEYCC0+ksAAEAABEAABEAABEAABEAABEAABEAABEAABKyZAAyE1qx99B0E4olAcNr08dQSmgEBELA2Aokc7K2ty+gvCIAACIAACIAACIAACBidAJ6qjY4UFYIACIQj8PQFZRo7MFwSDgwjEJI6LY28bEMj6ngaVgC5EiyBL19CafvCAwm2f3HtmJ093nfGlSHKgwAIgAAIgAAIgAAIWDcBGAitW//oPQiYlID/kTPkPbKjSdtIyJXbcefGzSWblCXJz7cDjRAbgvURsLWzoc4j8lhfx9FjEAABEAABEAABEAABEACBeCOAV+7xhhoNgYB1ERgpDFsHD5+xrk6boLdsFAx9fZq8PEuYoHZUaQkEbPFLbQlqgowgAAIgAAIgAAIgAAIgYNEE4EFo0eqD8CCgTgJsHPQTGxu2EIxDwNvjm4GQPTM5aKcZpxXUAgIgAAIgAAIgAAIgAAIgAAIgYI0E4JdgjVpHn0HAxATY6+3AjjkmbsV6q2fPTJ9aHYkNsQggAAIgAAIgAAIgAAIgAAIgAAIgEFcCMBDGlSDKm4VASEgI8YagXgLwbjOdbjDs2HRsUTMIgAAIgAAIgAAIgAAIgAAIWCMBGAgtVOt/Hz9M2QpkCre179qajp38O1Y9+vDxg6wr8PZN+vz5M63ZsJI4TTs9VhXrFNInt9KPR48f6uSO/HD42ME0Y860yDPgjFkIKENfzdK4FTaqbYRl9vAotMKLAF02KQFvzP1pUr6oHARAwLQEcuTIcdPf39+0jaB2EAABEACBBEMAcxBaqCpDQ0Ol5KcP/o9sxQz279+/p+nCYNa1byc66X+O7O1iptrEDolpzZKNlCF9Rnr/4T31H9aHvDx8KF3a9Jp0Y6BS5D6y60SE6tKmSRchDQmWQ4ANVDzsFfMOmkdnPOyY533kgNWO41cHbKj1wYI88Qs9HlrDC494gIwmQAAETErg7du3b0zagJ7KR44cSZ6enqsPHz6s5yySQAAEQAAE1EwAHoRq1o4BsqVJnZZ4y+Kaldr9+BM9e/6Url2/Koffzpg9jcpWKk7FPQvR6Akj6OPHj7LGXXt3UtM2jcm9dG7q7dud+NkhODiYJk8fTy9fvaSOPdvLfK07tqBHTx5p0jlxz4FdVKWelyz7c/e29ODRA5l3+ZqlNGXGROo1qJtsr3WnFkKWZ/Kcvg+WV3ezs7MjrmfMRD8pF8vO8R27t5FP7Qpy239wr6a6qwFX6LtWDaQsPQZ0pVevX2nOIRL/BOTCJGLuQQTzEFCGHZundbTKnmYwKCWs64CN7tpeugmrd+gNCICANRBwcnLy9fHxofjyIuR2/Pz8KCAg4L418EUfQQAEQCChEYCB0MI1qgwDvnf/Li1f/ac0FubJmZfWbFxFfyycSV1+7k5/TJ1H23dtpd/n/UbPXzynDj3bUasf2tDc3xbQ5f8u0cr1KyjkSwidPHNCGBE/UNcOPSSV3l37UfJkyTXpbHj8qVsbqljBi5bPX03BIcHUra/wGBPejDw8ePrsqZQ9Ww4aM2wcXbj4Dy1bvSRSupt3bCTt7fipozIv1zNvyRzKkjkr9ezSV8aHjval3l37U8niZWjEL0M1dXKfPMt70dhh4+nIsUM0cdovmnOIxD8BNlDBcy3+ueu2qK0DDDvWpYNjEAABEAABELAeAlevXt3h7u4+l42E7NlnqsCGQd64nezZs8948OBBH1O1hXpBAARAAARMRyBm41BNJwdqjiUB91K5NCXz53GnCaMmE3virVy7THoUtvqhrTzfSxjbps2aQs2btJTHT589IW/PSrRw1p8az0KlosIFisho0cLFKFGiREoybdy6nooVKUHDB46SaUP7j6BKdTzpwcP78rh0iTLUs3PY88C/ly7Q9ZvXNWV1I3MX/REuqZQw/pUtVV6mlSxWivp06y/jk34bL42Z9Wo1kMZHnhsxODhInmNZenTqLeNvhBfk+KljaezwCfIYH/FPAJ428c88uhYx7Dg6QsY7z4ZZ9qL13o7Vu41H1bw18ZB9TJlgXh2gdRAAgbgTuHTpUkdHR8fnhw4dymVjY9Mk7jVGrCFXrlw3xXDmd15eXqMOHjy4NmIOpIAACIAACFgCARgILUFLUch4cOdREj/2cs7BzC6umpwBNwOoW8eemmMezsvDgV0yZSY27A0fO0RulbyqCIOfYW8Ub90JpBLCKKeEzJnC2lOGEmfL4qacIvEgQh8/fdAc60a2r92tm6Q5dsuaXRNPlTIlseGTg2KsDAn5Io89ynnKPX+45ytA7969lUOr2UCKEH8E2Cii7bUWfy2jpegIsF4Uw1VkebEIQ2RkYpbOBnL+JmWvTRjLY8ZOjbkxZYIatQKZQAAEYkvgzZs3vvv3749t8WjLiSHFMs/Dh4YvOBhtpcgAAiAAAiAQ7wRsxepW19kl3BoC95P7m5D6mtU1G7FhTts4yP0r5F6YHn6dH5CPbwTeIK8K3vTi5QuqWqk6XTt3i1YtWifnH5w6cwpniTbky52f7j24p8nHKx5zyJs7n9zbGskwZ2evY7cWBlB94YUYLq2Em7duEHswwjioEDHOng0d7IUWWeA/0QjqJ6BrwFX0hjnzjKs75swL9SBYNgFpHMSLD8tWIqQHARAAARAAARAAARCIMQHbd+/evY9xKQsuIFb7DVupw4L7YIjo1SvXoHWb1tDtu7coKCiItog5/7w8KhHPVVirSTV6+PgBlStdQW5v3r4OVyWvisxBd9EPn4qV6dDRg3Tq7El5fvP2jVTFuxo5ODjI45h8XL56iXQ3ZREVQ+vZuG0DsZGS+7Rq3XJi+RCMS8ArioUXFCOTrvHJuBKgNmMTYKMgD51k/WERBuPSZYO6nzAS2qQsKfkat3bUZmoCfG/41O4ovUAP7MBQcVPzRv0gAAIgAAIgAAIgAALqImBfpEiRXwYNGjT/+PHjydUlmvGl2blz54vkyZMfMH7N8V8jDyuOKtStWZ/WbFhFntXLymxFChalBnUaykVMvD18ZDqvfsxDeCePnaapiutN6ZiSeB7AqvW96dBfx+Q5Ti+QvyBVFkOSm/xYn5InT0FJEieRcxhqCutEbG0iroGjyF2jYURj3l8b9+nUEHao21WljlzZc5FXzbB5C3kYcpP63+stj0TTEIBh0DRcTV0rG7F4XjXFS4oNWgjGI8D3BW/Mlw2FCl82tmPosfE4G6MmbQ9a1pe/8JZmfeG7zRh0UQcIgAAIgAAIgAAIgIClEbDfvXv3KiH0Sh5+6+3tbWnyx0jeiRMnOosC3WNUSKWZK5T1pFsXH0QqXbq06Wn7ut10595tckjkQJkyusi5CrnArF/nSq+7z0Gf5cIfSiXa9a1ftkV6EKZKmSpcOzMm/SHnMBSep5Q1azY59yGXVxYVUerq3L6bEg23j05uZb5BpdCB7X8rUcqXJ79GFmUxkk+fPtGTp4/lEGvFcKgpgEicCbBBA/OqxRmjKitgIwh7EirGEDaQKHFVCmxhQumyZL4+UQzXt7DuJQhxlfk3+XuO9TXCl2DETRCaRSdAAARAAARAAARAAARiQ0BO9la4cOH5AwcObHrixAnH2FRiCWV69ep1x93dfYdYycsSxDWKjDwfn/aCH9qV6s5ZqH1OibNxUF/IkD6jvmSzpCVOnJhcM2cxS9vW0ij/cWbjhrI6K8fhDWX52mc9Kt5t7EklvalEmq5hy/J7ar4eaLMcYT4x0DIIgAAIgAAIgAAIgAAIgAAIREtAjgH9559/fr4jghhq/DLaEhaYYfDgwe9EH08K42AnCxQfIoOAWQkowyLZoMQbG5KUNLMKhsZjTUAaB7WMgazPA9vD5lzjcwggAAIgAAIgAAIgAAIgAAIgAALWRUAzSdyDBw8KLFu27EHv3r3vJxQEPGy6VKlSr8T8in8dOHCgSULpF/oBAvFNQDEe8ZBUBMslwMbdqBZhCBtm+W1OQulVaLndheQgAAIgAAIgAAIgAAIgAAIgAAIGEpBDjJW89+7dc9+2bdvMadOmdfH19X1TrVo1OeTYkuYmZKMgB+E1+OHYsWNJxbDiVadPn4bnoKSCDxCIPQFluKSyAi7XxEONOcCjUGJQ3Ye2gU96f8ZgEQYuy6sc84Yh5apTLQQCARAAARAAARAAARAAARAAAaMSCGcg5JoDAgK6il3XpUuXTlu8eHHVpEmTJr5x40ZOo7Zqwspy5Mhx/f379x/Lli07QTSz1JrmHDQhVlQNApKA4mGmDENl45H0ShN7BPURiMsiDGz05Y11zXqGEVh9+oVEIAACIAACIAACIAACIAACIGAsAhEMhErFd+/e7aXELWkvjJlS3E2bNlmS2JAVBFRLQDEGKh6ELGi4uGolh2DGIKCta66PDcIwFhqDLOoAARAAARAAARAAARAAARAAAfUQ0MxBqB6RIAkIgICaCGDeQTVpw/yysDchz2OoGI7NLxEkAAEQAAEQAAEQAAEQAAEQAAEQiCsBGAjjShDlQSABE2BvMT/fDuE8BhNwd9E1AwiwRyF7EPK1gQACIAACIAACIAACIAACIAACIJAwCEQ6xDhhdA+9AAEQiAsBNgRhOGlcCCbMsmwkHKHVNQw71oKBKAiAAAiAAAiAAAiAAAiAAAhYIAF4EFqg0iAyCIAACKiJAA87tklZEsOO1aQUyAICIAACIAACIAACIAACIAACMSAAA2EMYCErCFgLAZ5jDkNIrUXbce8nexTyUHTMVxl3lqgBBEAABEAABEAABEAABEAABMxBAAZCc1BHmyCgYgK8+ASGFqtYQSoVjY2Eoa9Pq1Q6iAUCIAACIAACIAACIAACIAACIBAVARgIo6KDcyBgpQTY2IMAAnEhwB6oGHYcF4IoCwIgAAIgAAIgAAIgAAIgAALxRwAGwvhjjZZAwCIIwDhoEWpSvZDshYphx6pXEwQEARAAARAAARAAARAAARAAAUkABkJcCCAAAiAAAiYhgGHHJsGKSkEABEAABEAABEAABEAABEDA6ARgIDQ6UlQIApZHgOcd5A0BBExJAMOOTUkXdYMACIAACIAACIAACIAACIBA7AnYx74oSsaVQLIUSejaucC4VoPyRiCQVOjCWgMbbXj1WSwwYa1XQPz1Wxl2zNecl9j4GAEEQAAEQAAEQAAEQAAEQAAEQMD8BGAgNJMO3r58T3lKuZmpdTQbkUAo5SqWNWKyFaQcPCwMhFiUxAo0rY4u8rDjEeoQBVKAAAiAAAiAAAiAAAiAAAiAAAh8JQADoZkuhRROyaiIT14ztY5mQeAbASxK8o0FYvFPgL0JfWp1lEZqXIvxzx8tggAIgAAIgAAIgAAIgAAIgAATwByEuA5AwEoJsGEGAQTMTYCHGR/YMUeKgWvS3NpA+yAAAiAAAiAAAiAAAiAAAtZKAB6E1qp59NvqCbDXFuYdtPrLQBUA2EiI+QhVoQoIAQIgAAIgAAIgAAIgAAIgYKUE4EFopYpHt62bgE/tsCGd1k0BvVczAZuUJbGytpoVBNlAAARAAARAAARAAARAAAQSFAEYCBOUOtEZEDCMgFwoAguTGAYLucxCQPFuHSlW2EYAARAAARAAARAAARAAARAAARAwLQEYCE3LF7WDgCoJYDinKtUCoXQIsCHby7OETioOQQAEQAAEQAAEQAAEQAAEYkLAycnpckBAQEyKqC4vy+/s7PyfmgRLkjzx5WcPX6pJpBjLwvInTZ5YcoWBMMb4UAAELJcAvLEsV3fWKrm2MRvDjq31KkC/QQAEQAAEQAAEQAAE4kLAwcHh0I4dOz7FpQ5zl7169eqHpEmTHjG3HNrt2yWyO3T1zA2L5vrs/osPDokTSa4wEGprF3EQSMAEYBxMwMq1kq4pw47ZUIgAAiAAAkwgWSLHq0/e3AMMLQLMI1lix6taSfEWhT4iojanPiJKk7BTcP1F1K85rz/oQ136ePr06aqbN28mtmQvwj179iT18/P7OSJZ86W8f/Vh1fNHrxJbshfhtfOBSVsMqSe5wkBovmsJLYNAvBFg46D/kTPEQzYRQMCSCfA1fGDHHEvuAmQHARAwIoHg0KAPRqwuQVT19M1dSpooxUlzdAb6iEjdnPqIKE3CTsH1F1G/5rz+oA916WP27Nn+WbJkWbR169Y3ESVTf8qGDRtuu7i4zFebpJ0n/+CfKo3jossnAyyS68UT126nSp1CwxUGQrVdYZAHBExAQBpVtsOoYgK0qNIMBLSHHbPxG96xZlACmgQBlRCwsbXp/+/dIxb5UG4qhEHBnx+//vDcLG6V0EdErZpTHxGlSdgpuP4i6tec1x/0oS59sDTDhg1rZ2tru2Xz5s1m+Y2ISCT6FPZ4nDZt2rt79+4dVZv3oCJ9i8F129nZ2my5dDLAYriyx+PfW8+8e/3k7VHFe5D7AwOholXsQQAEQAAELI6A4hWLYccWpzoIDAJGITCw+m97Pgd9eoZhxmE4L98/QdcenUvvW2vGIKMAjmEl0Ed4YObWR3hpEv4Rrr/wOjb39Qd9qEsfijSDBw9uef369