Import 9 alphaear finance skills

- alphaear-deepear-lite: DeepEar Lite API integration
- alphaear-logic-visualizer: Draw.io XML finance diagrams
- alphaear-news: Real-time finance news (10+ sources)
- alphaear-predictor: Kronos time-series forecasting
- alphaear-reporter: Professional financial reports
- alphaear-search: Web search + local RAG
- alphaear-sentiment: FinBERT/LLM sentiment analysis
- alphaear-signal-tracker: Signal evolution tracking
- alphaear-stock: A-Share/HK/US stock data

Updates:
- All scripts updated to use universal .env path
- Added JINA_API_KEY, LLM_*, DEEPSEEK_API_KEY to .env.example
- Updated load_dotenv() to use ~/.config/opencode/.env

skills/alphaear-predictor/scripts/utils/predictor/evaluation.py

@@ -0,0 +1,137 @@
+import os
+import sys
+import torch
+import pandas as pd
+import numpy as np
+import glob
+from loguru import logger
+from datetime import datetime, timedelta
+
+# Setup paths
+KRONOS_DIR = os.path.dirname(os.path.abspath(__file__))
+SRC_DIR = os.path.dirname(os.path.dirname(KRONOS_DIR))
+if SRC_DIR not in sys.path:
+    sys.path.insert(0, SRC_DIR)
+
+from ..kronos.auto_synthesis_training import AutoSynthesisTrainer
+from ..kronos.model import KronosPredictor
+from ..visualizer import VisualizerTools
+from ..schema.models import ForecastResult, KLinePoint
+
+class NewsModelEvaluator:
+    def __init__(self, model_path=None):
+        self.trainer = AutoSynthesisTrainer()
+        self.device = self.trainer.device
+        
+        if model_path is None:
+            # Try to find the latest model in exports/models
+            model_files = glob.glob(os.path.join(SRC_DIR, "exports/models/*.pt"))
+            if not model_files:
+                logger.warning("⚠️ No trained models found in exports/models/. Using base model (zero-init proj).")
+            else:
+                model_path = max(model_files, key=os.path.getctime)
+        
+        if model_path:
+            self.load_weights(model_path)
+
+    def load_weights(self, path):
+        logger.info(f"🔄 Loading model weights from {path}...")
+        checkpoint = torch.load(path, map_location=self.device)
+        self.trainer.model.news_proj.load_state_dict(checkpoint['news_proj_state_dict'])
+        logger.success("✅ News projection layer loaded.")
+
+    def evaluate_range(self, start_idx=100, end_idx=200, pred_len=5):
+        # 1. Fetch Tickers
+        res = self.trainer.db.execute_query("SELECT code FROM stock_list")
+        all_tickers = [row['code'] for row in res]
+        test_tickers = all_tickers[start_idx:end_idx]
+        
+        if not test_tickers:
+            logger.error(f"No tickers found in range {start_idx}-{end_idx}")
+            return
+
+        logger.info(f"🚀 Evaluating News Model on stocks {start_idx} to {end_idx}...")
