Spaces:
Sleeping
Sleeping
Update app.py for Gradio Space
Browse files
app.py
CHANGED
|
@@ -1,3 +1,4 @@
|
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
import torch
|
| 3 |
import torch.nn as nn
|
|
@@ -12,12 +13,12 @@ class HolographicMasterCodeTransformer(nn.Module):
|
|
| 12 |
self.d_model = d_model
|
| 13 |
self.n_harmonics = n_harmonics
|
| 14 |
self.alpha_0 = 1.0 / 137.035
|
| 15 |
-
|
| 16 |
self.input_proj = nn.Linear(input_dim, d_model)
|
| 17 |
-
|
| 18 |
encoder_layer = nn.TransformerEncoderLayer(
|
| 19 |
-
d_model=d_model,
|
| 20 |
-
nhead=nhead,
|
| 21 |
dim_feedforward=d_model * 4,
|
| 22 |
dropout=0.1,
|
| 23 |
activation='gelu',
|
|
@@ -25,9 +26,9 @@ class HolographicMasterCodeTransformer(nn.Module):
|
|
| 25 |
norm_first=True
|
| 26 |
)
|
| 27 |
self.transformer = nn.TransformerEncoder(encoder_layer, num_layers=num_layers)
|
| 28 |
-
|
| 29 |
self.harmonic_weights = nn.Parameter(torch.randn(n_harmonics, d_model))
|
| 30 |
-
|
| 31 |
self.output_head = nn.Sequential(
|
| 32 |
nn.LayerNorm(d_model),
|
| 33 |
nn.Linear(d_model, d_model // 2),
|
|
@@ -35,27 +36,27 @@ class HolographicMasterCodeTransformer(nn.Module):
|
|
| 35 |
nn.Dropout(0.1),
|
| 36 |
nn.Linear(d_model // 2, output_dim)
|
| 37 |
)
|
| 38 |
-
|
| 39 |
def forward(self, x, alpha=None):
|
| 40 |
if alpha is None:
|
| 41 |
alpha = self.alpha_0
|
| 42 |
-
|
| 43 |
h = self.input_proj(x)
|
| 44 |
h = h.unsqueeze(1)
|
| 45 |
h = self.transformer(h)
|
| 46 |
h = h.squeeze(1)
|
| 47 |
-
|
| 48 |
psi = torch.zeros_like(h)
|
| 49 |
for n in range(self.n_harmonics):
|
| 50 |
phase = 2 * torch.pi * (n + 1) * (alpha - self.alpha_0) * 800.0
|
| 51 |
psi += self.harmonic_weights[n] * torch.cos(h * (n + 1) + phase)
|
| 52 |
-
|
| 53 |
alpha_dev = torch.abs(alpha - self.alpha_0)
|
| 54 |
security_factor = torch.exp(-300.0 * alpha_dev)
|
| 55 |
-
|
| 56 |
h = h + 0.35 * psi * security_factor
|
| 57 |
return self.output_head(h)
|
| 58 |
-
|
| 59 |
def holographic_reg(self):
|
| 60 |
reg = sum(torch.norm(w, p=1) for w in self.harmonic_weights)
|
| 61 |
return 0.0015 * reg / self.n_harmonics
|
|
@@ -67,27 +68,27 @@ def train_model(alpha_value, epochs):
|
|
| 67 |
optimizer = optim.AdamW(model.parameters(), lr=0.001, weight_decay=1e-5)
|
| 68 |
scheduler = CosineAnnealingLR(optimizer, T_max=epochs, eta_min=1e-6)
|
| 69 |
criterion = nn.MSELoss()
|
| 70 |
-
|
| 71 |
torch.manual_seed(42)
|
| 72 |
X = torch.randn(300, 8)
|
| 73 |
y = torch.sin(X.sum(dim=1, keepdim=True)) * 0.9 + 0.05 * torch.randn(300, 1)
|
| 74 |
-
|
| 75 |
losses = []
|
| 76 |
-
|
| 77 |
for epoch in range(epochs):
|
| 78 |
optimizer.zero_grad()
|
| 79 |
pred = model(X, alpha=alpha_value)
|
| 80 |
loss = criterion(pred, y)
|
| 81 |
reg = model.holographic_reg()
|
| 82 |
total_loss = loss + reg
|
| 83 |
-
|
| 84 |
total_loss.backward()
|
| 85 |
torch.nn.utils.clip_grad_norm_(model.parameters(), 1.0)
|
| 86 |
optimizer.step()