d29e/embdu2vVbrkGOWa8aMGcEzZ86koKCglYMGDfpB6YMa980G1mn5/MHL3dsXHqD/Tt94rdYhxyzXse1ng4/vOEchIV9WNhtYKxxXezXChUwgAALGIcDDirW9rYxTK2oBAXURYCMhVjtWl04gDQjEJ4FUKVKPEF6Ev/vkb+oYn+2qsa3PwZ/vJnVIscGcskEf3+irQR/fpLGOGK6/b3pWw/UHfahLH4o07Eko4u3Sp08/+8yZMxVfvnyZXzmnlj2vViy8HfeKbb3wHPRXi1xRycGehOJ8uyWjN82+d/1hxY/vPqmOK69WbCO42tjYrP9xSL0IXG2i6iDOGY9Ahw4dQqdOnWq8ClFTvBDgNytz5841x30SqizIENuO8rBLP7HFtZ7Yto9yIGAuAsqQY8W70FxyoF0QAIGIBL56+xr9d3XSrj6XsqbJl7mga/mUEVu1jpSLd4+9v/Xs8u3+1X81+x8S6INITfqwjjvgWy9x/anr+oM+1KWPb3cKYiAQkQCGGEdkghQQSBAE2DiIxRwShCrRiRgSUAyDbIhgL1oEEACBhE9AGMXc7zy/+ujSveOfE35vI/aQjVG3n1+5pwbjIEsHfahLHxGvmISdgutPXdcf9KEufSTsux+9iysBo7/BjatACbU8PAgtU7OW7EGI4cWWec1BauMRwD1gPJaoCQSMRcBUHoSKfOypkjl1rmQOdomz5XcpoyQn2D3PvXjh9pHXH4Pf3VeLcVAbNvShTQPx+CaA6y++iUfdHvQRNR+cBQE1EMAchGrQAmQAARMQwNyDJoCKKi2KgPY9gGHHFqU6CAsCsSbAnipTdvf//d2nV12DQj7fT2Tn4JLW0ZXSOWaOdZ1qK6gsyHLhzpF3QcGfnjgmdR7cvfLolWqTk+WBPtSoFeuRCdefunQNfahLH5AGBPQRgAehPiomSGMPQhNUiyrjgYC55iCMh66hCRAAARAAARAwB4F4ef6cuLPXpKSJHdN/CvpQ7v3n17nN0VFTtJkssePVkJDgj/a2DkP6Vp+0zRRtmKJO6MMUVFGnoQRw/RlKKn7yQR/xwxmtgAAIgAAIWDcBNkT7WTcC9B4EoiTgHeVZnAQBEAABEAABEAABEAABEAABEAABEAABCybgLWT3s2D5IToImIOAnzkaRZsgAAIgAAIgAAIgAAIgAAIgAAIgAAIgAAIgAALqIOAnxGDPW2+xIYAACIAACIAACIAACIAACIAACIAACIAACIAACFgpAW8r7Te6DQIgAAIgAAIgAAIgAAIgAAIgAAIgkAAIHBB98E4A/UAXQEAtBPzUIgjkAAEQAAEQAAEQAAEQAAEQAIH4IGAbH42gDRAAAZMR8BM1HxSbv9gQQAAEjEcAC/4YjyVqAgEQAAEQAAEQAAEQAAEQAAEQAAEQMBEBb1EvGzEQQAAETEPA2zTVolYQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAAEQAAFLI+BtaQJDXhAAARAAARAAARAAARAAARAwlACGGBtKCvlAAARA4P/s3QeYFEXawPF32SXnjCCwwBJUVJKKkhoRAyYEzAExoiiKOX406h3nnfnUU8xnRBQxYMZt1APJIibykhGUHJa4X73NzDC7DBtmJ/TM/Ot5arq6urvCrwfYfemAQCoLWGby3Hacyt8A5o4AAggggAACCCCAAAIIIICAhwRsMxbLQ+NhKAikkoCVSpNlrggggAACCCCAAAIIIIAAAggg4D0BywxJr2LSJQkBBOIrYJnuNZMQQAABBBBAAAEEEEAAgYQWSE/o0TN4BFJPINNMeYnJr5pMQgCB+Apcbrp/xeQ0kx2TSQgggAACCCCAAAIIIIAAAggggAACCCCQggJWCs6ZKSOAAAIIIIAAAggggAACCCCAQIwFrBj3R3cIIBCegGUO00xCAAEEEEAAAQQQQAABBBBAAAEEIirAcwcjykljCERNwDYt87bjqPHSMAIIIIAAAggggAACCCCAAAKpKWCbaWsmIYBA4gjYiTNURooAAggggAACCCCAAAIIIIAAAl4XsLw+QMaHAAJFCthF7sEOCCCAAAIIIIAAAggggAACCCCAAAIIIJCUApaZFbcdJ+WpZVIIIIAAAggggAACCCSHQJnkmAazQAABBBBAwLMCjhlZmm90lm/JAgEEEEAAAQQQQAABBBBAAAEEEChUwDZbNZMQQCB5BezknRozQwABBBBAAAEEEEAAAQQQQACB0gjY5uDs0jTAsQgg4HkBy4yQ2449f5oYIAIIIIAAAggggAACqSHALcapcZ6ZZeIJjEi8ITNiBBAogYBj9vXfdmyX4Dh2RQABBBBAAAEEEEAAAQQQQAABBBBAAIEkFrCSeG5MDQEEEEAAAQQQQAABBBBAAAEEihCwitjOZgQQSG4By0xPHy9gm0xCAAEEEEAAAQQQQAABBBBAAIEUE7DNfPV5ZCQEEEDANgSaSQgggAACCCCAAAIIIIAAAgggkEICvKwghU42U0WghAJWCfdndwQQQAABBBBAAAEEEEAAAQQQSEABKwHHzJARQCA2Av7bjq3YdEcvCCCAAAIIIIAAAgggkGoCvMU41c448/WqgOPVgTEuBBCIu0BP3wisuI+EASCAAAIIIIAAAggggAACCCCAQEQFLNOaHdEWaQwBBFJFwEqViTJPBBBAAAEEEEAAAQQQiL4AVxBG35geEDiYwPCDbaAeAQQQKELAMtv9tx4XsSubEUAAAQQQQAABBBBAAAEEEEDAiwK2GZRmEgIIIBCugG0O1BcckRBAAAEEEEAAAQQQQAABBBBAAAEEEEAAAbEwQAABBBBAAAEEEEAAAQTCEeAW43DUOAYBBBBAAAFvCugVhbY3h8aoEEAAAQQQQAABBBBAAAEEEEBABSwYEEAAgSgK2KZtbjuOIjBNI4AAAggggAACCCCAAAIIIFAaAdscrC8VICGAAAIIIIAAAggggAACCCCAAAIIIIBACgpwVU8KnnSmjECcBSzTP7cdx/kk0D0CCCCAAAIIIIAAAggggAACfgHLX2CJAAIIxFDANn1pkNAymYQAAggggAACCCCAAAIIIIAAAggggAACCCCAAAIIIIAAAggggAACCOwXSN9fpIQAAlEQsE2blsmOySQEEEDACwKWGcRik9NMdkwmIYAAAggggAACCCCAQIoL6C8HJAQQiJ6A3tbX02Qnel14t+W8jh11/qRwBNLS9qZNn85/4oRjxzHFEbDNTj1MHmGyYzIJAQQQQAABBBBAAAEEUlggI4XnztQRiLaAZTpI+V++54+fEm3npGs/bc8eyTq7S9LNiwl5SsD21GgYDAIIIIAAAggggAACCMRVoExce6dzBJJbwDHTs5N7iswuKgImQEhCIA4CesWvHYd+6RIBBBBAAAEEEEAAAQTiLECAMM4ngO4RQAABBBDwiIA+DkGT5X7ygQACCCCAAAIIIIAAAikjwC3GKXOqmWgMBVL6uYMxdKYrBBCIrIBjmtNMQgABBBBAAAEEEEAAgRQT4ArCFDvhTDfqArbpIeWfOxh1ZTpAAIFYCXDbcayk6QcBBBBAAAEEEEAAgTgKECCMIz5dJ52AZWakbwW1TSYhgAACySDgv+04OxkmwxwQQAABBBBAAAEEEEAgtAC3GId2oRaBcAQcc5BmEgIIIJAsAo6ZiGbLZBICCCCAAAIIIIAAAggkqQBXECbpiWVaCCCAAAIIRFDACWrLNmXNJAQQQAABBBBAAAEEEEgSAQKESXIimUZcBWzTuxXXEdA5AgggEDsB29eVPp+QhAACCCCAAAIIIIAAAkkgkJYEc2AKCMRTwDadDzc5pf4s5XXsSGAg3G9derrM/2hSoUen7dwpWf26702bPj290B3ZiEB8BSzTvRPfIdA7AggggAACCCCAAAIIREKAZxBGQpE2Ul1A31qccmn++CkpN+fSTjhtzx7JOqdbaZvheAS8IuAEDcT2lf3LoE0UEUAAAQQQQAABBBBAwOsC3GLs9TPE+LwuYJsBaiYhULSACRCSEEhSAds3L7262PKVWSCAAAIIIIAAAggggECCCBAgTJATxTA9J2B5bkQMCAEEEIivgG2672myYzIJAQQQQAABBBBAAAEEEkiAAGECnSyG6hkBy4wk2zOjYSAIIICAdwScoKHYpqyZhAACCCCAAAIIIIAAAh4XIEDo8RPE8DwpoC8lScnnDnrybDAoBBDwqoDtGxi3HXv1DDEuBBBAAAEEEEAAAQR8ArykhK8CAiUX0OCgU/LDOAIBBBBIOQHbN2Mn5WbOhBFAAAEEEEAAAQQQSCABriBMoJPFUD0j4HhmJAwEAQQQ8L6AHTREy5SD14M2UUQAAQQQQAABBBBAAIF4CRAgjJc8/SKAAAIIIJB6Ao5vytx2nHrnnhkjgAACCCCAAAIIeFiAW4w9fHIYmqcEbN9o/EtPDY7BIIAAAgkkYJuxOr5sFiQEEEAAAQQQQAABBBCItwABwnifAfpPBAHbN0j/MhHGHNYY8zp21Kt6ipVann5csfZLmZ3S02X+R5OSfro1a9acWb169So5OTktIzHZzMzM+aa9b2fPnn1VJNorbhvJMo/izteD+zkFxmSbdc0kBBBAAAEEEEAAAQQQiIMAAcI4oNNlQgrYCTnqMAY9f/yUMI5K7UPS9uyRrHO6JT1C06ZN5w4YMKDaGWec0cCyrIjM13Gclp988knVDRs2zF2yZEnriDRaRCPJMo8ippmIm/U/KPQlUHYiDp4xI4AAAggggAACCCCQyAIECBP57DH2WAnYseqIfhJUwAQIkz0dffTRL55zzjkthw8fnhbJuWqg0eQGVatWrf/kk0/OXL9+fYdItl+wrWSZR8F5JcG6beag2TKZhAACCCCAAAIIIIAAAjEW4CUlMQanu4QSsBJqtAwWgSgKbN682erRo0dEg4PBw9W2zW2/1YLrolFOlnlEw8YjbTpB47CCyhQRQAABBBBAAAEEEEAgigIECKOIS9MJLWCb0Q9P6BkweAQiKLBo0aIWkbqtONSwtG3tI9S2SNYlyzwiaeLhtiwzNr3t2DaZhAACCCCAAAIIIIAAAlEUIEAYRVyaTmgBDQ7qs7BICERUQKMdJAQQKJaAbfbSq1Ydk0kIIIAAAggggAACCCAQRQEChFHEpemEFuhpRu8k9AwYPAJxFNhjnstonicoeXmERON4GpKlaydoIlZQmSICCCCAAAIIIIAAAghESIAAYYQgaSbpBJykmxETQiAGAmPHjpWOHTtKRkaG1KpVS8qUKSPnn3++LFiwIAa900UKCFhmjtx2nAInmikigAACCCCAAAIIxFaAAGFsvenN2wK2GZ5mEgIIhCHw0EMPSf/+/aVRo0by/PPPi+M48vrrr8vixYulQ4cOsmLFijBa5RAE8gnYZo3bjvORsIIAAggggAACCCCAQOkFMkrfBC0gkDQCPHcwaU4lE4m1gN5OfP/998uYMWNkwIAB+bq/8MIL5bTTThMNIP7nP//Jt81rKzt37pRnn3220GG1bt3anU+hO7Ex2gJOUAeWKQevB22iiAACCCCAAAIIIIAAAsURIEBYHCX2SQUB20xSX0qiy6RLeR07FvtBcC1PPy7p5l+qCaWny/yPJpWqiVQ4eO7cue40zznnnAOmm24Mr7vuOrFt+4BtXqvYvn27DBs2TA4//HD3NulQ4zv99NMJEIaCiV+dZbrONjlp/w6PHy09I4AAAggggAACCKSKAAHCVDnTzLMoAbuoHRJ9+/zxUxJ9CjEff5p50UbWOd1i3m8idlitWjV32PpyEg0IFkx6heHu3bsLVnt2XW+Prlu3rmfHx8DyCdhmzZ/zbWAFAQQQQAABBBBAAAEEiifAMwiL58ReCCCQigIm2EUqnkDLli2lXr168uCDD8revXvzHaTBwZdfflm6dOmSr54VBCIsYAe1Z5myZhICCCCAAAIIIIAAAggUQ4AAYTGQ2CWpBeyknh2TQyBGAmXLlpV3331XRo0aJU2bNpWFCxe6Pe/YscNdnz9/vvuMwhgNh24QUAG97djWAgkBBBBAAAEEEEAAAQQKFyBAWLgPW5NbwE7u6TE7Lwro61eTNfXo0UOWLFkif//73yUtbd9My5QpIyNHjpRffvlFGjdunKxTZ17eE3DMkJL5j5v3xBkRAggggAACCCCAQEILECBM6NPH4EshYJlj9a3FtskkBEotUOy3wJS6J283sGXLFsnLy5Nx48bJN998I3pl4ZAhQ6ROnTreHrhvdFWrVpWff/5ZatasmRDjZZBFCtgF9ii4XmAzqwgggAACCCCAAAIIpKYALylJzfPOrEUcg9ATCAQQiJzAbbfdJo8++qgcddRR7huAb731Vjn77LPl/fffD/niksj1HLmWtm3bJv369QvZ4AsvvCCffvqp+7KVRx55JOQ+VHpawDKj0/8Y0mS7n3wggAACCCCAAAIIIICAK0CAkC9CKgs4qTx55o5AJAUWLVrkBge///77wMtIli