+        
+        # 2. Discover Shocks
+        shocks = self.trainer.discover_shocks(test_tickers, pred_len=pred_len)
+        
+        # 3. Associate News & Predict
+        self.trainer.model.eval()
+        predictor = KronosPredictor(self.trainer.model, self.trainer.tokenizer, device=self.device)
+        
+        save_dir = os.path.join(SRC_DIR, "exports/evaluation_results")
+        os.makedirs(save_dir, exist_ok=True)
+
+        count = 0
+        for shock in shocks:
+            summary = self.trainer.find_reason_and_verify(shock)
+            if not summary:
+                continue
+            
+            logger.info(f"📈 Testing shock: {shock['ticker']} on {shock['date']}")
+            
+            # Embedding news
+            news_emb = self.trainer.embedder.encode(summary)
+            
+            # Prediction
+            h = shock['history']
+            t = shock['target']
+            actuals = t['close'].values[:pred_len]
+            
+            x_ts = pd.to_datetime(h['date'])
+            future_dates = pd.date_range(start=x_ts.iloc[-1] + timedelta(days=1), periods=pred_len, freq='B')
+            y_ts = pd.Series(future_dates)
+            
+            # A. Base Prediction (No news)
+            p_base = predictor.predict(h, x_ts, y_ts, pred_len=pred_len, news_emb=None, verbose=False)
+            
+            # B. News-Aware Prediction
+            p_news = predictor.predict(h, x_ts, y_ts, pred_len=pred_len, news_emb=news_emb, verbose=False)
+            
+            # Calculate Improvement
+            b_preds = p_base['close'].values[:len(actuals)]
+            n_preds = p_news['close'].values[:len(actuals)]
+            b_mae = np.mean(np.abs(b_preds - actuals))
+            n_mae = np.mean(np.abs(n_preds - actuals))
+            improvement = (b_mae - n_mae) / (b_mae + 1e-6) * 100
+
+            # C. Visualize
+            try:
+                def to_kp_list(preds_df):
+                    points = []
+                    for idx, row in preds_df.iterrows():
+                        points.append(KLinePoint(
+                            date=str(idx)[:10], open=row['open'], high=row['high'],
+                            low=row['low'], close=row['close'], volume=row.get('volume', 0)
+                        ))
+                    return points
+
+                forecast_obj = ForecastResult(
+                    ticker=shock['ticker'],
+                    base_forecast=to_kp_list(p_base),
+                    adjusted_forecast=to_kp_list(p_news),
+                    rationale=summary
+                )
+
+                chart = VisualizerTools.generate_stock_chart(
+                    df=h, ticker=shock['ticker'],
+                    title=f"Test Eval: {shock['ticker']} ({shock['date']}) Imp: {improvement:.1f}%",
+                    forecast=forecast_obj,
+                    ground_truth=t[['date', 'open', 'high', 'low', 'close', 'volume']]
+                )
+                
+                safe_date = shock['date'].replace("-", "")
+                filename = f"test_{shock['ticker']}_{safe_date}.html"
+                VisualizerTools.render_chart_to_file(chart, os.path.join(save_dir, filename))
+                
+                logger.success(f"📊 Result for {shock['ticker']} saved. Base MAE: {b_mae:.4f}, News MAE: {n_mae:.4f}")
+                count += 1
+            except Exception as e:
+                logger.error(f"Visualization failed: {e}")
+
+        logger.info(f"🏁 Finished evaluation. {count} cases visualized in {save_dir}")
+
+if __name__ == "__main__":
+    # If you have a specific model, pass the path here. Otherwise it picks the latest.
+    evaluator = NewsModelEvaluator()
+    evaluator.evaluate_range(start_idx=100, end_idx=200, pred_len=1)

skills/alphaear-predictor/scripts/utils/predictor/kline_generate.py

@@ -0,0 +1,196 @@
+# Ref: https://github.com/shiyu-coder/Kronos
+
+from model import Kronos, KronosTokenizer, KronosPredictor
+import pandas as pd
+import sqlite3