|
| 87 |
scheduler.step()
|
| 88 |
-
|
| 89 |
losses.append(loss.item())
|
| 90 |
-
|
| 91 |
# Grafik oluştur
|
| 92 |
fig, ax = plt.subplots(figsize=(10, 6))
|
| 93 |
ax.plot(losses, linewidth=2, color='#FF6B6B')
|
|
@@ -96,9 +97,10 @@ def train_model(alpha_value, epochs):
|
|
| 96 |
ax.set_title(f'Eğitim Süreci (α = {alpha_value:.8f})', fontsize=14)
|
| 97 |
ax.grid(True, alpha=0.3)
|
| 98 |
ax.set_yscale('log')
|
| 99 |
-
|
| 100 |
final_loss = losses[-1]
|
| 101 |
-
|
|
|
|
| 102 |
return fig, f"Final Loss: {final_loss:.6f}"
|
| 103 |
|
| 104 |
|
|
@@ -110,74 +112,78 @@ def compare_alphas(epochs):
|
|
| 110 |
"Sapma1 (1/137.000)": 1/137.000,
|
| 111 |
"Sapma2 (1/137.070)": 1/137.070
|
| 112 |
}
|
| 113 |
-
|
| 114 |
fig, ax = plt.subplots(figsize=(12, 7))
|
| 115 |
colors = ['#FF6B6B', '#4ECDC4', '#45B7D1', '#FFA07A']
|
| 116 |
-
|
| 117 |
results = {}
|
| 118 |
-
|
| 119 |
for (name, alpha_val), color in zip(alpha_values.items(), colors):
|
| 120 |
model = HolographicMasterCodeTransformer()
|
| 121 |
optimizer = optim.AdamW(model.parameters(), lr=0.001, weight_decay=1e-5)
|
| 122 |
scheduler = CosineAnnealingLR(optimizer, T_max=epochs, eta_min=1e-6)
|
| 123 |
criterion = nn.MSELoss()
|
| 124 |
-
|
| 125 |
torch.manual_seed(42)
|
| 126 |
X = torch.randn(300, 8)
|
| 127 |
y = torch.sin(X.sum(dim=1, keepdim=True)) * 0.9 + 0.05 * torch.randn(300, 1)
|
| 128 |
-
|
| 129 |
losses = []
|
| 130 |
-
|
| 131 |
for epoch in range(epochs):
|
| 132 |
optimizer.zero_grad()
|
| 133 |
pred = model(X, alpha=alpha_val)
|
| 134 |
loss = criterion(pred, y)
|
| 135 |
reg = model.holographic_reg()
|
| 136 |
total_loss = loss + reg
|
| 137 |
-
|
| 138 |
total_loss.backward()
|
| 139 |
torch.nn.utils.clip_grad_norm_(model.parameters(), 1.0)
|
| 140 |
optimizer.step()
|
| 141 |
scheduler.step()
|
| 142 |
-
|
| 143 |
losses.append(loss.item())
|
| 144 |
-
|
| 145 |
ax.plot(losses, label=name, linewidth=2.5, color=color)
|
| 146 |
results[name] = losses[-1]
|
| 147 |
-
|
| 148 |
ax.set_xlabel('Epoch', fontsize=12)
|
| 149 |
ax.set_ylabel('Loss', fontsize=12)
|
| 150 |
ax.set_title('Alpha Değerleri Karşılaştırması', fontsize=14)
|
| 151 |
ax.legend(fontsize=10)
|
| 152 |
ax.grid(True, alpha=0.3)
|
| 153 |
ax.set_yscale('log')
|
| 154 |
-
|
| 155 |
# Sonuçları metne dönüştür
|
| 156 |
-
results_text = "**Final Loss Değerleri:**
|
|
|
|
|
|
|
| 157 |
for name, loss in sorted(results.items(), key=lambda x: x[1]):
|
| 158 |
-
results_text += f"• {name}: {loss:.6f}
|
| 159 |
-
|
|
|
|
|
|
|
| 160 |
return fig, results_text
|
| 161 |
|
| 162 |
|
| 163 |
# Gradio Interface
|
| 164 |
-
with gr.Blocks(title="OmegaCode: Holographic Master Code"
|
| 165 |
gr.Markdown("""
|
| 166 |
# 🌌 OmegaCode: Evrensel Kaynak Kodun Holografik Korunumu
|
| 167 |
-
|
| 168 |
Bu uygulama, **Ω_v57** teorisini temel alan Holografik Master Code Transformer modelini etkileşimli bir şekilde test etmenizi sağlar.