1bJk2aNJHx48fLWWedFcnuotZWhQoV5OGHHw7ZfqtWraRy5coHvIQl5M5UelHAMYPS247toGyKJAQQQAABBBBAAAEEECBAyHcg1QRsM2HHl82ChAACkRBYtWqV+xbj448/PtCcPnPwiiuuEH1BSaKk9PR0OfbYY0MOt0aNGtKgQYOQ26hMKAG7wGh1vWBdgV1YRQABBBBAAAEEEEAguQV4BmFyn19ml1/AMqt6e5ljMgkBBCIooFfXrVmzRl555RX3FlxtesWKFeI4jrRu3TqCPUW3qU2bNkmjRo1C5s8//zy6ndN6PAQs0ynPo42HPH0igAACCCCAAAIIeEqAAKGnTgeDibKAZdofEeU+aD6OAnv37pU9u3fHcQSp23XdunVFn9E3bNgw98UkLVq0kEMPPVTat28vffr0SRiY6tWry7p16wJ59erVMmbMGGnevLl069YtYebBQIst4Jg9/W87tot9FDsigAACCCCAAAIIIJBkAgQIk+yEMp1CBWyzVTOpFAK3Xn+ptGqQIT98n52vlZnTJrv1N119Qb764q5s3rTRPX7jxvXFPeSA/Ua//oIMvSa8/g9oLLUqLDNdu7RTvuqqq2T9+vUyZcoUGT58uEyePFnee+89KVMmZv/URGQe+gZjf65fv74MGDBAOnXqJJ999llpiUpzfETmVpoBJPmxtpmfZhICCCCAAAIIIIAAAikpELPf2lJSl0kjkIQCepWepi8+GZtvdl9/9qG7vjdv3/Z8G0uykpdXkr3ZNzICjmlmoMmKb5tc4rR9+3a58cYb5eabb5bXX39dpk2bJm+++aZbN3v27BK3F+YBjjmuVPPQfjds2BDIGvCcNWtWvIODOizH5LNNDvscmWNJCCCAAAIIIIAAAggggEBIAQKEIVmoTDIB/YWaFEGBdh2Pk3HvvRG4nVeDhh+8+1/Ren/asP4vuf/26+T4tg3lgrO6y5g3X/Jvkq8+HSeX9usl7bNqyh03Xi5bNm8KbNOCtjfSvl30asXdu3bJ2jWr5eZrLnTbGnT+qfLrnB/d/XW/fz/ygJzc5XDR+tkzp+Zrh5USCQzy7T3cLEschEpLS5NKlSoFsl41+O2338q7774r5cuXL9FASrlzqeaxcePGwNWDehVhrVq1pEOHDtKmTRsv3Co9zNhsMDmsc1RKVw5HAAEEEEAAAQQQQACBJBbISOK5MTUEVCDbZJ47GOHvwjGdu8uyJYtl2pTvpXMXS36ePUNyc7dL564nyuKFc93e/v3IgzJ/7i/y9EtjZPmyHLn9hoHSo9dpUrZcORlyxQB58oV3pHr1miYQeJsbPBxw0RXucXnmCsKHR9whn330nowe/52kZ2TI4Mv6SrXqNeTRZ1+Xyd99I317d5Lpc/+U77K/kJefe1yG3j5cdu3aKY88dI/07tM3wrMtXnP+h5gVb29P7uWYUeWYnGmyJg1CadY/P7bJhaYKFSrIww8/nG+fnTt3us8g1CvyYpgc01eOyZkmayrRPKpVqyY5OTnugf4PfXuxPpvQA8kxY8gxuZ3Jmko0t32H8IkAAggggAACCCCAAAIIHChAgPBAE2qSR8AyU5losm0yKYICerXYWf0vki/Hf+AGCPWKwL4DLnFfTuHvpvdpZ8tlV90ghzRqLBUrVXarFy2YK02atXDLf639Q3qceKo8//qHssMEF/3p8X/8n3z0/lvy4dfT5ZCGjeXnn2bKnB+nS/a0BdKocaYc3+1Eee/tV9xA4cQJn8l5l1wlg6692T08GlcQ5vkHFuOluVFbr/COV/fBs/UHoYLrilUuZ4LBtWvXdm/R7dy5c7GOMTtFa87Fmod+tzXfe++98uWXX7pjtixLHnjggZK+jTla8wjl6J9bT7PRCbUDdQgggAACCCCAAAIIIIBAYQLcYlyYDtsSXcAxE7ATfRJeHb9eqae3FestwB++9+YBV+6lmVtMb7jyXGnbpLJ7BaB/Hg0bNZG7hv9THrjnJmnXoobYd95gAjL7/yp6+7XnZeuWzbJp476rzpYtWeQe2vOYLPclJq0PKSt//blG1v21Vr4yzz1s32l/4KlDp+P93ST80ojowxzTYpxDuY3wjSHUtkBdbm6u3HXXXYF85513Sr9+/eS7776To48+OrBfMQqRmHOoboo1D73qsXfv3vLHH3/I+eefL3Xq1JE9e/ZI9+7dZe3ataHaPVhdJOYRqo3g/vyXZvrn5gRvpIwAAggggAACCCCAAAIIFFdg/2/lxT2C/RBAAAEj0L7jvsDcK6OekM2bN8oxx3cLuGhA5fYhA0VvRZ42d61MnLlYKlep6m7XZxP2OvUs+XnpVvnv+1+7zx/U5wj609gvp8qZ/S6U+28b7AZmqlbdd2vnpDkrZPbiTW4e+8UUOeWMftL68CPlj1Ur/IfK4kXzA2UKJRawzRH+gJMe7A86aX2RSW8NX7duXSDryz1at27tvtzjhBNOKPL4CO5gm7bCnsecOXNk1apV8v7778u5557rzkHfxHzooYcGriiM4FhL2pTOTZN/fk+asgYRbZNJCCCAAAIIIIAAAggggEDYAtxiHDYdB3pYwDZjc3zZLEjRENBnA+ptxf968G4ZcOEgc3txuUA3u3fvCgQNNcCnVxrqVYH6UpEVy5fKJeecKB9NmOHenjy1a0/3GYb+gxs3yZTb7x8p3dtnyrtvvCinntnf3fThmDfksqtvlP9N/FquueQsGT/xJznl9H4y7t3XxTqpj+iVXxM+/0g6HBvTYJR/2Mmw1NtUNWlg0NZCSVLFihVl1KhR7iELFixw3wKclZUl+vy+GKdSzUMDnZr0JSv+tHv3btmyZYv4t/nr47D0z00Dg3Yc+qdLBBBAAAEEEEAAAQQQSFKB/b8BJekEmVbKCdhmxj1MdkwmRUFAAyf6jDZNJ59+Tr6lrpQxtwuXL19BBg+9U266+gLp2Kq2jB83Wrqb5w3qy0mOOLK9dO95ivQ6rpX7VuJPzLbBN93ttuN+mLYbHHKo3Hrv3+SfD97lBmWeeP4tefiBO+WIxpXc4OCt9zwkLVsfLmf0PV927NghvY9vI6f3OEqObNcpX2Bnf6OUihCwzfYSXTEYqr0PP/zQvZ24ZcuWcswxx7hvA37kkUdC7RqtOts0XKp56BuLmzVrJldccYVoYHDhwoVy7bXXulcVnnjiidEad3HaLfXcitMJ+yCAAAIIIIAAAggggEBqCuz7LT81586sk1PANtNyfNksSCqQ17Fj3vzxU2KOsXHDOvN24V1Sp259N9Cnzw7UsqYVy5fILnPVX2bzlsUa1/ZtW2XZ0hypf0hD9+3H/oP2mCDO0iULpV79hoHbmP3bSrtMM+NrMcCSBR9NKrQp3S/L7Dc/kvv16743bfr09EI7ju1GcwHdvqvrQnW7ceNG92rBf/zjH3L55ZeLvqBk4sSJctlll8mnn34qXbt2DXVYvjpf4Dna/y4VOg8d0NKlS+W+++6Tq6++Wq666io58sgj3WcrdurUKd94D7YSo3kcrHvqEUAAAQQQQAABBBBAAIESC3CLcYnJOMDjArbHx5dSw6teo1Zgvho08QcHtbLRoU0D24pT0Dcht2pzxAG76q3OzVq0PqCeitgKLFmyRKpWrSq33XabpKfvi2v27dvXfVHJtGnTihUgjO2ID95bkyZN5Mknn3SvgJw7d+7Bd2QLAggggAACCCCAAAIIIJAkAtxinCQnkmkggAAC8RQ4/PDD3e7Hjh0beFafXok3YcKEkr7FOJ7TcPu+5pprpFatWnLEEUfIihUr5O9//7t89NFHgXnFfYAMAAEEEEAAAQQQQAABBBCIsAABwgiD0hwCCCCQigIZ5krOV155RW655RbJzs52Cbp06SInn3xyQl09qM8cfOedd+T777+Xww47TF5++WU57rjj5JJLLhF9wzEJAQQQQAABBBBAAAEEEEhGAW4xTsazypwQQKBQAX1uod6aXJwU7QfiFWcMibJP//79pXv37lKpUiV3yD/99JN7m26ijN8/zkMOOUQ0uKkvwPnnP/8p999/v5x77rkyadIkOeqoo/y7sUQAAQQQQAABBBBAAAEEkkaAKwiT5lQyEQRKJ7Bw/m/SqkGGXHBW9wMauvLCPu62ub8WfgXVc0+OlNtvGOge/9Wn42RpzkK3/O9HHpA7h15xQLsFK/TNxqNffyFf9S9zZrl9r1/3Z7760qx0alNXfv/lp9I0wbEhBPTW3Hr16smxxx7r3pr7n//8J+FuzW3RooU0bNhQxowZI23atJFZs2bJ+vXrRZ9F2LRpyZ6bGYKIKgQQQAABBBBAAAEEEEDAkwIECD15WhgUArEX8L+hdubUSfLHqhWBAWhg7rvsLwPrhRW0jb1797q7PPXICNHgnqZzzrtUrh16h1su6sM/joL7Hay+4H6sx0cgWW7N1asG16xZI+edd540atTILevzCLdv3y6dO3eODy69IoAAAggggAACCCCAAAJRFiBAGGVgmkcg0QTadTxOvv7io8Cws78aL1rnTzOnTZbBl/X1r4oGFG+86rzAuhaeeHi46NWGI//vNpn07QSZ/H22fPHJ2Hz7vP7S0zJ9yvf56oqz8tOsaXLR2ZYc37ah3HHj5bJh/V+Bw55/6h/Sp/tRcnKXw2Xk8Ntkz5497rZ5v/8ilw3o7dY/9a8RsnXL5sAxFCIn4L819/rrr5f//e9/0qtXr8CtuZHrJbotLVu2TKpVqyYzZsyQmTNnunn27Nmib2KuWbNmdDundQQQQAABBBBAAAEEEEAgTgIECOMET7cIeFXg1DP6y/gPRgeG9/nH74vW+dOWzRvl1zk/+ldl06YN+dZ1w9kDLpLGTZvLeZdcKYe1PVpWrVgmixfOCxyjhf9N/FqWLF6Qr05XnK8/lVFP/zOQ33/71cA+f679Qwacdrxps5089eJoExxcJ9dfvm9s8377WV4d9ZTccs9D8rdHn5eP3n9LvvniY/fY6waeI2XLlpVhdz0gk76bEGiPQuQEkuXW3KysLPnkk0/c24mbNGkimvVKQr3NeOfOnZEDoyUEEEAAAQQQQAABBBBAwEMCxXtKv4cGzFAQQCC6Aj1PPl3+MeIOWbN6pZSvWNEN2N3zwKNuXXF7btaitVSqXFlatDpMataqU9zD3P0WmUDibvMSEX/atHG9vyiffPCONDjkULnvocclLS1N6tStJ6d0OULWrlltXjqSLs+88r4c3eFYWb1quTQ8tIn8/utP0rLNEbJsySJ577NJ7liaZmZJ396dAm16spCe7slhFTao4Ftz/fvprbkdOnRIqFtzN27cKHXqhP7OfvDBB9K37/6rZ/3zZIkAAggggAACCCCAAAIIJLoAAcJEP4OMH4EIC9Sp20A6d+0pE8zVd5UqV3FvL27Q8NCD9hIczDvoTr4NenvykEH7rvj76881MmXSRHnkoXvcfiZM2XeF4RXX3iwXXHZNoCl9juE5vY9x15fmLHKDf60PKRvYroX16/6Squa20OefelguOLObVK5S1d1undRHpv3wnWS2aBUIVLY+/Mh8x8Z8xTynseXp+2/ZPmj/GgFNoBR8a65/6Okm0Nm2bVspUyZxLlbX24tXrVoVkNdnD77xxhsyatQo6dGjR6CeAgIIIIAAAggggAACCCCQTAIECJPpbDIXBCIkcNqZA+TTD991A219zs7/fEHtYteu/bdarlqxtNi9tmjVRh599nV3/0f/fq90Oq6r9Oh1WrEDSNWq15Aj23WSNz74xm1j965dMt88X7BZi5byrwfvlpxF8yV7+kJpdGhTuenqC9x9GjdpJjnmqsQ95qrE9IwMN8BY7AFHY8e8vL1pM2Yk3iWCRVjorbmTJ08+YK9//etfcsIJJ0iXLl0O2ObFCg1uNmjQIN/Q7r33XjdA+PHHH8tll12WbxsrCCCAAAIIIIAAAggggEAyCCTOZR3JoM0cEEgQgV6nnOle3ffNl5+IloNT/QaNRK/+0yvz9A3Hb782KnhzoJyRniEbzTMCg1P16jXlhO693Fy3XgNpdVhbt6xXLBYnaUBxzo/T5efZM9xbjF974SkZcsUAKZNWxg0Otj26kxsc1KsOv83+wn2jcsfjuriBzrf/+7x5E+02efvV54vTFfuEIaDBs/6LgjAAAEAASURBVPr16+fLd9xxh5x22mlu3bp1+b8PYXQRl0P0CsiffvpJBgwYEJf+6RQBBBBAAAEEEEAAAQQQiLYAVxBGW5j2EUgQAf9tobqs16ChdDj2BNlhbq/Ul43k5m4PzCLLPFewS4+T5OK+Pd3A2/EmuDfXvCBEkx7rv520a89T5P/uuF6q16zl23bg/0f4+3R3KOJD9+1q9ZZBg4e5fevutevUk8efe9O9MvCyq25036b8wbv/lVq168o5510mzzz2kJxyej+55e4H5YF7bnKzzktvQS5J30UMjc1GYMmSJfL666/Lhx9+KJUqVQqYjBw5Urp27SrdunWTqlX33fod2JhABd5gnEAni6EigAACCCCAAAIIIIBAiQUS6hlXJZ4dByCAgCuQ17Fj3vzxUyKqoS8GqWGCf2XLljtouxs3rJMqVauLPosukmndX2vlz7VrpGmzFlK+fIVA0zt37nBfrtKocaYbANT9qteo5favVztu3rTRDXgWNziYZt5amzXAkvkfTQr0EapQov36dd+bNn16ZEFCDSrydXkmHbRVfcPvlClT3EBg8E5z5syRevXquVcQBteHKvvOS7T/XSp0HqHGVdK6GM2jpMNifwQQQAABBBBAAAEEEEDgoAJcQXhQGjYggEBhAnqLcFFJg3PRSHqFoOaCqVy58nKoeeagPwXvo29TLukblf3tsCxaoFy5cgcEB/WoI4+M80thih46eyCAAAIIIIAAAggggAACKS9w4D1/KU8CAAIIIIAAAggggAACCCCAAAIIIIAAAqkjQIAwdc41M0UAAQTCFmjWrNkCx3HCPr6oA7Vt7aOo/Uq7PVnmUVoHjkcAAQQQQAABBBBAAAEEggUIEAZrUEYAAQQQCCmwZcuWrV988cXGkBsjUKlt161bt/CHPUagn2SZRwQoaAIBBBBAAAEEEEAAAQQQCAjwDMIABQUEEEAgygIRfllLlEebr/m1a9e2e+utt37XylNOOaW6ZVn5toe7olcOanDQtL166dKlA8Ntp7jHJcs8ijtf9kMAAQQQQAABBBBAAAEEiiNAgLA4SuyDAAKuwF9/rpHj2zaUE08+Q57777iAyorlS+SnWdPktDMHuHVffTpOWh9+pDTJbBHYpySF4Pa2b98mRzerJl9N/t28tTirJM3Edl/zht+Wpx9XdJ/FfYVy0S3FfA8TwGvzzTffvDZ69OgTFi9eHJGTobf86lV9GriL1YSSZR6x8qIfBBBAAAEEEEAAAQQQSH4BAoTJf46ZIQIRE/j8k/elcpWq8s2Xn8i6v9YG3iQ899c58shD9wQChE89MkKuH3Zv2AHC4PbKl68gb47LlnoNGkZsHlFpKC9vb9qMGelRadtDjU6dOjWiV/mZQGNcZpcs84gLHp0igAACCCCAAAIIIIBA0gnwDMKkO6VMCIHoCbz35sty+30jpXadeqJXCWpavnSx/O3+W2TZkkVy41XnyRMPDxcN8I38v9tk0rcTZNeunW7wsNdxreSc3sfI+HGj3eN27MiV/qd2ljFvvSx9uh8lun3Mmy8d0N7u3bvk8ZH3y8YN62TPnj3y7ON/k+7tM90rGUcOv01yc7e77d17yzXy5ivPyqX9ernbHrjnJnd/dyMfCCCAAAIIIIAAAggggAACCCBwUAEChAelYQMCCAQLLJz/m/wyZ5acemZ/6Xf+QPlgzOvu5tp168t5F1/hBg0H33S3nD3gImnctLmcd8mVcljbo+XJh2357OP35Pb7R8rAq4fKsMEXy5RJE2Xv3r0y58fp8sIzj8g9DzwqnbtYcu+t10rVatXztadBwelTvpfc7dvl/bdfcfe/dugd8tSLo+Wzj96T5578hzuOhQt+lxF3D5UBF10h1w69U954+Rn3uOA5UEYAAQQQQAABBBBAAAEEEEAAgQMFCBAeaEINAgiEEPjkg9Huswdr1a4rvfv0lZlTJ8mKZTlSsWIladmmrVSqXEWOOLK9NGvR2pQrS4tWh0n1GrVk1NP/lKG3D5dTz+gvfc+7VPqee4l8Of6DQA8P/PNZ6Wr1doOEWqnPOQxuL7CjKYx+40U3yHjxoOvlmM7d5Mbb7pex77wW2EX7OXvAxXL5NTfJke06ydLFCwLbKCCAAAIIIIAAAggggAACCCCAQGgBAoShXahFAIEgAb2K753/jnKv/NNbgYdedb679fOP3w/a68Din2tWu5W33zBQWjXIcPO4MW/IX2v/COx8SMND3bI+21DTzh073WWoj0UL5krbozsGNh3apJmsXrU8sN7w0CaBco2atWTnzh2BdQoIIIAAAggggAACCCCAAAIIIBBagJeUhHahFgEEggT0Fl+9su/ldz6TjIx9f218PPZted9cvXfl9bcG7Zm/qLcLa9I3Hh/f7US3vHrlMilTZv+7PNLT95fdHQr5aHtUB/kjKCC4eOE86dbz5MAR6en8lRbAoIAAAggggAACCCCAAAIIIIBAMQW4grCYUOyGQCoLfPTem3LKGf3cW4E7d+0pmi8adJ0smPerzPvtZxPwKyPbtm6RPbt3u0wZJlC3cf06qVipsrTreJzo8bt27nRvSR444BT54fvsQjkLtuff+aTTzpaxo//rvhBFX37yyQfvmADhKf7NLBFAAAEEEEAAAQQQQAABBBBAIAwBAoRhoHEIAqkksH37NvdNw6f33XdbsX/uh7dtJw0OOVS+GD9WtKzptB5HucuuJmj3f3dc776c5MF/PSczp02WTq3ruG8rPrJ9J+l/weXufqE+0tLSDmhP99P6Pmef5942rLc5H9G4kuibkM/sd2GoZqhDAAEEEEAAAQQQQAABBBBAAIFiCqQVcz92QwCBBBbI69gxb/74KVGdgV49mJu7XfzPEty4YZ1UqVpd9BZi3bYkZ4H7IhMNKhYnFWzPf4w+D3H50sVStlw5OaRhYzdw6N8W6WWaueoxa4Al8z+aVGjTaWZMWWd32Zs2fXrx75cutEU2IoAAAggggAACCCCAAAIIIBA7AR7YFTtrekIgqQXSzbMJ/cFBnai+wdifdFvzrDb+1WItC7bnP0gDjk2bZflXo7/My5OWpx9XdD96iSMJAQQQQAABBBBAAAEEEEAAgQQUIECYgCeNISOAQAwF8vL2ps2YwZWBMSSnKwQQQAABBBBAAAEEEEAAgdgK8AzC2HrTGwIIIIAAAggggAACCCCAAAIIIIAAAp4SIEDoqdPBYBBAAAEEEEAAAQQQQAABBBBAAAEEEIitAAHC2HrTGwIIIIAAAggggAACCCCAAAIIIIAAAp4SIEDoqdPBYBBAAAEEEEAAAQQQQAABBBBAAAEEEIitAAHC2HrTGwIIIIAAAggggAACCCCAAAIIIIAAAp4SIEDoqdPBYBBAAAEEEEAAAQQQQAABBBBAAAEEEIitAAHC2HrTGwIIIIAAAggggAACCCCAAAIIIIAAAp4SIEDoqdPBYBBAwFMC6emeGg6DQQABBBBAAAEEEEAAAQQQQCAaAhnRaJQ2EUDAewItTz/Oe4NKhBGlpaWZYdom9zB5osmOL5sFCQEEEEAAAQQQQAABBBBAAAEEEEAAAW8KWGZY2SbrkhQZAcs0Y5usS3+yTcGf/XUsEUAAAQQQQAABBBBAAAEEEEgoAe6fS6jTxWARKJaAbfYaaPIIkx2TSZERyDHNOCbrMjhlBq3k+MqWWV7uK+f4liwQQAABBBBAAAEEEEAAAQQQQAABBBCIioBlWrWj0jKNhiug50Nzwas4LVOnmYQAAggggAACCCCAAAIIIIAAAggggECpBSzTQp7JBYNQpW6YBqImYJuW9XzpeQtOVvAKZQQQQAABBBBAAAEEEEAAAQQQQAABBA4mYB1sA/UJLeAPGhYMHCb0pBg8AggggAACCCCAAAIIIIAAAggggEDkBGzTlAaPNJBESh0BPeec99Q538wUAQQQQAABBBBAAAEEEEAAAQQQQACBkAJWUK2W/UFDO6ieIgIIIIAAAggggAACCCCAAAIIIIBAkgjoVYIaALKTZD5MIzoClmlWsz9ZpqDfHduXzYKEAAIIIIAAAggggAACCCCAQNECaUXvwh4IIBBDAcvXlxPDPukqeQQsMxXNji+bRb5Ac3C9biMhgAACCCCAAAIIIIAAAgggIAQI+RIgED8By3Q93GRd8mfRIJCiJmAHtewvW6ZOs+PLZkFCAAEEEEAAAQQQQAABBBBAAAEEEIilgG06s2LZIX0hECRgmbJtsv+2ZFN0k2U+NZMQQAABBBBAAAEEEEAAAQQQQAABBCIoYJm2NBCjSxICXhawzeD0u5pnspb9yfIXWCKAAAIIIIAAAggggAACCCCAAAIIlEzAMrsTHCyZGXt7T8A2Q9KgoWbLZBICCCCAAAIIIIAAAggggAACCCCAwEEELFNvH2Qb1Qgkm4A/aEgQPNnOLPNBAAEEEEAAAQQQQAABBBBAAIGwBDRIQqAkLDoOSnABq8D4NXCofxbsAvWsIoAAAggggAACCCCAAAIIIIAAAkklYBWYTcH1AptZRSClBCwzW7vAjP1Bw4L1BXZjFQEEEEAAAQQQQAABBBBAIFYCabHqiH4QSDIBy8xnuG9OPZNsbkwHgWgKWKZxzY4vm4WbbN/SMUvNJAQQQAABBBBAAAEEEEAAgRgJECCMETTdJJ2A5ZuRk3QzY0IIxEfADuq2YNkx2zSTEEAAAQQQQAABBBBAAAEEEEAAgbgIWKZX/22RcRkAnSKQogKWmbdtcqg/f7qNhAACCCCAAAIIIIAAAggggAACCMREQIMTVkx6ohMEECiOgGV20j+XeSbbJvuT5S+wRAABBBBAAAEEEEAAAQQQQAABBMIVsMyBBATD1eM4BGIvYAV1qWUNGmrWMgkBBBBAAAEEEEAAAQQQQAABBBAokYAGBgkOloiMnRFICAHbjFKDhvz5TojTxSARQAABBBBAAAEEEEAAAQQQiJ2AZbqyY9cdPSGAQJwF7AL9+/9TQOutAttYRQABBBBAAAEEEEAAAQQQQACBJBawzNy4miiJTzBTQ6AEApbZ1y6wv677c4FNrCKAAAIIIIAAAggggAACySeQlnxTYkYIhBSwTK0TcguVCCCAQH4By6xq1mS7n/s+/GXHrGomIYAAAggggAACCCCAAAIIIIBAAgjoLYT+qwUTYLgMEQEEPCxgm7Fp1r9XgpNtVqzgCsoIIIAAAggggAACCCCAAAIIIOAdAcs7Q2EkCCCQhAKWmZNtsv8/I0wxkKxAiQICCCCAAAIIIIAAAggggAACCERdwDI9+H9Bt6PeGx0ggAAChQtYZjN/JxVuxFYEEEAAAQQQQAABBBBAAAEEIipgmdY0kxBAAAGvCVhBA9Jyni/bZklCAAEEEEAAAQQQQAABBBBAAIEwBCxzjF6Zo5mEAAIIJKKAZQat2Z9sU9DAoS4tk0kIIIAAAggggAACCCCAAAIIIFCIgAYGrUK2swkBBBBIVAHbDNwKGryu6995urRMJiGAAAIIIIAAAggggAACERdIi3iLNIhAZAUs09xwk0eY7JhMQgABBFJNwA6asL/sX+qm4LKukxBAAAEEEEAAAQQQQACBEgmkl2hvdkYgtgK26W6gyQQHY+tObwgg4C0BxwzHn4NHlhm0kuMrW2Z5ua+c41uyQAABBBBAAAEEEEAAAQQQQCBhBCwzUjthRstAEUAAAe8J2GZImvNMtkz2J8sUNJMQQAABBBBAAAEEEEAAAQQQ8KyAPmNLs+XZETIwBBBAIHEFbDN0/TtWA4fByQpeoYwAAggggAACCCCAAAIIIIBALAWsAp0VXC+wmVUEEEAAgSgI+IOGuiQhgAACCCCAAAIIIIAAAgggEBMBy/Siv4jyy2hMuOkEAQQQKJaAVWCvPLOuf0/bBepZRQABBBBAAAEEEEAAAQQQQKDUArZpwSp1KzSAAAIIIBBtAct0oNmfLFPQwKHty2ZBQgABBBBAAAEEEEAAgWQQSEuGSTAHzwpYZmTDTZ5osm0yCQEEEEAgsQUsM3zNji+bhbtuacEkx5e1TEIAAQQQQAABBBBAAIEEESBAmCAnKgGHaZkxa3BwhMmOySQEEEAAgeQVsIOm5i/7l7opuKzrJAQQQAABBBBAAAEEEEAAgSQUsMyc9HlVJAQQQAABBFTAMtk2Wf9t0KU/WaZg+1dYIoAAAggggAACCCCAAAIIJIeA/vKn2UqO6TALBBBAAIEoClimbdvkPN/SLNxkmU/NJAQQQAABBBBAAAEEEEAAgQQQsMwYNZMQQAABBBCIlIBtGtKgoWbLZH+y/AWWCCCAAAIIIIAAAggggAAC8RewzBD0F7dsk7VMQgABBBBAINoCtumAf3uirUz7CCCAAAIIIIAAAggggEAhAlYh29iEAAIIIIBArASsAh35g4a2qS+4rcCurCKAAAIIIIAAAggggAACCIQjoFcJ+n/5Cud4jkEAAQQQQCDaApbpwC7Qif77pXWaSQgggAACCCCAAAIIIFAMgbRi7MMuqSlgmWk7qTl1Zo0AAgggkMAClhm7Zk22+7nvw192zKpmEgIIIIAAAggggAACCCCAgE/AMkv/1YJaJiGAAAIIIJCMAraZlD8Hz0/rrOAKyggggAACCCCAAAIIIIBAqglYZsKaSQgggAACCKSagGUmbJus/1GmOTjZwSuUEUAAAQQQQAABBBBIZgFuMU7ms3vg3CxTNdxX3fPAzYlRM3jwYKtChQqPmlxt3bp1WbEedc2aNedXqVJl0ooVK1597rnnnFj3X8r+rDp16jxhxl8pJyenZSnbCuvwzMzM+bVr1540Y8aMV00DTliNcBACCCAQXQHLNK//XupyhMm2ySQEEEAAAQQQQAABBJJWgABh0p7aAyZmmRr9ZUd/0XFMTshk2/aLK1euvHLIkCHu+LOyYh4flAULFsi8efNyv/rqqwqNGzd+7f777788ETCPPfbY16ZOnXrZ119/vS09Pb2SZVlxGbbjODJhwoTchx56qELHjh1fM4HChPCLCxadIoCA1wQsMyD/lYb6H22OyZoskx2TSQgggAACCCCAAAIIJKQAAcKEPG3FGrRl9kr4gGDwTDU4mJGRce6wYcOqBdfHs/zYY49t2LFjx2gT7Bocz3EU1fcRRxwxqmLFiudNmzatelH7xnJ7p06dNmzdunX077//7mm/WJrQFwIIJIyAZUbq+EZrm6X+m+uYnND/EWfGT0IAAQQQQAABBBBIQQEChMl50m0zrR4mJ80vKXpb8d69e7Mff/xxz50xE7AUc8vs6SNHjvzUc4MzAzLBwTN/+eWXj/Ly8rw4PElLS5PmzZufvmjRIk/6eRKNQSGAgFcFLN/AHN/SNkv993iiyY4vmwUJAQQQQAABBBBAAAFvCZTx1nAYTZgCljlOsz/ZptDTZMfkpEj169e/+NRTT/XkXPR259zc3L95cnBmUOZ5g+eaqy+9OjzJzs6WTZs2edbPs3AMDAEEvCjgmEFp9ifbFPQ/6zQ57ue+D9ss/HlfDZ8IIIAAAggggAACCMRRgABhHPEj1HW2aWd4hNrybDPmNl6rRYsWnh1f+fLlq3h1cGvWrOnSo4dewOLdVLVqVc/6eVeNkSGAQIIIOGacdoGxal2oZJlK22RdkhBAAAEEEEAAAQQQiJlARsx6oqNICVimISeosaS5jThoTgcU9W3F8XghyQEDCVGh44rH25RDDCVk1eLFi5vH64UkIQdUoFLHZsaYVaCaVQQQQCCZBRwzOc0Fk+Wr0P/4c3xlXdgmO75sFiQEEEAAAQQQQAABBCIrQIAwsp7RbM0yjfuvFHSCOgouB1VTRAABBBBAAIEEE7ALGa/+DJBtclrQPpYpO0HrFBFAAAEEEEAAAQQQCEuAAGFYbHE5yDK9psTVgnHRpVMEEEAAAQS8K2CHGJpl6vxBQ90cHDjUdRICCCCAAAIIIIAAAsUW4BmExaaK6Y6W6U2vEtClP9mm4PhXWCKAAAKlEbj58ezM0hxf8Fivt9fncrtdwTGzjkCCCzhm/D1N1sBgcHDQMut5vmybJQkBBBBAAAEEEEAAgSIFuIKwSKKY72CZHvWKgBEmOyaTEEAgxQUGDP33rAaZbSMW4CpfLkN27NwtNzym/w9R+uT19qpVriAzJ33umJlqMIUUYQHzPdJgVMTS07ek5GkKDvCV1tIxDfjbs4Ias01Zf75wTJ5osm0yCQEEEEAAAQQQQAABV4AAYfy/CJYZgmbbZE2OL2uZhAACCMiWjX9u6N25tRzWvF5ENMZOmCOHNaufMu1NmbPUBAgjQkcjBxEYelG3g2wpefXTt5jL3/IiF3NMS0vzfHslVyr2EU7QnrYpa7ZMDk62WelhsgYNHV82CxICCCCAAAIIIIBAKgkQIIzv2dbfgByTR8R3GPSOAAJeF9i0NTeiQ0y19iKKR2MIJLaAU2D4tlm3fNksAskOlPb/J2ZQFUUEEEAAAQQQQACBZBIgQBjbs2n5unN8S/8tQL5VFggggAACCCCAQMwFHNOj5uBkmxXNmmxf1rLly45ZaiYhgAACCCCAAAIIJIFAmSSYQyJMwTaD1KsFhyfCYBN9jLt27ZLu3bvLp59+GpjK559/7tbNnj07UPfyyy/LBRdcEFingAACCCCAAAL5BGyz5s/+DZavoD/T+MtapWXNJAQQQAABBBBAAIEEFCBAGL2TZgU1bZuyXi3Y02THZFIUBcqWLSudOnWSOXPmBHqZPn26W545c2agToOFuh+paIErr7xSbrzxxsCOW7dulV69eknjxo1lyZIlgXoKCCCAAAJJL2CbGWou+DONZeo0aKj/Iaplf7L8BZYIIIAAAggggAAC3hUgQBj5c5Ntmiz4w3Hke0muFi0zHTuSU+rYsaPMmjXLbXLv3r3y/fffS79+/WTy5MlunV5lqNvbt28fyW7j0ZZtOtUc1aSGmjVpcPCMM86QnJwcmTRpkjRt2jSqfdM4AggggEBCCNhmlBo0TDPZMdmfhpuC/lykmYQAAggggAACCCDgUQEChJE/MSNMk/rDsR35ppO2RcfMbKDJ+suDbXKpU7t27WTlypWyYcMGWbhwoWzbtk0uuugi+f3332XTpk2yaNEit4+jjjqq1H3FuQHb9D/c5IjZFTafLVu2uMHBNWvWyP/+9z/3CsLC9mcbAggggEDKC/iDhvqzUXDSf7c0ZwdXUkYAAQQQQAABBBCIjwABwvDdLXOo/lCrP9xq2Z8cf4FliQQG+fYebpZqavvWw1q0atXKPU4DgjNmzHCfP1ivXj1p0aKFe+Xgr7/+Kg0bNpS6detKbm6uzJ07N6x+PHLQCN84Bpplqe0ONicNtvbp00ccxxF9pmODBg0Otiv1CCCAAAIIFCWgAUPN/n/D/Pvrv2PZJtv+CpYIIIAAAggggAAC0RcgQBi+sWUO1R9q9Ydbx2RS6QQcc3hOUBPDTTnsYJc+h7Bz586igcCpU6cGnjWodfo8Qn0+oZY1/fHHH/Lvf//bLSfoh+0bd6ZvOdAsw7bztXHA4q233nKtdMMTTzxxwHYqEEAAAQQQCEPAKXCMP2gYXG+ZfbJNtn3ZLEgIIIAAAggggAACkRQgQFg8Tcvspj+YavYn2xQc/wrLIgUss4ffMM+UQ+VMU18wDTcVum+Jkz6HUJ+RpwHBDh06uMfrUuv0qsJQzx/88ssvZfjw4aLPKCxJ0vZMCjWnWNUFDzfTtxK2XXBj/rJelanPcnz++eflsccec8v+bRFaxsoq24xXs+3LZkFCAAEEEPCQgGPGotmfHFPwX2moZX+yTMH2ZS2TEEAAAQQQQAABBMIUIEBYNJxldtFAi/5g2tNkUskFNBjjN1THtIPkUC379w+1rdC6o48+WubNmyc1a9aUJk2auPu2bdtW1q5dK+vXr5eCzx/UwKEGv4YMGSJ6BWJJkgYjTTrYvGJRH2q4YduFauykk05yb8m+6qqr5MQTT5SBAweKPpMwgikWTtpHT5PVxp/yTMH2r7BEAAEEEPCkgGNGZZusS39y/AWztAqU7QJ1QZspIoAAAggggAACCBQUIEBYUGTfD5h2ULVjyhpQ0CWp5AIaHJxost/QOUgTtqnfELRNAzhpJttBdSUqtmzZ0t3/hBNOCBxXsWJFNzCYmZnpBg79GxYsWCB33XWXXHfddaLPKkywZJvxRtSusPmXKVNGRo0a5b7o5Z577ilsVy9vc8zgbF/W75kmDRRaWiAhgAACCCSMgG1G6s8FB63/Oanb/MkyBc0kBBBAAAEEEEAAgQICGQXWU33VNgA9TNbgFKn0ArZpYqLJuiwq6Q/xmtTe1kJpU3p6unz77bcHNPP0008fUKcV559/vrz55pvSs2dP0WMTKEXcrqi568te1PGGG26Q/v37S48e+scmoZPtG71aOr4yCwQQQACBxBRwzLA1h0r693y2yT1NdkzWZJnsmExCAAEEEEAAAQRSVqBMys5838Qts9DsT44pBP/A6K9nGZ6A/hBuF+NQ3UcDg2kmaznmKSsry716UG8t/vjjj2Pefyk6tM2xUbd75ZVX5Jlnnsk3TL0VOy8vLxmCg/55qaUm2/3kAwEEEEAg2QQcMyH9OU9/3tCyP1mmkOfLWiYhgAACCCCAAAIpJ5CRcjPeN2HLLDR4pUmDK/7k+AssSy1gmxaCbQtrUPeNa9IrBvXW2aFDh8ptt93mPmOvWrVqcR1TMTu3i7kfuxVPQL+zemWJ48tmQUKgdALfDrn27D3btv9jx+ZNbUrX0v6ja2Q2fT93w4YV1ouv3bS/lhICBwq8+Y+Pzt6xbdc/crfuiNj3r3aDGu9v35q74nK7X7J8/2wjp9ky2THZn/TfA8tk/bfB8WWzICGAAAIIIIAAAsknkEoBQsucPifoFPp/2AuqopiKAk2bNpUnn3zSnbq+xOTzzz9PRQbmvE/AMQvNJAQiIjDxmiue3vbHH0PaXXKp1GjSNCJtaiM5333bPydniUwYePGuXq+9eVvEGqahpBJ4dcS4pzeu3TKkc5/2YoJ6EZvb/Fk5/efNWiwv3Tdm15UPnZtM3z+nAFJP37ptlo6vrIs8kx2TJ/qWWiYhgAACCCCAAAIJLZAKtxhb5gzp/wAPDzpTjilrJkVPoIdp2ole87SMQNQE9D8PrKi1TsMpJbB93bqIBwcVMLNbd9Gg466tW29NKVAmWyKBbZu2RTw4qANo2T5TNOi4Y/uuVPn+2QXg08y6Bgc1Oe7nvg/9edP25X01fCKAAAIIIIAAAgkikCoBQv2Fv2eCnBOGiQAC8RfQADcJgVIJfDagr93k+BPWR/LKweABabsm52VfefmpwfWUEVCBZ257025xVJP1RV05+P3kb+WtMW8E0FauWiG//Paz7Ny5M1AXqqDt1j6kZt4r9thU/f7ZxkVzcNKfN0Ml21RqtkwmIYAAAggggAACnhRItgChZZT1f2916U+2KTj+FZYxE7BMT07MeqMjBJJcoFrlCkk+w+hOr1yFSnEBLJORUTOaM6vRtGla7sYNnaPZB20nroB5vm6h379H/v2wXHzV+fLDtMnuJD/6dJwcf1In6TOgt/Q6s7v8sWZ1oZM3QcK0bVu28/3br+SYou3LZnFAGl6gxjbrVoE6VhFAAAEEEEAAgbgIZMSl1+h0aptm9aof/d9bx2QSAgggEI6AYw7S/2jwTGrcsmO7r36YK5ojkcqXy5Dlf2yQKXOWRKI58Xp7GlytXqtRg4hMlkZCCjz11nch68OpvOGxbLnxcSecQw96TFpa2kG3hbMh0u2FM4bSHvPnX2vl3889IW+/PEZOOK6r29yzL/xbzu9/kdx83S1y1/Db5LOvxsvlF19Z2q44/sArDdXE8sFo0FBzT9+6LiyTHZNJCCCAAAIIIIBAzASSLUAYMzg6QgABBGIlsGz+jB+vu/Za67Dm9SLS5dgJc+SwZvUlVdqbMmep/Dzz28IvhYqIbOo2MvSibhGbvAYbI9ne07eYN0rk5UVsfBocjHR7ERtcCRpauHih1K5VJxAcXLZ8qfw271d54uGnpeEhjaR/3/Pkv2+9QoCwBKYl3NUx+2sumCxToQHDbJM1aOiYTEIAAQQQQAABBKIuUCbqPdABAggggECpBTZtzS11G8ENpFp7O3O3RRYwGJMyAgkoULNGTSlXtmxg5D9MnyyVK1eRVlmt3bolS3OkWrXqge0UYibgmJ40MFjwslfL1GmkWwOHWiYhgAACCCCAAAIRFSBAGFFOGkMAAQQQQAABBLwvkNW8pWzasllm/DhNtudul3fee1NOsk6WMmXKuC8peevd16Vnt17en0hyj9AJmp6WNWg4IqhOi7bJGjjUpWUyCQEEEEAAAQQQCEsgmW4xDguAgxBAAAEEEEAAgVQT0EDgNZcPln4XnxWY+n23D3fL+pKSo9u2kzNO3b8tsBOFeAs4BQZgm3Wts0wOTnbQSnA5qJoiAggggAACCCCwX4AA4X4LSh4WqFmz5vwFCxa0zMrK8twozbikVq1aCzw3MN+AmjdvvtBxnBaWZXlyiGZs0rRp0wVLlizx5PgYFALRFvhmylRZvGK5XNmvn9vVstWr5a8NG80zIpuZF8CUi3b3tJ/CAjdff6t0Pb67LF22RNod2V6aN2vhakzNniX16/FenwT6ajhmrJqDk21WNGuyTHZM1mS7n/s+gstB1RQRQAABBBBAIBUFyqTipJlz4gns3Llzq3ko/BYvjnzhwoVStWrVH7w4Nh3T1q1bt3l1bDquiRMnyiGHHOJZPy/bMbbEFxjx7H+kz3XXy3czZrqTefeLL6VlnzOk80UXS7v+58qqtWsTf5LMwLMCemvxxk0bZNKU7+XBf9nyxLOPyrr16wgOevaMlXhgtjlCs2NywdSjQIVl1u0CdawigAACCCCAQAoJECBMoZOdyFPdvn37sGeffbaKF+fw+eefy7Jly17y4th0TLm5ubf07NnTq8MT27blhx9+8KyfZ+EYWMILrPlrnYx88SX5/Pn/yKt/e8idz79eNm+N7Xu2zP/0E8lq0kQ+mPBNws+TCXhTYPee3XLJVeeL/ff7pYZ5YUnrrDby+VefSp/+J8mmzZu8OWhGFQkB2zSiueAPBpap06TPMwxOtlmxgisoI4AAAggggEByChAgTM7zmnSzeu6555zGjRu/9thjj23w0uQeffTRdQ0bNnxJx+elcQWPZePGjV936NDhlY4dO64PrvdC2Yxr3dFHH63BQccL42EMCMRSYN6SHKlbq5ZYxxzjdpuzYqXMmT9fbrjoQmncoIFccuYZMubLL2M5JPpKIYEFC+fL9FnTZPRrY0WfPXjXLffKuLfHy85du2SC81UKSTBVn4BtlprTTC6Ysk1FwcChVXAn1hFAAAEEEEAgsQUIECb2+Uup0d9///2X79ixY/SwYcPkiy++EH32XzyS9qtZx7F3794