+import torch
+import matplotlib.pyplot as plt
+import matplotlib.gridspec as gridspec
+from pandas.tseries.offsets import BusinessDay
+import numpy as np
+
+def get_device():
+    device = "cuda" if torch.cuda.is_available() else "mps" if torch.backends.mps.is_available() else "cpu"
+    print(f"Using device: {device}")
+    return device
+
+def load_predictor():
+    tokenizer = KronosTokenizer.from_pretrained("NeoQuasar/Kronos-Tokenizer-base")
+    model = Kronos.from_pretrained("NeoQuasar/Kronos-base")
+    device = get_device()
+    tokenizer = tokenizer.to(device)
+    model = model.to(device)
+    return KronosPredictor(model, tokenizer, device=device, max_context=512)
+
+def load_data(ticker="002111", db_path="AlphaEar/data/signal_flux.db"):
+    with sqlite3.connect(db_path) as conn:
+        df = pd.read_sql_query(f"SELECT * FROM stock_prices WHERE ticker = '{ticker}'", conn)
+    df['date'] = pd.to_datetime(df['date'])
+    df = df.sort_values('date').reset_index(drop=True)
+    return df
+
+def plot_kline_matplotlib(ax, ax_vol, dates, df, label_suffix="", color_up='#ef4444', color_down='#22c55e', alpha=1.0, is_prediction=False):
+    """
+    绘制 K 线图和成交量
+    """
+    # X axis mapping to integers for consistent spacing
+    x = np.arange(len(dates))
+    
+    # K-line data
+    opens = df['open'].values
+    closes = df['close'].values
+    highs = df['high'].values
+    lows = df['low'].values
+    volumes = df['volume'].values
+    
+    # Width of the candlestick
+    width = 0.6
+    
+    for i in range(len(x)):
+        color = color_up if closes[i] >= opens[i] else color_down
+        linestyle = '--' if is_prediction else '-'
+        
+        # Wick
+        ax.vlines(x[i], lows[i], highs[i], color=color, linewidth=1, alpha=alpha, linestyle=linestyle)
+        
+        # Body
+        rect_bottom = min(opens[i], closes[i])
+        rect_height = abs(opens[i] - closes[i])
+        if rect_height == 0: rect_height = 0.001 # Visual hair
+        
+        ax.add_patch(plt.Rectangle((x[i] - width/2, rect_bottom), width, rect_height, 
+                                 edgecolor=color, facecolor=color if not is_prediction else 'none', 
+                                 alpha=alpha, linewidth=1, linestyle=linestyle))
+        
+        # Volume
+        ax_vol.bar(x[i], volumes[i], color=color, alpha=alpha * 0.5, width=width)
+
+def render_comparison_chart(history_df, actual_df, pred_df, title):
+    """
+    渲染组合图：历史 K 线 + 真值 K 线 + 预测 K 线
+    """
+    # Combine all dates for X axis
+    all_dates = pd.concat([history_df['date'], actual_df['date'] if actual_df is not None else pred_df.index.to_series()]).unique()
+    all_dates = sorted(all_dates)
+    date_to_idx = {date: i for i, date in enumerate(all_dates)}
+    
+    fig = plt.figure(figsize=(14, 8), facecolor='white')
+    gs = gridspec.GridSpec(2, 1, height_ratios=[3, 1], hspace=0.1)
+    ax_main = fig.add_subplot(gs[0])
+    ax_vol = fig.add_subplot(gs[1], sharex=ax_main)
+    
+    # 1. Plot History
+    hist_indices = [date_to_idx[d] for d in history_df['date']]
+    # We use a custom x for plotting to ensure continuity
+    plot_kline_matplotlib(ax_main, ax_vol, history_df['date'], history_df, alpha=0.8)
+    
+    offset = len(history_df)
+    
+    # 2. Plot Actual if exists
+    if actual_df is not None:
+        # Shift indices
+        actual_x = np.arange(len(actual_df)) + offset
+        # Plotting manually to handle offset
+        for i in range(len(actual_df)):
+            idx = actual_x[i]
+            row = actual_df.iloc[i]
+            color = '#ef4444' if row['close'] >= row['open'] else '#22c55e'
+            ax_main.vlines(idx, row['low'], row['high'], color=color, linewidth=1, alpha=0.9)
+            ax_main.add_patch(plt.Rectangle((idx - 0.3, min(row['open'], row['close'])), 0.6, abs(row['open']-row['close']), 
+                                         edgecolor=color, facecolor=color, alpha=0.9))
+            ax_vol.bar(idx, row['volume'], color=color, alpha=0.4)