|
| 169 |
-
|
| 170 |
**Teori Özeti:**
|
| 171 |
- **Master Code (Ψ)**: Evrenin temel titreşim dokusunu temsil eden harmonik dalga süperpozisyonu
|
| 172 |
- **Alpha (α) Güvenlik Kilidi**: İnce yapı sabiti değerindeki sapmalara karşı sistemi koruyan mekanizma
|
| 173 |
- **Ryu-Takayanagi Entanglement**: Karadelik ufkundaki minimal yüzey alanı üzerinden bilgi korunumu
|
| 174 |
- **Holografik AI**: Yapay zeka eğitiminde kuantum hata düzeltme benzeri etkileri
|
| 175 |
""")
|
| 176 |
-
|
| 177 |
with gr.Tabs():
|
| 178 |
with gr.Tab("Tek Alpha Eğitimi"):
|
| 179 |
gr.Markdown("### Belirli Bir Alpha Değeriyle Eğitim Yap")
|
| 180 |
-
|
| 181 |
with gr.Row():
|
| 182 |
alpha_input = gr.Number(
|
| 183 |
value=1/137.035,
|
|
@@ -191,22 +197,22 @@ with gr.Blocks(title="OmegaCode: Holographic Master Code", theme=gr.themes.Soft(
|
|
| 191 |
step=50,
|
| 192 |
label="Epoch Sayısı"
|
| 193 |
)
|
| 194 |
-
|
| 195 |
train_btn = gr.Button("Eğitimi Başlat", variant="primary", size="lg")
|
| 196 |
-
|
| 197 |
with gr.Row():
|
| 198 |
output_plot = gr.Plot(label="Loss Grafiği")
|
| 199 |
output_text = gr.Textbox(label="Sonuç", interactive=False)
|
| 200 |
-
|
| 201 |
train_btn.click(
|
| 202 |
fn=train_model,
|
| 203 |
inputs=[alpha_input, epochs_input],
|
| 204 |
outputs=[output_plot, output_text]
|
| 205 |
)
|
| 206 |
-
|
| 207 |
with gr.Tab("Alpha Karşılaştırması"):
|
| 208 |
gr.Markdown("### Farklı Alpha Değerlerini Karşılaştır")
|
| 209 |
-
|
| 210 |
epochs_compare = gr.Slider(
|
| 211 |
minimum=50,
|
| 212 |
maximum=300,
|
|
@@ -214,45 +220,45 @@ with gr.Blocks(title="OmegaCode: Holographic Master Code", theme=gr.themes.Soft(
|
|
| 214 |
step=50,
|
| 215 |
label="Epoch Sayısı"
|
| 216 |
)
|
| 217 |
-
|
| 218 |
compare_btn = gr.Button("Karşılaştırmayı Başlat", variant="primary", size="lg")
|
| 219 |
-
|
| 220 |
with gr.Row():
|
| 221 |
compare_plot = gr.Plot(label="Karşılaştırma Grafiği")
|
| 222 |
compare_text = gr.Markdown(label="Sonuçlar")
|
| 223 |
-
|
| 224 |
compare_btn.click(
|
| 225 |
fn=compare_alphas,
|
| 226 |
inputs=[epochs_compare],
|
| 227 |
outputs=[compare_plot, compare_text]
|
| 228 |
)
|
| 229 |
-
|
| 230 |
with gr.Tab("Teori Hakkında"):
|
| 231 |
gr.Markdown("""
|
| 232 |
## Ω_v57 Teorisi Detayları
|
| 233 |
-
|
| 234 |
### Master Code (Ψ)
|
| 235 |
Evrenin temel titreşim dokusunu temsil eden harmonik dalga süperpozisyonu:
|
| 236 |
-
|
| 237 |
-
$$\
|
| 238 |
-
|
| 239 |
### Alpha Güvenlik Kilidi
|
