Delete app.py

app.py

@@ -1,233 +0,0 @@
-import torch
-import torch.nn as nn
-import pandas as pd
-import csv
-import os
-import json
-from tqdm import tqdm
-import gradio as gr
-
-# =========================================================
-# MODEL ARCHITECTURE
-# =========================================================
-class DeepRegressor(nn.Module):
-    def __init__(self):
-        super().__init__()
-        self.net = nn.Sequential(
-            nn.Linear(1, 128),
-            nn.ReLU(),
-            nn.Linear(128, 256),
-            nn.ReLU(),
-            nn.Linear(256, 256),
-            nn.ReLU(),
-            nn.Linear(256, 128),
-            nn.ReLU(),
-            nn.Linear(128, 1)
-        )
-
-    def forward(self, x):
-        return self.net(x)
-
-
-# =========================================================
-# MODELS (HARDCODED — USER DOES NOT INPUT)
-# =========================================================
-
-high_mod = f"from_smartsbyviva_mark2_11_12_1Dec25_epoch_124_r2_1.00000000000000000000_rmse_0.00000090259020531177.pt"
-mid_mod  = f"from_smartsbyviva_mark2_11_12_1Dec25_epoch_143_r2_1.00000000000000000000_rmse_0.00000076154023975762.pt"
-low_mod  = f"from_smartsbyviva_mark2_11_12_1Dec25_epoch_176_r2_1.00000000000000000000_rmse_0.00000068535079250664.pt"
-
-ultra1 = f"from_smartsbyviva_mark2_11_12_1Dec25_epoch_1_r2_1.00000000000000000000_rmse_0.00000329740002026770 (1).pt"
-ultra2 = f"from_smartsbyviva_mark2_11_12_1Dec25_epoch_197_r2_1.00000000000000000000_rmse_0.00000115787004480322.pt"
-
-
-# =========================================================
-# PREDICTION
-# =========================================================
-def predict_value(input_value, model_path):
-    model = DeepRegressor()
-    model.load_state_dict(torch.load(model_path, map_location="cpu"))
-    model.eval()
-    x = torch.tensor([[float(input_value)]], dtype=torch.float32)
-    with torch.no_grad():
-        y_pred = model(x).item()
-    return float(f"{y_pred:.13f}")
-
-
-# =========================================================
-# POST PROCESS
-# =========================================================
-def post_process1(inputz, pred):
-    input_last_6 = inputz[-6:]
-    output_last_8 = f"{pred:.8f}"[-8:]
-
-    a_u = input_last_6[:2]
-    b_u = input_last_6[2:4]
-    c_u = input_last_6[4:]
-
-    a_l = output_last_8[:2]
-    b_l = output_last_8[2:4]
-    c_l = output_last_8[4:6]
-    d_l = output_last_8[6:]
-
-    def calc_diff(u, l):
-        if int(u) > int(l):
-            return int(u) - int(l)
-        return (100 + int(u)) - int(l)
-
-    diff1 = calc_diff(a_u, a_l)
-    diff2 = calc_diff(b_u, b_l)
-    diff3 = calc_diff(c_u, c_l)
-
-    sigma = str(int(d_l) + diff1 + diff2 + diff3)
-    return sigma
-
-
-# =========================================================
-# TRANSFORM
-# =========================================================
-def transform_and_average(a, b, c, d, e):
-    def transform(code: str) -> float:
-        first_digit = int(code[0])
-        base = (first_digit + 1) * 10
-        decimal_part = int(code[1:]) / 10 if len(code) > 1 else 0
-        return base + decimal_part
-
-    return (
-        transform(a),
-        transform(b),
-        transform(c),
-        transform(d),
-        transform(e),
-    )
-
-
-# =========================================================
-# CLOSEST VALUE FINDER
-# =========================================================
-def closest_code(v1, v2, v3, v4, v5, answer):
-    avg = (v1 + v2 + v3 + v4 + v5) / 5
-    target_last_digit = int(answer) % 10
-
-    closest = None
-    min_diff = float("inf")
-