PzBVwV8VjHCXpc+bMmVeYsY5LS0uTESNGiD73L15J+9asY1G/2bNne94vXlb0m9wCtarXkHJlMwKT/G7GDKlauZIc3ry5W7dw2TKpUbVqYDsFBCIpsGPnDqlcuYrUNFcP+lOFChXk5BNPcW8z9texTHkB2who0DA4cGiZ9eEma9BQg4ckBBBAAAEEEEgCgf2/mSTBZJhC8gs89NBDg+++++6Pfvnll/OmTp3aZd26dVmxnrW+kGTXrl3bzBsgbzLBQSfW/Yfb348//niF+eXvdXNL9JWvvfbacYsXL465nY5dX0hibhnX5yLeZIKDjtaREEhFgTbNMmXTli3yw+yf5KjWreTlcePk9O7dRd8u++Pvc+WlsWPljkGDUpGGOcdAoGWLVm4vD/3TlvP7XyR169STmT9Ol48++1DeeundGIyALhJYwDFj16zJcj/3fWhZs20yCQEEEEAAAQQSTIAAYYKdMIYrMnLkyE+Ng2ZSCQXM8wizzfP+4vq//QnytmLH0Fom65KEQFQENBB40yWXiDXoikD7Dw+72S3rS0o6HXG4DDi5d2AbBQQiKVCpYiU3EPjCq8/JmeedGmj6wfv+7r7ROFBBAYHCBZygzVoOXg/aRBEBBBBAAAEEvC5AgNDrZ4jxIYAAAggkrcB9114jvY47ThavWGECgkdIq8ym7lwXf/GZHFK3btLOm4l5Q6Ddke3lmUeflyf+8bSsXL1CGjZoJKv/WCWbzUtKqlat5o1BMgoEEEAAAQQQQACBmAgQIIwJM50ggECCCUxMsPEy3AQV2JabK39t3CjfTJ0q73/1tbRr01quO/88goMJej4Tadj6FuOvs/O/BOe3ub/KS/8dJcd0OE5OOK6LdGx/jFSsUDGRpsVYEUAAAQQQQAABBMIUIEAYJhyHIYBAUgs4ZnbDTdYlCYGoCJiX9EjfoTfJslWrpU/3blLPvNH402+/k1c//FCmvfOO1KrOFVxRgadRV8C89EueeObRAzSWrlgqCxcvlGdeeEruvuU+GXzlkAP2oQIBBBBAAAEEEEAg+QQIECbfOWVGCCAQGQHLNKPZMZmEQMQFFixdJt9OnyELPhsvh9av77a/e88e6X31NfLJREcuO+usiPdJgwj4BSpXqiyffzDBvxpYjnzsIenZrZds27ZVPvjk/UA9BQQQQAABBBBAAIHkFiiT3NNjdggggEBYAo45aoTJehVhLJJlOrFj0RF9RFTAMq3Z4bZYvWoVqVq5ktSuUSPQREZ6utQx63v35gXq4lywTP92nMdA96EFLFNth95UvFq9zXjp8iWyZFlOIC9eskjmL5wnLZpnuc8mLF5L7IUAAggggAACCCCQ6AIZiT4Bxo8AAghEScA27WabrEvN0U4ajOxs8g8m2yaTEkNAz1umyTkm2yYXO9WvXVuOPfJIOXHQlfLWvx6WZo0ayV2PPyEfOxPlsTtuL3Y7Ud7RMe2/YrLOU4Pmtskk7wiE/f3T4GDX3sfKKvNSkoLpi68/c6tmfvez1K5Vu+Bm1hFAAAEEEEAAAQSSUCAjCefElBBAAIFICWhAJNvXmB2pRkO045i6DSaf4ssECg1EAiTHjDHH5IEma8o0Ocdk2+RipQ+efELGfPGlbN2+3d2/ZrVq8utH46RxgwbFOj5GOw0y/eifAw1GaSZQaBA8kBwzhhyTw/r+rV69SjZt2Sw//u9XqVChgmlmX7rbvl1O7H6S9D7xFF5Q4kdhiQACCCCAAAIIpIAAAcIUOMlMEQEEwhZwzJFpJtsm55msgRFNji9rOVLpHNNQtq+xgoFCqVZ5/y/wkeowldopV6FStAD9wTPlLBioKZJ43pIl8tuiRVKtShU5rHlzGXByb/dKwiIPjO0Ojukux+RMkzUN92X982CbTIqfQNjfv4aHNJLP3v9Kdu7c4Wb/FG678U6pWbMWwUE/CEsEEEAAAQQQQCBFBNJSZJ5MM/YCGkzh+xV7d3qMroAG8KzodnFg61fY7+VWqlY7YgGu8uUyZMfO3Qd2FGaN19vT4Opvs3+QD54dFuYMwzvs3s7HOJdeeKGV2a17yAZ27Nwph/Y6yb3NeMIPU0SvJvwwO1tm/PKrfPXiKKlRtWrI44Irc777Vh56fpS88evvwdUxLd/wmP6xSJ309C09E2KyV515k3PB+RdYLdtnhhzv5s2bpG3n1iG3jXrqFTml16khtwVXzp+VI48/+4iMnxS1l5k4vv4m+pa2b8kCAQQQQAABBBBAIMICXEEYYVCaQwCBpBSwzKz0qilNI0x2tGCS435G5sM2zdxpsj8QqLccazpn5eI5w6+79lrrsOb19tWU8nPshDlyWLP65oq11Ghvypyl8vPMnT8YtuNLSRfqcNtUDjQ502RNOe6nyKCuhzayTFlzyPTzggXSoE4dGf/sM/LmJ+NlzJdfyssPPiDdBw6S72fOkjN6hA4sFmzs4sPbjDABQrtgfYTX80K0p38WbJPzhl7ULcTm8Kqeeus78XJ7T99iJpwXiiO8+aalpemB7kcYLdjmmJDfv/atj7XMNs0hU9Wq1eSnyfsDy+s3rJPvJ38njz39L+nYrmPIY0JV9jmh3wgTILRDbYtQnWXa0axJ4f3fO7eCDwQQQAABBBBAAIHICBAgjIwjrSCAQPIKZPumpr+UOlGc5nBf24HAYFB/wzdtzY1o16nW3s7cbZEF3H82/Octx1c1yCwdX9nyLUMu2jRrJqv//FM2b90mR7VuJf8ZPdrdr23LLFm7bl3IY+JUaZt+9XtZw9c/ARofhAcWYX//dOzVq1UPTEHLmU2ayZcTPpPvJn0r55zZP7AtzgXH9K9Zk+3LeWbZ02THZBICCCCAAAIIIIBABAQIEEYAkSYQQCBpBWwzs4km6zKayfY1rkEYfRah41tn4W0B2ze8HLMMDgz6qgtfVK5YUfr37i2nDR4sp/foIdPNrcX3PvmUeWnJF3Ld+ecVfnBst/qDUAQGY+teVG+2b4ccsyzx92/Xrl0yZty+oLSvHflr3V8y8X+OXHHZNf4qLy5t36D0e+n4yiwQQAABBBBAAAEESilAgLCUgByOAAJJK2CbmekvoGkxmGEP00dPk50Y9EUXkRMo1Xlb/scf8uq4D+VMq4fM/PVXd7lo+XJ59W8PSdusrMiNsnQt2eZwAoOlM4zW0aX6/u3avUs+HP9BYGw7zMtKZs2eIYe1Oly6du4aqPdowTbjyjZZl5pJCCCAAAIIIIAAAqUUIEBYSkAORwCBpBXQX741MBKLpMFBUuIJlOq8Va9SVd5/4jE5vXvxnjUYJx47Tv3SbdECpfr+VapYSUa/mv/lIhs2bpATz+gmP875UTq1P6boEcR3D/37WYOEji+bBQkBBBBAAAEEEEAgXAEChOHKcRwCCCSzgOWbnJ3Mk2Ru8RUoWzZDvp0+w80HG0m7Nq3lwj59DraZegQiKlCjeg1pe1hb+X3eb4kQIHTM5DWTEEAAAQQQQAABBCIgQIAwAog0gQACSSdgmRlNTLpZMSFPCeibcHft3l3omHbv3lPodjYiEK7Atu3b5MohA/MdvvqPlbIoZ5Hcfet9+eo9vKJXEVomOyaTEEAAAQQQQAABBEohQICwFHgcigACCCCAQLgCFcuXl8fuuF1+X7zYPIPwN9GA4dHmikEPPX8w3KlxXAIIlM0oK6f0OjXfSCtXriLtj+ogWc1b5qv3+Io+DoKEAAIIIIAAAgggUEoBAoSlBORwBBBAAAEEwhV49LX/um8uDj5+6MUXyT9vvSW4ijICERcoW7asXHTupfL95G/lyCOOkrp16snuPbslI50fDSOOTYMIIIAAAggggEACCPBTYBxO0m+fXWClZZR/ND29bLVduZuyYj2EsuWrz0/PqDhpx7Y1rx522jtOrPsvbX+DBw+2KlSo8KjJ1datWxdzv5o1a86vUqXKpBUrVrz63HPPJZqfVadOnSfM+Cvl5OTE5RKRzMzM+bVr1540Y8aMV813wat+sXxBiWEgpaLAqrVr3eDgh/9+SipVqCBPv/W23Df4Gjnm/Aulb68T5YR27VKRhTnHSECvWD3lnJ7uLcXa5afvfSXPv/IfqVenrtw57F7RAGICJMeMUV9UEpXEz2tRYaVRBBBAAAEEEPCoAAHCGJ+YRRNveDF365orm7Tp6/ZcqXrjGI9AZNvGZS23blzSOHfr6oGLJg59rXmPpy6P+SDC7NC27RdXrlx55aBBg9wWsrJiHh+UBQsWtJw3b17jZcuWDXzwwQdfu//++xPC79hjj31t6tSpl73zzjvb0tPTK1mWFeZZKN1hjuO0nDBhQmMTIBzYsWPH18wyIfxKN2uORuBAgUXLl0vdWrXklC4nyP9mzXJ3OLJlS/etxr8uXESA8EAyaiIo8Pu8X2Xjpk0yb2aOPPPCU/LuB+/Ifbf/n5x1/mlycq/T5NiOx0Wwt8Rrip/XEu+cMWIEEEAAAQQQKJ1AmdIdztElEdAfNtPKpJ97WNc7RAOD8QgO6ni137pNulbQcUianL0ge/BzJZlHvPbV4GBGRsa5jz/+uGhgMB7BQZ279tunT58KOg6Tzr7vvvs873fEEUeM2rt379l6xUivXr3iFhxUMA1MmsBqBR2L+rVp08bzfjpQEgKRFmjRuLGsXbdOfpj9U6DpeTlLZPy330rrzMxAHQUEoiFQr24Dt9ly5cpJz+695Nfff5Z6devLST1Pdt9iHI0+E6VNfl5LlDPFOBFAAAEEEEAgkgIECCOpWUhbepuKXjmYefSl1QrZLeabmrW7rMbO7euunfvNVX1i3nkJOtTbivXKwWHDhnnK75ZbbqmxZs2aa++++27P+png4Jm//PLL1dOmTateAvKY7Dp9+vQav//++7XNmzf3rF9MIOgkJQUa1KkjI2++SaxBV8j23B3yYXa2HNWvv1x0eh/p2qF9Spow6dgJ1K5VW040gcHhf79PNm/ZLFNnTJFvJn4tE7935PA2R8RuIB7riZ/XPHZCGA4CCCCAAAIIxEyAAGGMqMtXrH1x3SZdYtRbybppeuSFkrZn199KdlRs965fv/7Fp56a/22LsR3BwXsbMmSI5ObmetbPPG/wXHP15cEnEOct2SYosmnTJs/6xZmH7pNcYNhll8rP48ZK25ZZ8tY/H5Zpo9+Wlx98QNLS0pJ85kwv3gIbNm6QMeNGy2tvvSyXXn2BO5xB118qp5x0mhx1xNHxHl7c+ufntbjR0zECCCCAAAIIxFmAZxDG6ATs2bvDqlS9SYx6K3k3ZdLLVSn5UbE7YseOHVaLFi1i12EJeypfvrxn/cwVjl169NB3bng3Va1atcqff/7p3QEyMgSiKJDVZN+/Df1O6hXFXmgagfwCO3fukBuvvVluG3pn/g0pvsbPayn+BWD6CCCAAAIIpLAAVxDG6OTr24rj9czBoqao49qZuzH2b/soamBB2/VtxfF65mDQMEIWdVzxeJtyyMGEqFy8eHHzeL2QJMRwDqjSsZkxevr7d8CgqUAAAQQSXKBihYoy48fpsmLl8sBM9uzZIy/+d5Q4330TqEu1Aj+vpdoZZ74IIIAAAggg4BcgQOiXYIkAAggggAACCKSIQAUTIDRvtJcTeh8j7384RpYuXyIXDOovDz48XGrXqpMiCkwTAQQQQAABBBBAwC/ALcZ+CZYIIIAAAggggECKCJQtW1ZeH/W2vPfhu3LLPUPdWXc7oYdM/nq6NDykUYooME0EEEAAAQQQQAABvwBXEPolWCKAAAIeFqhWuYKHR+f9oZWrUAlA758mRhhjgT//WisTnK/cXptnNpclS3Nk5eqVMR4F3SGAAAIIIIAAAgh4QYArCL1wFhgDAgggUIhA45Yd2331w1zRHIlUvlyGLP9jg0yZsyQSzYnX29PgavVajRpEZLI0ElLgqbe+C1kfbqWX27vhsWy58XEn3KkdcJy29/QtPQ+oj3bF9tzt0qNPF8lqliXffPKdNDm0qTz53GPS/5KzZPSrY6XzMcdHewi0jwACCCCAAAIIIOAhAQKEHjoZSTYUJ8nmw3RSS8Ay0439b+wHMV42f8aP1117rXVY83oH2aNk1WMnzJHDmtWXVGlvypyl8vPMb1eXTCkye+ft3bvDtFQ+Mq2FbqVshQpxvzpy6EXdQg8ujFoNDqZae2EwFeuQvYV8//Ly8uS6K4bIdVffIBnp+34cvO3GO8XqeqIJ+hf/K1u2XLm4f/+KhcFOCCCAAAIIIIAAAoUKcItxoTxsDFPACvM4DkMAgYMIbNqae5At4VWnWns7c7dFFrAY7BWq1/hh3eJFO4uxa9i77N29e316ubLFj+aE3RMHJppApSoVf1i7ct1Bv3+VKlaSGwffHAgO+ufXqf0xcuQRR/lXC12atx6vL1u2TLy/f44ZpFXoQNmIAAIIIIAAAgggUKQAAcIiidghDAHLHDMxjOM4BAEvCFhmEI4XBsIYElug50uvfr555cqqG5ZG5lbughra7tLJk2r2fPn1YQW3sY7AILvf5xvXbq761+oNUcHQdhf+tLTmoBH9+f5FRZhGEUAAAQQQQACB2AoQIIytN73FQODnn3+W7t27y/LlywO9Pfzww25dbu7+i4iGDBkijz76aGAfCkULbN68WX777TfZvn170Tsn7h6WGToB7sQ9f54aeaW69V788Y3XZcn33+VFcmAaHNR2K9aq9Uwk26Wt5BKoWrPKiz98Okvm/5gT0e+fBge13UrVKvH9O8hXxtzeLYtyVsjKVWsPsgfVCCCAAAIIIICAtwR4BqG3zsdBRzPmg6/lvIF3yRfjnpaTT/z/9u4FOqr6TuD4LyaEEBNDeFZWMDxku4oSXr5QmXRbCL4fK7qwLWZVHutuKbo9LV2Ei9pqtwLaAk09hcJWqLbUyh7rsoJmYH2gCIIoByRIEIQDbgOICS9J9v+7ZrITMgmZmXvv3Jn5/s/5c+/87//+H587J7n8ch9X2vUqP94jFxXfJk/99CGZMvnvW9w3ARtmmj4zEtCv3WX//v3t5datW+WCCy4Qfc7Sm2++aZd9+OGHMmTIEDlx4oRs2bJFbrvttkQNM6n6rayslLFjx8r69esbx/2Nb3xDFi9eLD179mwsS5GVEWYeBAhT5GAmehrX/fKZ+//r72799NAnVQN2/WTNHU6NJ/vcvA9yu3efft38X61wqk3aST2Bb0+/+f75/7r007/sOzTgo427HPv+te+Q/UFB17zp4354sx++f7PMkdPzjqAfjqCeczwxZ7E8Pvs3cvSLWntI+Xm58rPHvicT//F2PwyRMSCAAAIIIIAAAhEFCBBGZPFf4Z23fVPGrPiWfPv+GbJtw3LJzztX7plkybVXD5J/nnCXnwZsmcHoyXo0KWAqa7ZMjjtlZ2fL5ZdfLhoMHDlypOzatUsOHTokN9xwg7z77rt2gHDHjh12P5dd1rbnLMU9KPcasBqaDi0d72n//v0yePBgKS0tlSeeeMK+EnPPnj0yefJkGTp0qOzcuVPy8vIc7zdBDaqjBgd1SULAEYHRy1+0HGmIRhCIQeCBJ8dZMeyWTLsEzWArTA6YHDS5tRQwGzVbJruSZvy4XJ5e8DtZtujHMupvrxK9knD5ilflH+57WI6Zuxi+909jXem3hUYDplyzZTIJAQQQQAABBBBoVeCcVrey0VcCC+b80B7Pd7//pDy1YJm8sW6zLPnVLMnM9M1htMwA9a/4uowmBU3l8SbXm2yZHHfSgNZ7771nt6PL4cOHy1VXXSXr1q2zy/Q22R49eki3bs68FTbuAcfegGV2VXPH7M4cSnl5uXTt2lWWLl0qetVgVlaW9O7dW5599lk5ePCgLFu27MxdkvVzwAxcLS2TSQgggAACySMQuorwbCMOmgpTTHbld+ahw5/LY/++UOY8/qDcWHqttGuXJe3bZ8u4MaPl4R/cJzMeK7fvajjbIB3cHjRtjTfZlfk6OE6aQgABBBBAAAEfCPgmsuQDC98PoXOnAllcbsmzz78s35/+tCxaMEN6X9jDD+MOmEFUmDzC5BKTY0llDTvNNMu4T2QHDhwoVVVV8sUXX8jbb78tw4YNEy3Tq92qq6tFn1N4xRVXxDJOP+6j/zHSNN7kuO3slsL+0duKR40aZf6j0y6sVKRLly52wFCDrSmQLDOHCpNj/f6mAAFTQAABBJJWwGoYeWjZ2kRCzxaZaSo5+jtz+47ddr965eCZSR8Po7ccf/a/h87c5PZnR8+v3B4s7SOAAAIIIIBA4gQIECbOPqaer7mq2NxenGvvW/rNq2Nqw4GdAqaNUNagiuY1JmtwJWhyLClodqoK2zGuE/fQcwg3b95sXzVYXFwsBQUFouV6RaEGDbUsRZLVMI+ihuV4s6w32Wr4HNfiwIEDtl2kRgoLC0VfXJJkKWDGG8qWWdfvr6YMk4O6QkIAAQQQSDoBPQcZYbLVkANmGSkFTWH4q51nms+O/M78S/URu7+OHfPtZfg/55zz1Sl3Te3/vywtfLuL60HTdlVY+47NN6xNVhGMG79LAAAXbklEQVRAAAEEEEAgBQQIECbZQfyRNc8Ea/Kkf79eMuUHTzo9ej1BbkuuMPVCOdAwiJlt3Le19osa2gpfhNoNLzvrul7tprcU622wGsTSW2I16VWDy5cvl9raWvuKQi07ffq0zJ07V2655RZ5+umnZcKECVocS2ptbm5vCx9vUcOHmOzCG9J1fQnJq6++emax/Vmf5dirV6+I22IodNso1L66hLIOc5bJlq6QEEAAAQSSWiAUJNSf8RUm17eQO0aYpe5Tv35L7G8d7vG1rnazb6//oFnzr7+1yS77q/O/qtOsQnQFLc2rpfKiCM3b8zXlgQjbKEIAAQQQQACBNBQgQJhEB33tGxtl3jO/l2d+Pl0Wzp8h+mbjF18KOjmDDNNYW7OehM9qyDoGXW/rvi3V03bOTKF2zyw/62d9DqG+qESDghkZ2qXYL9vQMg16derUyS7buHGj6G2y8+fPt6+U27Ztm10ewz8tzcuL8kjDjdkuvDF94YveZqwvK9FUU1MjDz30kLzyyivy/vvvy0033RRePZ51L5y0j5KwbJn1oMkkBBBAAIHkF6homELo53xLv1cizXSWKcwYdmnsAbyL+vW0233lta+edxzeyeqKtyVw7RDJzm76uI7wOlGstzSvlsojNW3P12wIRtpIGQIIIIAAAgiknwABwiQ55rXHjkvZ5Fky9s5SGf2tq0VvNb73O7fIxCk/EX0odgJS0PRpNWQ9IdWkf7kO6EoMyTL7hN/yEzpx1fKYkj5zUJO+aTeULrnkEnv1yiuvDBXJhg0bZPTo0XLBBRfYVxE2bkieFcsM1VG78KlPnDjR/jh+/HjZvn27nDhxQlavXm0/l/COO+6QQYMGhVdnHQEEEEAAgUQIaHBwjcklJgcbslk0S5Yp+TKsNO7zjVBbeefmijVtgvzs6f+wnxf9+dEaqT70ufzkyUWyygQIrWlf/T4N1fdoaZl+XDtH8GgOdIMAAggggAACHghkedAHXTggMOvxZ+wHW8994sHG1n76yHfl9y+skh/Nmi+/nDutsTxBK1ZDv3rLSrBhPZqF7qdJT9QtXYk3ff3rX5e1a9c2aSYnJ6dZmV4ZN2DAALtednZ2k/pJ8sFxu/B5d+7c2b5S8O677xY1DaWLL77YvopQ3wwdHnANbWeJAAIIIICARwKW6UeDg7o8W3L1d+a/ff9eu/9v3z+jcRzdunaS/35xnoy4ZnBjmYcrrs7Xw3nQFQIIIIAAAgi4LECA0GVgp5rXYKDm8KRvNf58X9MAWPj2BKxbps8Kk3Wpua3JMhUdCwy2tdNQPQ1u6ctMrrnmGtm06atnBIW2JcHSMmN03e7SSy+1b9fWF5bs3bvXvkW7W7dusnTpUunQoUMSMDFEBBBAAIEUFtAgWEYb5meZOq7+zszKypSZ5irCaQ+VyUeVn8h5550rvS74WhuG5koV1+fryqhpFAEEEEAAAQQSIkCAMCHsKd2pnnhrkDDYkM3irMk6aw0XK+jLTFauXCllZWX2C0tc7MqNpi03Gm2pze7du4vmUBo3blxolSUCCCCAAAKJELBMp3ru0ZZktaWSE3X0WYMDLu7rRFPxtGHFszP7IoAAAggggEB6CRAgTK/j7cVsg6YTzUmTOnbsaL+9+NChQ/bVcCNHjkyasTNQBBBAAAEEEEAAAQQQQAABBBBAIF4BXlISryD7RxLQv+QHIm3wc1lhYaFkZmb6eYiMDQEEEEAAAQSaCowwH4NNi/iEAAIIIIAAAgggEK0AAcJoxajfVgE9YU+61K5dO/nzn/+cdONmwAgggAACCCCAAAIIIIAAAggggECsAgQIY5Vjv5QVyM/PT9m5MbHkFTjv3BxHB59u7TmKR2MIIOAngYAZTNBPA2IsCCCAAAIIIIBAMgrwDMJkPGr+H3PQDFFfVEJCAAEHBPIKunRctW67aHYitc/Okr0HDqdNe04HQ504BqnWxs+X/Y+jU0q39hzFozEEEEAAAQQQQAABBGIQIEAYAxq7IIAAAl4KLP/5vwxyur/vza0oempqSZVT7fq9vevvsYqdmivtNBWY92BJRtMSPiGAAAIIIIAAAggggECyCXCLcbIdMcaLAAIIOCDgZHBQh+P39l5ebG1ygI0mEEAAAQQQQAABBBBAAIGUFCBAmJKHlUkhgAACCCCAAAIIIIAAAggggAACCCDQNgEChG1zohYCCCCAAAIIIIAAAggggAACCCCAAAIpKUCA0KPD2q59wY7aI3s86i26bnRc7XLOq4xuL29rFxYW7qis9OcQdVydOnXy5+DMYerTp8/OYDDo7QGLojcd24UXXuhbvyimQlUEEEAAgSQX4HwtyQ8gw0cAAQQQQACBmAUIEMZMF92O9XUna0TqvohuL29q1x75RNpl56/zprfYejl58mRNfX29L/127twp+fn+9aupqamNTd2bvdasWSPnn3++r79/3kjQCwIIIIBAogU4X0v0EaB/BBBAAAEEEEiUAAFCj+RPf3li6u4tz+d51F1U3Xz2yRty7Oj+hVHt5HHlY8eOTV2wYIEv/VauXCl79uzxrd/x48cfLCkp8fiItb07y7Jk3bp1vvVr+0yoiQACCCCQ7AKcryX7EWT8CCCAAAIIIBCrQEasO7Jf9AIfr/nuYsmQW3oXf6dj9Hu7s0fVpiXV5sq8P/UZMe8+h3uoN+05+v169NFHF5s2b3nwwQd94zd79uzqurq6P5kgl9N+jh6OwYMHLzIN3rphw4ZCRxuOszEzLttv8+bNvvaLc5rsjgACCCDgnoDj5xtpdr7m3pGhZQQQQAABBBBIKgFHAzhJNfMEDbayYlL5yWPVE7v2Gi65Bb1M7un5SELPQty95XeSc263hS4EB3VOjp+wa6PTp08vP3jw4MTS0lLp27ev9OvXT4s9TaFnIc6fP1969Oix0O/BwRBOcXHxok2bNpXpFXsjRoyQQCAQ2uTpMvQ8RL2qceDAgQsJDnrKT2cIIIBAqgm4cr6RRudrqfZ9YD4IIIAAAgggEKMAAcIY4eLZbftr913fLjN3TN3pY8NPHj/ieYRLX0hSd/rL2rovj0/5m9HPBeOZSyv7unLCrv1Nmzbt+tzc3DHm2XrDq6urPffTF5KcOnWq1vQ/pby8PNiKge825eTklJhA4b0HDhy4YteuXZ7bKYi+kMTcMl5rAr1TzMeglpEQQAABBBCIUcC18400OV+LkZ3dEEAAAQQQQAABBBBom4CesJMQQAABBBBAAAE3BTjfcFOXthFAAAEEEEAgbQR4SUnaHGomigACCCCAAAIIIIAAAggggAACCCCAQHMBAoTNTShBAAEEEEAAAQQQQAABBBBAAAEEEEAgbQQIEKbNoWaiCCCAAAIIIIAAAggggAACCCCAAAIINBcgQNjchBIEEEAAAQQQQAABBBBAAAEEEEAAAQTSRoAAYdocaiaKAAIIIIAAAggggAACCCCAAAIIIIBAcwEChM1NKEEAAQQQQAABBBBAAAEEEEAAAQQQQCBtBAgQps2hZqIIIIAAAggggAACCCCAAAIIIIAAAgg0FyBA2NyEEgQQQAABBBBAAAEEEEAAAQQQQAABBNJGgABh2hxqJooAAggggAACCCCAAAIIIIAAAggggEBzAQKEzU0oQQABBBBAAAEEEEAAAQQQQAABBBBAIG0ECBCmzaFmoggggAACCCCAAAIIIIAAAggggAACCDQXIEDY3IQSBBBAAAEEEEAAAQQQQAABBBBAAAEE0kaAAGHaHGomigACCCCAAAIIIIAAAggggAACCCCAQHMBAoTNTShBAAEEEEAAAQQQQAABBBBAAAEEEEAgbQQIEKbNoWaiCCCAAAIIIIAAAggggAACCCCAAAIINBcgQNjchBIEEEAAAQQQQAABBBBAAAEEEEAAAQTSRoAAYdocaiaKAAIIIIAAAggggAACCCCAAAIIIIBAcwEChM1NKEEAAQQQQAABBBBAAAEEEEAAAQQQQCBtBLLSZqZMNGUEJk2aFMjJyZlt8nnV1dX9vJ5YYWHhjry8vDc//fTTxeXl5UGv+4+zv0CXLl2eMuPPraqquijOtqLevaioaEd+fn5wy5YtE6LemR0QQAABBBBAAAEEEEAAAQQQQMAVgQxXWqVRBETqDYLj3y/Lsn69b9++ex944AHbuF8/z+ODUllZKR999NHxVatW5fTs2XPJww8/fE8yHPDLL798yTvvvPOd1atX12ZmZuYGAgHPhx0MBqWioqL+kUceyRgwYMCvP/jgg/s9HwQdIoAAAgikkoAr5xupBMRcEEAAAQQQQACBtgg4HsBpS6fUSQsBx0/YNTiYlZV159SpU8/zi+CcOXMOnzhx4vnHHntskl/GFGkcl1xyyTMdOnQYs379+oJI2xNRNmzYsCPG7vdcTZgIffpEAAEEUkbA8fONlJFhIggggAACCCCAQBQCBAijwKJqVAKOnrDrbcV1dXUVc+fOjWoQXlQ2AUvp3LnzDY8//vjLXvQXbR8mOHjThx9++J/19XpI/JP0asKSkhIdkP4T1BUSAggggAACUQo4er4RZd9URwABBBBAAAEEUkaAl5SkzKFM7Yl07959XGlpqS8nqbc7Hz9+/Me+HJwZlHne4J3m6kvfDU9vcZ4xY8bpgoKCn/lucAwIAQQQQAABBBBAAAEEEEAAgTQSIECYRgc7madqbkUN9O3b17dTaN++fZ5fB3fw4MHhI0aM8OXwzBWEmR07dvTNLeO+RGJQCCCAAAIIIIAAAggggAACCLgsQIDQZWCad0ZA31aciBeStGX0Oq5EvE25LWPTOrt27eqTiBeStGV8Oq7du3f3b0td6iCAAAIIIIAAAggggAACCCCAgDsCBAjdcaVVBBBAAAEEEEAAAQQQQAABBBBAAAEEkkKAAGFSHCYGiQACCCCAAAIIIIAAAggggAACCCCAgDsCBAjdcaVVBBBAAAEEEEAAAQQQQAABBBBAAAEEkkKAAGFSHCYGiQACCCCAAAIIIIAAAggggAACCCCAgDsCBAjdcaVVBBBAAAEEEEAAAfcFgu53QQ8IIIAAAggggEDqCxAgTP1jnKgZBk3HgUR1Tr8IIIAAAgggkPICgZSfIRNEAAEEEEAAAQQ8EiBA6BE03SCAAAIIIIAAAgg4KhAwra1xtEUaQwABBBBAAAEE0lSAAGGaHvhUnvapU6fkuuuuk5dffrlxmitXrrTLNm/e3Fi2aNEiufvuuxs/s4IAAggggAACCCCAAAIIIIAAAgikowABwnQ86t7MeZbpZqY3XTXtpV27djJ06FDZsmVL44Z3333XXt+4cWNjmQYLtR6pdYHJkyfLkCFDmlW69dZbpbS0tFk5BQgggAACCHgkoOcZlkd90Q0CCCCAAAIIIJDSAgQIU/rwJnRyQdN7oCGbRaspYLZardaIcqMGtN577z17r7q6Onn99dfl9ttvl7feessu06sMdfugQYOibNl31S0zIs2upTFjxogGVnfu3NnYx5EjR2TFihUyduzYxjJWEEAAAQQQ8FDAMn3pHyNJCCCAAAIIIIAAAg4IECB0AJEmWhRo61WEQdPCeJPrTbZMjjsVFxfLvn375PDhw3Zgq7a21g5mbdu2TT7//HP5+OOP7T4uu+yyuPtKcAOW6V+voHDM7sz5XHvttdKtWzc7IBjaprdsa7r55ptDRSwRQAABBBDwSsAyHenvPl2SEEAAAQQQQAABBBwQIEDoACJNtChgNWwJLVusaDaUNWx0JNjVv39/uzkNCG7YsMF+/qAGufr27WtfObh161bp0aOHdO3atbUxJcu20BUU482AHQ8UZmVlyT333CPPPfdco8cf//hH0SsLO3bs2FjGCgIIIIAAAi4LBEz7FSaPMLnEZBICCCCAAAIIIICAQwIECB2CpJkWBfQEXk/krYYcMMtIKWgKq8I2zDTr9SZbYWVtXtXnEF555ZWigcB33nmn8VmDWqbPI9TnE+p6iiSrYR5FDcvxZhmzXUMbTRZ33XWXrF+/XqqqqqSmpkb+8Ic/cHtxEyE+IIAAAgi4IBAwbYZyhVnXvMZkPbcImkxCAAEEEEAAAQQQcEiAAKFDkDTTqkD4X/lnmpoavIqUiyK0EqofYVPrRfocwjfffNMOCA4ePNiurEst06sKQ88f1LcZv/TSS/Z2fTbhhAkT5OjRo603fsZWbc+kSHPyqix8REUNH2K2C29M19VNr8rU5w6uWrVK8vPzZdSoUWdWi+ezV070k9jvKf748x3gOxDNd2Cm+cWiOWDyLJP1fMIymYQAAggggAACCCDgsECWw+3RHAItCVgtbQgr1/80nJn0PwSWyZG2nVm3yeeBAwfK/PnzpbCwUHr16mVvGzBggHz22Wf2euj5gxdddJE8//zzcuONN4q+2Tg7O9sOgDVp7CwfNBhpgoQZZ6nm5uZIPjHbRRro+PHj7duM1WvcuHGSk5MTqVqsZYm0i3XM7IcAAggggAACCCCAAAIIIIBASghwBWFKHMaUmIRlZnE4bCYa3NKgkRVWFtWqBrI0XX311Y37dejQQTQwWFRUZAcOdcOwYcPk/ffft19e8sYbb8jIkSMb6yfJimXG6ahdpHnrMwfXrVsnv/3tb0VvOSYhgAACCCCAAAIIIIAAAggggEBqCHAFYWocx1SYhd5CpEkDg5auxJsyMzNl7dq1zZqZN29ekzK9Ek5vl9VnFb722muyePHiJtuT4IPjdpHm3K9fPzuYunv3btE3G5MQQAABBBBAAAEEEEAAAQQQQCA1BAgQpsZxTPZZWGYCjgUGY8EIBALyi1/8QvSqQ70lOYmSZcbqmZ0GUUkIIIAAAggggAACCCCAAAIIIJBaAgQIU+t4JutsrEQPfOjQobJv3z4pKytL9FCi7T/hdtEOmPoIIIAAAggggAACCCCAAAIIIOAvAQKE/joejCZBAu3bt494O3KChkO3CCCAAAIIIIAAAggggAACCCCAgGcCvKTEM2o6QgABBBBAAAEEEEAAAQQQQAABBBBAwH8CBAj9d0wYEQIIIIAAAggggAACCCCAAAIIIIAAAp4JECD0jJqOEEAAAQQQQAABBBBAAAEEEEAAAQQQ8J8AAUL/HRNGhAACCCCAAAIIIIAAAggggAACCCCAgGcCBAg9o6YjBBBAAAEEEEAAAQQQQAABBBBAAAEE/CdAgNB/x4QRIYAAAggggAACCCCAAAIIIIAAAggg4JkAAULPqOkIAQQQQAABBBBAAAEEEEAAAQQQQAAB/wkQIPTfMWFECCCAAAIIIIAAAggggAACCCCAAAIIeCZAgNAzajqKR6CwsHBHZWVlPE24tq+Oq1OnTv4cnJl1nz59dgaDQdfmH0/DOq5evXptj6cN9kUAAQQQQAABBBBAAAEEEEAAgfgECBDG58feHgmcPHmypr6+/guPuouqm507d0p+fv66qHbysHJNTU2th91F1dWqVatqTfD3jah2ojICCCCAAAIIIIAAAggggAACCDgqkOFoazSGgEsCkyZNCtTV1VXMnTvXpR5ib3bq1KlyzjnnlJSXlwdjb8W9PQsKCr555MiRVSbA6l4nMbackWH/COLnUIx+7IYAAggggAACCCCAAAIIIICAEwJcQeiEIm24LqDBt549ey6ZM2fOYdc7i6KD2bNnV/fo0WOhX4ODOhUTHFw9ePDg3wwZMuRQFFNztareWqzjKS4u/o2rHdE4AggggAACCCCAAAIIIIAAAgicVSDzrDWogIBPBCoqKl4cPnx47xdeeGFow5Vn+uw/z0enzxysrq6WRx99VPLy8pZZlnWf54OIssP9+/ev6N69+19Pnjx5UMiuqKgoylbir66BwSVLlkhZWZnaLd66deuE+FulBQQQQAABBBBAAAEEEEAAAQQQiEeAW/vi0WPfhAhMmzbt+tzc3DHm2XrDTaCun9eD0BeSnDp1qtb0P8XPVw5GcsnJySkxV+3de+DAgSt27drluV3v3r0rjx49+oUZx7S9e/eujDRGyhBAAAEEEEAAAQQQQAABBBBAwFuB/wNmf0I1BBRSvAAAAABJRU5ErkJggg=="
+ }
+ },
"cell_type": "markdown",
"id": "5b04dff2",
"metadata": {},
"source": [
"## 2. Внимание (Attention)\n",
"\n",
+ "\n",
"\n",
- "\n",
"\n",
"Механизм внимания — ключевая идея трансформеров. \n",
"Он позволяет модели **взвешивать важность других токенов** при обработке текущего, то есть решать, на какие слова нужно обратить внимание при генерации следующего.\n",
@@ -341,7 +370,7 @@
},
{
"cell_type": "code",
- "execution_count": 5,
+ "execution_count": null,
"id": "8fe8d3bb",
"metadata": {},
"outputs": [],
@@ -406,7 +435,7 @@
},
{
"cell_type": "code",
- "execution_count": 6,
+ "execution_count": null,
"id": "d55276a9",
"metadata": {},
"outputs": [],
@@ -442,13 +471,19 @@
]
},
{
+ "attachments": {
+ "image.png": {
+ "image/png": "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"
+ }
+ },
"cell_type": "markdown",
"id": "3ffafb56",
"metadata": {},
"source": [
"## 3. Feed Forward Network (FFN)\n",
"\n",
- "\n",
+ "\n",
+ "\n",
"\n",
"После блока внимания каждый токен независимо проходит через двухслойную нейронную сеть — **Feed Forward Network**. \n",
"Она добавляет модели способность нелинейно преобразовывать информацию.\n",
@@ -463,7 +498,7 @@
},
{
"cell_type": "code",
- "execution_count": 7,
+ "execution_count": null,
"id": "84f57562",
"metadata": {},
"outputs": [],
@@ -484,14 +519,19 @@
]
},
{
+ "attachments": {
+ "image.png": {
+ "image/png": "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"
+ }
+ },
"cell_type": "markdown",
"id": "8d9ce9d8",
"metadata": {},
"source": [
"## 4. Блок Декодера\n",
"\n",
+ "\n",
"\n",
- "\n",
"\n",
"\n",
"Каждый слой GPT-1 — это **декодер**, состоящий из следующих элементов:\n",
@@ -515,7 +555,7 @@
},
{
"cell_type": "code",
- "execution_count": 8,
+ "execution_count": null,
"id": "300acc96",
"metadata": {},
"outputs": [],
@@ -589,7 +629,7 @@
},
{
"cell_type": "code",
- "execution_count": 9,
+ "execution_count": null,
"id": "0eb26ef3",
"metadata": {},
"outputs": [],
@@ -732,7 +772,7 @@
},
{
"cell_type": "code",
- "execution_count": 10,
+ "execution_count": null,
"id": "632eec77",
"metadata": {},
"outputs": [],
@@ -777,7 +817,7 @@
},
{
"cell_type": "code",
- "execution_count": 11,
+ "execution_count": null,
"id": "8003ea24",
"metadata": {},
"outputs": [],
@@ -826,7 +866,7 @@
},
{
"cell_type": "code",
- "execution_count": 12,
+ "execution_count": null,
"id": "dd700a5c",
"metadata": {},
"outputs": [
@@ -1032,7 +1072,7 @@
},
{
"cell_type": "code",
- "execution_count": 13,
+ "execution_count": null,
"id": "4afd7733",
"metadata": {},
"outputs": [],
@@ -1071,7 +1111,7 @@
},
{
"cell_type": "code",
- "execution_count": 14,
+ "execution_count": null,
"id": "71bb6b24",
"metadata": {},
"outputs": [
@@ -1117,7 +1157,7 @@
},
{
"cell_type": "code",
- "execution_count": 15,
+ "execution_count": null,
"id": "ccb9621a",
"metadata": {},
"outputs": [],
@@ -1132,7 +1172,7 @@
},
{
"cell_type": "code",
- "execution_count": 16,
+ "execution_count": null,
"id": "f1b82472",
"metadata": {},
"outputs": [