| 240 |
İnce yapı sabiti α ≈ 1/137.035 değerindeki sapmalara karşı sistem kendini otomatik olarak korur:
|
| 241 |
-
|
| 242 |
-
$$\
|
| 243 |
-
|
| 244 |
### Ryu-Takayanagi Entanglement
|
| 245 |
Karadelik ufkundaki minimal yüzey alanı üzerinden kuantum dolaşıklığının geometrik temsili:
|
| 246 |
-
|
| 247 |
-
$$S_{
|
| 248 |
-
|
| 249 |
### Holografik AI Etkisi
|
| 250 |
Bu mekanizmalar, yapay zeka eğitiminde Loss Zeroing hızını %140 artırmaktadır.
|
| 251 |
""")
|
| 252 |
-
|
| 253 |
gr.Markdown("""
|
| 254 |
---
|
| 255 |
-
|
| 256 |
**Proje Kaynakları:**
|
| 257 |
- GitHub: [OmegaCode Repository](https://github.com/KULLANICI_ADI/OmegaCode)
|
| 258 |
- Teori: Ω_v57 Holographic Master Code Framework
|
|
@@ -260,4 +266,4 @@ with gr.Blocks(title="OmegaCode: Holographic Master Code", theme=gr.themes.Soft(
|
|
| 260 |
""")
|
| 261 |
|
| 262 |
if __name__ == "__main__":
|
| 263 |
-
demo.launch()
|
|
|
|
| 1 |
+
|
| 2 |
import gradio as gr
|
| 3 |
import torch
|
| 4 |
import torch.nn as nn
|
|
|
|
| 13 |
self.d_model = d_model
|
| 14 |
self.n_harmonics = n_harmonics
|
| 15 |
self.alpha_0 = 1.0 / 137.035
|
| 16 |
+
|
| 17 |
self.input_proj = nn.Linear(input_dim, d_model)
|
| 18 |
+
|
| 19 |
encoder_layer = nn.TransformerEncoderLayer(
|
| 20 |
+
d_model=d_model,
|
| 21 |
+
nhead=nhead,
|
| 22 |
dim_feedforward=d_model * 4,
|
| 23 |
dropout=0.1,
|
| 24 |
activation='gelu',
|
|
|
|
| 26 |
norm_first=True
|
| 27 |
)
|
| 28 |
self.transformer = nn.TransformerEncoder(encoder_layer, num_layers=num_layers)
|
| 29 |
+
|
| 30 |
self.harmonic_weights = nn.Parameter(torch.randn(n_harmonics, d_model))
|
| 31 |
+
|
| 32 |
self.output_head = nn.Sequential(
|
| 33 |
nn.LayerNorm(d_model),
|
| 34 |
nn.Linear(d_model, d_model // 2),
|
|
|
|
| 36 |
nn.Dropout(0.1),
|
| 37 |
nn.Linear(d_model // 2, output_dim)
|
| 38 |
)
|
| 39 |
+
|
| 40 |
def forward(self, x, alpha=None):
|
| 41 |
if alpha is None:
|
| 42 |
alpha = self.alpha_0
|
| 43 |
+
|
| 44 |
h = self.input_proj(x)
|
| 45 |
h = h.unsqueeze(1)
|
| 46 |
h = self.transformer(h)
|
| 47 |
h = h.squeeze(1)
|
| 48 |
+
|
| 49 |
psi = torch.zeros_like(h)
|
| 50 |
for n in range(self.n_harmonics):
|
| 51 |
phase = 2 * torch.pi * (n + 1) * (alpha - self.alpha_0) * 800.0
|
| 52 |
psi += self.harmonic_weights[n] * torch.cos(h * (n + 1) + phase)
|
| 53 |
+
|
| 54 |
alpha_dev = torch.abs(alpha - self.alpha_0)
|
| 55 |
security_factor = torch.exp(-300.0 * alpha_dev)
|
| 56 |