-    for n in range(0, 1000):
-        if n % 10 == target_last_digit:
-            diff = abs(n - avg)
-            if diff < min_diff:
-                closest = n
-                min_diff = diff
-
-    return closest
-
-def load_file_content(file_obj):
-    if isinstance(file_obj, dict) and "path" in file_obj:
-        # Standard Gradio object
-        with open(file_obj["path"], "r") as f:
-            return f.read()
-    elif hasattr(file_obj, "name"):
-        # Older file interface
-        with open(file_obj.name, "r") as f:
-            return f.read()
-    else:
-        raise ValueError("Unsupported file type")
-
-
-
-# =========================================================
-# MAIN PROCESS FOR GRADIO
-# =========================================================
-import io
-import csv
-from tqdm import tqdm
-
-def run_pipeline(txt_file):
-    # ---- FIX START ----
-    text = load_file_content(txt_file)
-    data = text.strip()
-    # ---- FIX END ----
-
-    window_size = 12
-    total_windows = len(data) - window_size + 1
-
-    fieldnames = ["tier1", "tier2", "tier3", "tier4", "tier5", "tier6"]
-
-    # ------------------------------------------------------------
-    # ⬇️ IN-MEMORY CSV BUFFER (NO FILESYSTEM)
-    # ------------------------------------------------------------
-    buffer = io.StringIO()
-    writer = csv.DictWriter(buffer, fieldnames=fieldnames)
-    writer.writeheader()
-
-    rows_buffer = []
-    update_counter = 0
-    updates_msg = ""
-
-    for i in tqdm(range(total_windows)):
-        input_data_norm = data[i:i+12]
-        input_to_send = input_data_norm[:11]
-        answer = input_data_norm[-1]
-        input_data = "0." + input_to_send
-
-        pred1 = predict_value(input_data, high_mod)
-        pred2 = predict_value(input_data, mid_mod)
-        pred3 = predict_value(input_data, low_mod)
-        pred4 = predict_value(input_data, ultra1)
-        pred5 = predict_value(input_data, ultra2)
-
-        s1 = post_process1(input_data_norm, pred1)
-        s2 = post_process1(input_data_norm, pred2)
-        s3 = post_process1(input_data_norm, pred3)
-        s4 = post_process1(input_data_norm, pred4)
-        s5 = post_process1(input_data_norm, pred5)
-
-        v1, v2, v3, v4, v5 = transform_and_average(s1, s2, s3, s4, s5)
-        cudo = closest_code(v1, v2, v3, v4, v5, answer)
-
-        rows_buffer.append({
-            "tier1": v1,
-            "tier2": v2,
-            "tier3": v3,
-            "tier4": v4,
-            "tier5": v5,
-            "tier6": cudo
-        })
-
-        # ------------------ BATCH WRITE ------------------
-        if len(rows_buffer) >= 1000:
-            update_counter += 1
-            writer.writerows(rows_buffer)
-            updates_msg += f"Updated #{update_counter} (saved {i+1} rows)\n"
-            rows_buffer = []
-
-    # -------- final save --------
-    if rows_buffer:
-        writer.writerows(rows_buffer)
-
-    updates_msg += "\n🎉 Processing Complete"
-
-    # ------------------------------------------------------------
-    # IMPORTANT: reset pointer to start of buffer
-    # ------------------------------------------------------------
-    buffer.seek(0)
-
-    # Return buffer → gr.File will turn into downloadable CSV
-    return buffer, updates_msg
-
-
-
-# =========================================================
-# GRADIO UI
-# =========================================================
-gr.Interface(
-    fn=run_pipeline,
-    inputs=gr.File(label="Upload TXT"),
-    outputs=[
-        gr.File(label="Download CSV Output"),
-        gr.Textbox(label="Logs")
-    ]
-).launch()
-
-
-ui.launch()