+            
+    # 3. Plot Prediction
+    pred_x = np.arange(len(pred_df)) + offset
+    for i in range(len(pred_df)):
+        idx = pred_x[i]
+        row = pred_df.iloc[i]
+        color = '#ff8c00' # Orange for prediction to distinguish
+        ax_main.vlines(idx, row['low'], row['high'], color=color, linewidth=1.5, linestyle='--')
+        ax_main.add_patch(plt.Rectangle((idx - 0.3, min(row['open'], row['close'])), 0.6, abs(row['open']-row['close']), 
+                                     edgecolor=color, facecolor='none', linewidth=1.5, linestyle='--'))
+        # Plot secondary prediction line for close
+        if i == 0:
+            # Connect to history
+            ax_main.plot([offset-1, idx], [history_df['close'].iloc[-1], row['close']], color=color, linestyle='--', alpha=0.6)
+        elif i > 0:
+            ax_main.plot([idx-1, idx], [pred_df['close'].iloc[i-1], row['close']], color=color, linestyle='--', alpha=0.6)
+
+    # Styling
+    ax_main.set_title(title, fontsize=14, fontweight='bold')
+    ax_main.grid(True, linestyle=':', alpha=0.6)
+    ax_vol.grid(True, linestyle=':', alpha=0.6)
+    ax_vol.set_ylabel('Volume')
+    ax_main.set_ylabel('Price')
+    
+    # Set X ticks
+    step = max(1, len(all_dates) // 10)
+    ax_vol.set_xticks(np.arange(0, len(all_dates), step))
+    ax_vol.set_xticklabels([all_dates[i].strftime('%Y-%m-%d') for i in range(0, len(all_dates), step)], rotation=45)
+    
+    plt.tight_layout()
+    plt.show()
+    plt.close()
+
+def run_backtest(df, predictor, lookback, pred_len, start_index=0):
+    total_len = len(df)
+    history_start = start_index
+    history_end = start_index + lookback 
+    pred_start = history_end
+    
+    available_pred_len = total_len - pred_start
+    if available_pred_len <= 0: return
+    actual_pred_len = min(pred_len, available_pred_len)
+    pred_end = pred_start + actual_pred_len
+    
+    x_df = df.iloc[history_start : history_end].copy()
+    y_true_df = df.iloc[pred_start : pred_end].copy()
+    y_timestamp = y_true_df['date']
+    
+    print(f"Backtesting: {x_df['date'].iloc[0].date()} to {y_timestamp.iloc[-1].date()}")
+    
+    pred_df = predictor.predict(
+        df=x_df[['open', 'high', 'low', 'close', 'volume']],
+        x_timestamp=x_df['date'],
+        y_timestamp=y_timestamp,
+        pred_len=actual_pred_len,
+        T=1.0, top_p=0.9, sample_count=1
+    )
+    
+    render_comparison_chart(x_df, y_true_df, pred_df, f"Backtest: {TICKER} K-Line Comparison")
+
+def run_forecast(df, predictor, lookback, pred_len):
+    if len(df) < lookback: return
+    x_df = df.iloc[-lookback:].copy()
+    last_date = x_df['date'].iloc[-1]
+    future_dates = pd.date_range(start=last_date + BusinessDay(1), periods=pred_len, freq='B')
+    future_dates = pd.Series(future_dates)
+    
+    print(f"Forecasting: Starting from {future_dates.iloc[0].date()}")
+    
+    pred_df = predictor.predict(
+        df=x_df[['open', 'high', 'low', 'close', 'volume']],
+        x_timestamp=x_df['date'],
+        y_timestamp=future_dates,
+        pred_len=pred_len,
+        T=1.0, top_p=0.9, sample_count=1
+    )
+    
+    render_comparison_chart(x_df, None, pred_df, f"Forecast: {TICKER} Future K-Line")
+
+if __name__ == "__main__":
+    LOOKBACK = 20
+    PRED_LEN = 10
+    TICKER = '002111'
+    
+    pred_model = load_predictor()
+    stock_data = load_data(TICKER)
+    
+    total_rows = len(stock_data)
+    backtest_start = max(0, total_rows - LOOKBACK - PRED_LEN - 10) # Leave some space to see trend
+    
+    print("\n--- Running Backtest ---")
+    run_backtest(stock_data, pred_model, LOOKBACK, PRED_LEN, start_index=backtest_start)
+    
+    print("\n--- Running Forecast ---")
+    run_forecast(stock_data, pred_model, LOOKBACK, PRED_LEN)

skills/alphaear-predictor/scripts/utils/predictor/model/__init__.py

@@ -0,0 +1,16 @@
+from .kronos import KronosTokenizer, Kronos, KronosPredictor
+
+model_dict = {
+    'kronos_tokenizer': KronosTokenizer,