+
|
| 57 |
h = h + 0.35 * psi * security_factor
|
| 58 |
return self.output_head(h)
|
| 59 |
+
|
| 60 |
def holographic_reg(self):
|
| 61 |
reg = sum(torch.norm(w, p=1) for w in self.harmonic_weights)
|
| 62 |
return 0.0015 * reg / self.n_harmonics
|
|
|
|
| 68 |
optimizer = optim.AdamW(model.parameters(), lr=0.001, weight_decay=1e-5)
|
| 69 |
scheduler = CosineAnnealingLR(optimizer, T_max=epochs, eta_min=1e-6)
|
| 70 |
criterion = nn.MSELoss()
|
| 71 |
+
|
| 72 |
torch.manual_seed(42)
|
| 73 |
X = torch.randn(300, 8)
|
| 74 |
y = torch.sin(X.sum(dim=1, keepdim=True)) * 0.9 + 0.05 * torch.randn(300, 1)
|
| 75 |
+
|
| 76 |
losses = []
|
| 77 |
+
|
| 78 |
for epoch in range(epochs):
|
| 79 |
optimizer.zero_grad()
|
| 80 |
pred = model(X, alpha=alpha_value)
|
| 81 |
loss = criterion(pred, y)
|
| 82 |
reg = model.holographic_reg()
|
| 83 |
total_loss = loss + reg
|
| 84 |
+
|
| 85 |
total_loss.backward()
|
| 86 |
torch.nn.utils.clip_grad_norm_(model.parameters(), 1.0)
|
| 87 |
optimizer.step()
|
| 88 |
scheduler.step()
|
| 89 |
+
|
| 90 |
losses.append(loss.item())
|
| 91 |
+
|
| 92 |
# Grafik oluştur
|
| 93 |
fig, ax = plt.subplots(figsize=(10, 6))
|
| 94 |
ax.plot(losses, linewidth=2, color='#FF6B6B')
|
|
|
|
| 97 |
ax.set_title(f'Eğitim Süreci (α = {alpha_value:.8f})', fontsize=14)
|
| 98 |
ax.grid(True, alpha=0.3)
|
| 99 |
ax.set_yscale('log')
|
| 100 |
+
|
| 101 |
final_loss = losses[-1]
|
| 102 |
+
|
| 103 |
+
plt.close(fig) # Close the figure to free up memory
|
| 104 |
return fig, f"Final Loss: {final_loss:.6f}"
|
| 105 |
|
| 106 |
|
|
|
|
| 112 |
"Sapma1 (1/137.000)": 1/137.000,
|
| 113 |
"Sapma2 (1/137.070)": 1/137.070
|
| 114 |
}
|
| 115 |
+
|
| 116 |
fig, ax = plt.subplots(figsize=(12, 7))
|
| 117 |
colors = ['#FF6B6B', '#4ECDC4', '#45B7D1', '#FFA07A']
|
| 118 |
+
|
| 119 |
results = {}
|
| 120 |
+
|
| 121 |
for (name, alpha_val), color in zip(alpha_values.items(), colors):
|
| 122 |
model = HolographicMasterCodeTransformer()
|
| 123 |
optimizer = optim.AdamW(model.parameters(), lr=0.001, weight_decay=1e-5)
|
| 124 |
scheduler = CosineAnnealingLR(optimizer, T_max=epochs, eta_min=1e-6)
|
| 125 |
criterion = nn.MSELoss()
|
| 126 |
+
|
| 127 |
torch.manual_seed(42)
|
| 128 |
X = torch.randn(300, 8)
|
| 129 |
y = torch.sin(X.sum(dim=1, keepdim=True)) * 0.9 + 0.05 * torch.randn(300, 1)
|
| 130 |
+
|
| 131 |
losses = []
|
| 132 |
+
|
| 133 |
for epoch in range(epochs):
|
| 134 |
optimizer.zero_grad()
|
| 135 |
pred = model(X, alpha=alpha_val)
|
| 136 |
loss = criterion(pred, y)
|
| 137 |
reg = model.holographic_reg()
|
| 138 |
total_loss = loss + reg
|
| 139 |
+
|
| 140 |
total_loss.backward()
|
| 141 |
torch.nn.utils.clip_grad_norm_(model.parameters(), 1.0)
|
| 142 |
optimizer.step()
|
| 143 |
scheduler.step()
|
| 144 |
+
|
| 145 |
losses.append(loss.item())
|
| 146 |
+
|
| 147 |
ax.plot(losses, label=name, linewidth=2.5, color=color)
|
| 148 |
results[name] = losses[-1]
|
| 149 |
+
|
| 150 |
ax.set_xlabel('Epoch', fontsize=12)
|
| 151 |
ax.set_ylabel('Loss', fontsize=12)
|
| 152 |
ax.set_title('Alpha Değerleri Karşılaştırması', fontsize=14)
|
| 153 |
ax.legend(fontsize=10)
|
| 154 |
ax.grid(True, alpha=0.3)
|
| 155 |
ax.set_yscale('log')
|
| 156 |
+
|
| 157 |
# Sonuçları metne dönüştür
|
| 158 |
+
results_text = "**Final Loss Değerleri:**
|
| 159 |
+
|
| 160 |
+
"
|
| 161 |
for name, loss in sorted(results.items(), key=lambda x: x[1]):
|
| 162 |
+
results_text += f"• {name}: {loss:.6f}
|
| 163 |
+
"
|
| 164 |
+
|
| 165 |
+
plt.close(fig) # Close the figure to free up memory
|
| 166 |
return fig, results_text
|
| 167 |
|
| 168 |
|
| 169 |
# Gradio Interface
|
| 170 |
+
with gr.Blocks(title="OmegaCode: Holographic Master Code") as demo: # Moved theme parameter here
|
| 171 |
gr.Markdown("""
|
| 172 |
# 🌌 OmegaCode: Evrensel Kaynak Kodun Holografik Korunumu
|
| 173 |
+
|
| 174 |
Bu uygulama, **Ω_v57** teorisini temel alan Holografik Master Code Transformer modelini etkileşimli bir şekilde test etmenizi sağlar.
|
| 175 |
+
|
| 176 |
**Teori Özeti:**
|
| 177 |
- **Master Code (Ψ)**: Evrenin temel titreşim dokusunu temsil eden harmonik dalga süperpozisyonu
|
| 178 |
- **Alpha (α) Güvenlik Kilidi**: İnce yapı sabiti değerindeki sapmalara karşı sistemi koruyan mekanizma
|
| 179 |
- **Ryu-Takayanagi Entanglement**: Karadelik ufkundaki minimal yüzey alanı üzerinden bilgi korunumu
|
| 180 |
- **Holografik AI**: Yapay zeka eğitiminde kuantum hata düzeltme benzeri etkileri
|
| 181 |
""")
|
| 182 |
+
|
| 183 |
with gr.Tabs():
|
| 184 |
with gr.Tab("Tek Alpha Eğitimi"):
|
| 185 |
gr.Markdown("### Belirli Bir Alpha Değeriyle Eğitim Yap")
|
| 186 |
+
|
| 187 |
with gr.Row():
|
| 188 |
alpha_input = gr.Number(
|
| 189 |
value=1/137.035,
|
|
|
|
| 197 |
step=50,
|
| 198 |
label="Epoch Sayısı"
|
| 199 |
)
|
| 200 |
+
|
| 201 |
train_btn = gr.Button("Eğitimi Başlat", variant="primary", size="lg")
|
| 202 |
+