+    'kronos': Kronos,
+    'kronos_predictor': KronosPredictor
+}
+
+
+def get_model_class(model_name):
+    if model_name in model_dict:
+        return model_dict[model_name]
+    else:
+        print(f"Model {model_name} not found in model_dict")
+        raise NotImplementedError
+

skills/alphaear-predictor/scripts/utils/predictor/model/kronos.py

@@ -0,0 +1,676 @@
+import numpy as np
+import pandas as pd
+import torch
+from huggingface_hub import PyTorchModelHubMixin
+import sys
+
+from tqdm import trange
+
+sys.path.append("../")
+from model.module import *
+
+
+class KronosTokenizer(nn.Module, PyTorchModelHubMixin):
+    """
+    KronosTokenizer module for tokenizing input data using a hybrid quantization approach.
+
+    This tokenizer utilizes a combination of encoder and decoder Transformer blocks
+    along with the Binary Spherical Quantization (BSQuantizer) to compress and decompress input data.
+
+    Args:
+           d_in (int): Input dimension.
+           d_model (int): Model dimension.
+           n_heads (int): Number of attention heads.
+           ff_dim (int): Feed-forward dimension.
+           n_enc_layers (int): Number of encoder layers.
+           n_dec_layers (int): Number of decoder layers.
+           ffn_dropout_p (float): Dropout probability for feed-forward networks.
+           attn_dropout_p (float): Dropout probability for attention mechanisms.
+           resid_dropout_p (float): Dropout probability for residual connections.
+           s1_bits (int): Number of bits for the pre token in BSQuantizer.
+           s2_bits (int): Number of bits for the post token in BSQuantizer.
+           beta (float): Beta parameter for BSQuantizer.
+           gamma0 (float): Gamma0 parameter for BSQuantizer.
+           gamma (float): Gamma parameter for BSQuantizer.
+           zeta (float): Zeta parameter for BSQuantizer.
+           group_size (int): Group size parameter for BSQuantizer.
+
+    """
+
+    def __init__(self, d_in, d_model, n_heads, ff_dim, n_enc_layers, n_dec_layers, ffn_dropout_p, attn_dropout_p, resid_dropout_p, s1_bits, s2_bits, beta, gamma0, gamma, zeta, group_size):
+
+        super().__init__()
+        self.d_in = d_in
+        self.d_model = d_model
+        self.n_heads = n_heads
+        self.ff_dim = ff_dim
+        self.enc_layers = n_enc_layers
+        self.dec_layers = n_dec_layers
+        self.ffn_dropout_p = ffn_dropout_p
+        self.attn_dropout_p = attn_dropout_p
+        self.resid_dropout_p = resid_dropout_p
+
+        self.s1_bits = s1_bits
+        self.s2_bits = s2_bits
+        self.codebook_dim = s1_bits + s2_bits # Total dimension of the codebook after quantization
+        self.embed = nn.Linear(self.d_in, self.d_model)
+        self.head = nn.Linear(self.d_model, self.d_in)
+
+        # Encoder Transformer Blocks
+        self.encoder = nn.ModuleList([
+            TransformerBlock(self.d_model, self.n_heads, self.ff_dim, self.ffn_dropout_p, self.attn_dropout_p, self.resid_dropout_p)
+            for _ in range(self.enc_layers - 1)
+        ])
+        # Decoder Transformer Blocks
+        self.decoder = nn.ModuleList([
+            TransformerBlock(self.d_model, self.n_heads, self.ff_dim, self.ffn_dropout_p, self.attn_dropout_p, self.resid_dropout_p)
+            for _ in range(self.dec_layers - 1)
+        ])
+        self.quant_embed = nn.Linear(in_features=self.d_model, out_features=self.codebook_dim) # Linear layer before quantization
+        self.post_quant_embed_pre = nn.Linear(in_features=self.s1_bits, out_features=self.d_model) # Linear layer after quantization (pre part - s1 bits)
+        self.post_quant_embed = nn.Linear(in_features=self.codebook_dim, out_features=self.d_model) # Linear layer after quantization (full codebook)
+        self.tokenizer = BSQuantizer(self.s1_bits, self.s2_bits, beta, gamma0, gamma, zeta, group_size) # BSQuantizer module
+
+    def forward(self, x):
+        """
+        Forward pass of the KronosTokenizer.
+
+        Args:
+            x (torch.Tensor): Input tensor of shape (batch_size, seq_len, d_in).
+
+        Returns:
+            tuple: A tuple containing:
+                - tuple: (z_pre, z) - Reconstructed outputs from decoder with s1_bits and full codebook respectively,
+                         both of shape (batch_size, seq_len, d_in).
+                - torch.Tensor: bsq_loss - Loss from the BSQuantizer.
+                - torch.Tensor: quantized - Quantized representation from BSQuantizer.
+                - torch.Tensor: z_indices - Indices from the BSQuantizer.
+        """
+        z = self.embed(x)
+
+        for layer in self.encoder:
+            z = layer(z)
+
+        z = self.quant_embed(z) # (B, T, codebook)
+
+        bsq_loss, quantized, z_indices = self.tokenizer(z)
+
+        quantized_pre = quantized[:, :, :self.s1_bits] # Extract the first part of quantized representation (s1_bits)
+        z_pre = self.post_quant_embed_pre(quantized_pre)
+
+        z = self.post_quant_embed(quantized)
+
+        # Decoder layers (for pre part - s1 bits)
+        for layer in self.decoder:
+            z_pre = layer(z_pre)
+        z_pre = self.head(z_pre)
+
+        # Decoder layers (for full codebook)
+        for layer in self.decoder:
+            z = layer(z)
+        z = self.head(z)
+
+        return (z_pre, z), bsq_loss, quantized, z_indices
+
+    def indices_to_bits(self, x, half=False):
+        """
+        Converts indices to bit representations and scales them.
+
+        Args:
+            x (torch.Tensor): Indices tensor.
+            half (bool, optional): Whether to process only half of the codebook dimension. Defaults to False.
+
+        Returns:
+            torch.Tensor: Bit representation tensor.
+        """
+        if half:
+            x1 = x[0] # Assuming x is a tuple of indices if half is True
+            x2 = x[1]
+            mask = 2 ** torch.arange(self.codebook_dim//2, device=x1.device, dtype=torch.long) # Create a mask for bit extraction
+            x1 = (x1.unsqueeze(-1) & mask) != 0 # Extract bits for the first half
+            x2 = (x2.unsqueeze(-1) & mask) != 0 # Extract bits for the second half
+            x = torch.cat([x1, x2], dim=-1) # Concatenate the bit representations
+        else:
+            mask = 2 ** torch.arange(self.codebook_dim, device=x.device, dtype=torch.long) # Create a mask for bit extraction
+            x = (x.unsqueeze(-1) & mask) != 0 # Extract bits
+
+        x = x.float() * 2 - 1 # Convert boolean to bipolar (-1, 1)
+        q_scale = 1. / (self.codebook_dim ** 0.5) # Scaling factor
+        x = x * q_scale
+        return x
+
+    def encode(self, x, half=False):
+        """
+        Encodes the input data into quantized indices.
+
+        Args:
+            x (torch.Tensor): Input tensor of shape (batch_size, seq_len, d_in).
+            half (bool, optional): Whether to use half quantization in BSQuantizer. Defaults to False.
+
+        Returns:
+            torch.Tensor: Quantized indices from BSQuantizer.
+        """
+        z = self.embed(x)
+        for layer in self.encoder:
+            z = layer(z)
+        z = self.quant_embed(z)
+
+        bsq_loss, quantized, z_indices = self.tokenizer(z, half=half, collect_metrics=False)
+        return z_indices
+
+    def decode(self, x, half=False):
+        """
+        Decodes quantized indices back to the input data space.
+
+        Args:
+            x (torch.Tensor): Quantized indices tensor.
+            half (bool, optional): Whether the indices were generated with half quantization. Defaults to False.
+
+        Returns:
+            torch.Tensor: Reconstructed output tensor of shape (batch_size, seq_len, d_in).
+        """
+        quantized = self.indices_to_bits(x, half)
+        z = self.post_quant_embed(quantized)
+        for layer in self.decoder:
+            z = layer(z)
+        z = self.head(z)
+        return z
+
+
+class Kronos(nn.Module, PyTorchModelHubMixin):
+    """
+    Kronos Model.
+
+    Args:
+        s1_bits (int): Number of bits for pre tokens.
+        s2_bits (int): Number of bits for post tokens.
+        n_layers (int): Number of Transformer blocks.
+        d_model (int): Dimension of the model's embeddings and hidden states.
+        n_heads (int): Number of attention heads in the MultiheadAttention layers.
+        ff_dim (int): Dimension of the feedforward network in the Transformer blocks.
+        ffn_dropout_p (float): Dropout probability for the feedforward network.
+        attn_dropout_p (float): Dropout probability for the attention layers.
+        resid_dropout_p (float): Dropout probability for residual connections.
+        token_dropout_p (float): Dropout probability for token embeddings.
+        learn_te (bool): Whether to use learnable temporal embeddings.
+    """
+
+    def __init__(self, s1_bits, s2_bits, n_layers, d_model, n_heads, ff_dim, ffn_dropout_p, attn_dropout_p, resid_dropout_p, token_dropout_p, learn_te, news_dim=None):
+        super().__init__()
+        self.s1_bits = s1_bits
+        self.s2_bits = s2_bits
+        self.n_layers = n_layers
+        self.d_model = d_model
+        self.n_heads = n_heads
+        self.learn_te = learn_te
+        self.ff_dim = ff_dim
+        self.ffn_dropout_p = ffn_dropout_p
+        self.attn_dropout_p = attn_dropout_p
+        self.resid_dropout_p = resid_dropout_p
+        self.token_dropout_p = token_dropout_p
+        self.news_dim = news_dim
+
+        self.s1_vocab_size = 2 ** self.s1_bits
+        self.token_drop = nn.Dropout(self.token_dropout_p)
+        self.embedding = HierarchicalEmbedding(self.s1_bits, self.s2_bits, self.d_model)
+        self.time_emb = TemporalEmbedding(self.d_model, self.learn_te)
+        self.transformer = nn.ModuleList([
+            TransformerBlock(self.d_model, self.n_heads, self.ff_dim, self.ffn_dropout_p, self.attn_dropout_p, self.resid_dropout_p)
+            for _ in range(self.n_layers)
+        ])
+        self.norm = RMSNorm(self.d_model)
+        self.dep_layer = DependencyAwareLayer(self.d_model)
+        self.head = DualHead(self.s1_bits, self.s2_bits, self.d_model)
+
+        if self.news_dim is not None:
+            self.news_proj = nn.Linear(self.news_dim, self.d_model)
+        else:
+            self.news_proj = None
+
+        self.apply(self._init_weights)
+
+    def _init_weights(self, module):
+
+        if isinstance(module, nn.Linear):
+            nn.init.xavier_normal_(module.weight)
+            if module.bias is not None:
+                nn.init.zeros_(module.bias)
+        elif isinstance(module, nn.Embedding):
+            nn.init.normal_(module.weight, mean=0, std=self.embedding.d_model ** -0.5)
+        elif isinstance(module, nn.LayerNorm):
+            nn.init.ones_(module.weight)
+            nn.init.zeros_(module.bias)
+        elif isinstance(module, RMSNorm):
+            nn.init.ones_(module.weight)
+
+    def forward(self, s1_ids, s2_ids, stamp=None, padding_mask=None, use_teacher_forcing=False, s1_targets=None, news_emb=None):
+        """
+        Args:
+            s1_ids (torch.Tensor): Input tensor of s1 token IDs. Shape: [batch_size, seq_len]
+            s2_ids (torch.Tensor): Input tensor of s2 token IDs. Shape: [batch_size, seq_len]
+            stamp (torch.Tensor, optional): Temporal stamp tensor. Shape: [batch_size, seq_len]. Defaults to None.