|
| 203 |
with gr.Row():
|
| 204 |
output_plot = gr.Plot(label="Loss Grafiği")
|
| 205 |
output_text = gr.Textbox(label="Sonuç", interactive=False)
|
| 206 |
+
|
| 207 |
train_btn.click(
|
| 208 |
fn=train_model,
|
| 209 |
inputs=[alpha_input, epochs_input],
|
| 210 |
outputs=[output_plot, output_text]
|
| 211 |
)
|
| 212 |
+
|
| 213 |
with gr.Tab("Alpha Karşılaştırması"):
|
| 214 |
gr.Markdown("### Farklı Alpha Değerlerini Karşılaştır")
|
| 215 |
+
|
| 216 |
epochs_compare = gr.Slider(
|
| 217 |
minimum=50,
|
| 218 |
maximum=300,
|
|
|
|
| 220 |
step=50,
|
| 221 |
label="Epoch Sayısı"
|
| 222 |
)
|
| 223 |
+
|
| 224 |
compare_btn = gr.Button("Karşılaştırmayı Başlat", variant="primary", size="lg")
|
| 225 |
+
|
| 226 |
with gr.Row():
|
| 227 |
compare_plot = gr.Plot(label="Karşılaştırma Grafiği")
|
| 228 |
compare_text = gr.Markdown(label="Sonuçlar")
|
| 229 |
+
|
| 230 |
compare_btn.click(
|
| 231 |
fn=compare_alphas,
|
| 232 |
inputs=[epochs_compare],
|
| 233 |
outputs=[compare_plot, compare_text]
|
| 234 |
)
|
| 235 |
+
|
| 236 |
with gr.Tab("Teori Hakkında"):
|
| 237 |
gr.Markdown("""
|
| 238 |
## Ω_v57 Teorisi Detayları
|
| 239 |
+
|
| 240 |
### Master Code (Ψ)
|
| 241 |
Evrenin temel titreşim dokusunu temsil eden harmonik dalga süperpozisyonu:
|
| 242 |
+
|
| 243 |
+
$$\Psi(\mathbf{x}, t, lpha) = \sum_{n=1}^{N} A_n \cos(\mathbf{k}_n \cdot \mathbf{x} + \omega_n t + \phi_n(lpha))$$
|
| 244 |
+
|
| 245 |
### Alpha Güvenlik Kilidi
|
| 246 |
İnce yapı sabiti α ≈ 1/137.035 değerindeki sapmalara karşı sistem kendini otomatik olarak korur:
|
| 247 |
+
|
| 248 |
+
$$\delta(lpha - lpha_0) imes \exp\left(-rac{d\Psi}{dlpha}
|
| 249 |
+
|
| 250 |
### Ryu-Takayanagi Entanglement
|
| 251 |
Karadelik ufkundaki minimal yüzey alanı üzerinden kuantum dolaşıklığının geometrik temsili:
|
| 252 |
+
|
| 253 |
+
$$S_{ ext{ent}}^{ ext{RT}} = rac{ ext{Area}(\gamma_{ ext{min}})}{4 G_{13}}$$
|
| 254 |
+
|
| 255 |
### Holografik AI Etkisi
|
| 256 |
Bu mekanizmalar, yapay zeka eğitiminde Loss Zeroing hızını %140 artırmaktadır.
|
| 257 |
""")
|
| 258 |
+
|
| 259 |
gr.Markdown("""
|
| 260 |
---
|
| 261 |
+
|
| 262 |
**Proje Kaynakları:**
|
| 263 |
- GitHub: [OmegaCode Repository](https://github.com/KULLANICI_ADI/OmegaCode)
|
| 264 |
- Teori: Ω_v57 Holographic Master Code Framework
|
|
|
|
| 266 |
""")
|
| 267 |
|
| 268 |
if __name__ == "__main__":
|
| 269 |
+
demo.launch(theme=gr.themes.Soft()) # Moved theme parameter here
|