+            padding_mask (torch.Tensor, optional): Mask for padding tokens. Shape: [batch_size, seq_len]. Defaults to None.
+            use_teacher_forcing (bool, optional): Whether to use teacher forcing for s1 decoding. Defaults to False.
+            s1_targets (torch.Tensor, optional): Target s1 token IDs for teacher forcing. Shape: [batch_size, seq_len]. Defaults to None.
+            news_emb (torch.Tensor, optional): News embedding tensor. Shape: [batch_size, news_dim]. Defaults to None.
+
+        Returns:
+            Tuple[torch.Tensor, torch.Tensor]:
+                - s1 logits: Logits for s1 token predictions. Shape: [batch_size, seq_len, s1_vocab_size]
+                - s2_logits: Logits for s2 token predictions, conditioned on s1. Shape: [batch_size, seq_len, s2_vocab_size]
+        """
+        x = self.embedding([s1_ids, s2_ids])
+        if stamp is not None:
+            time_embedding = self.time_emb(stamp)
+            x = x + time_embedding
+        x = self.token_drop(x)
+
+        for layer in self.transformer:
+            x = layer(x, key_padding_mask=padding_mask)
+
+        x = self.norm(x)
+
+        if news_emb is not None and self.news_proj is not None:
+            news_bias = self.news_proj(news_emb).unsqueeze(1) # [B, 1, d_model]
+            x = x + news_bias
+
+        s1_logits = self.head(x)
+
+        if use_teacher_forcing:
+            sibling_embed = self.embedding.emb_s1(s1_targets)
+        else:
+            s1_probs = F.softmax(s1_logits.detach(), dim=-1)
+            sample_s1_ids = torch.multinomial(s1_probs.view(-1, self.s1_vocab_size), 1).view(s1_ids.shape)
+            sibling_embed = self.embedding.emb_s1(sample_s1_ids)
+
+        x2 = self.dep_layer(x, sibling_embed, key_padding_mask=padding_mask) # Dependency Aware Layer: Condition on s1 embeddings
+        s2_logits = self.head.cond_forward(x2)
+        return s1_logits, s2_logits
+
+    def decode_s1(self, s1_ids, s2_ids, stamp=None, padding_mask=None, news_emb=None):
+        """
+        Decodes only the s1 tokens.
+
+        This method performs a forward pass to predict only s1 tokens. It returns the s1 logits
+        and the context representation from the Transformer, which can be used for subsequent s2 decoding.
+
+        Args:
+            s1_ids (torch.Tensor): Input tensor of s1 token IDs. Shape: [batch_size, seq_len]
+            s2_ids (torch.Tensor): Input tensor of s2 token IDs. Shape: [batch_size, seq_len]
+            stamp (torch.Tensor, optional): Temporal stamp tensor. Shape: [batch_size, seq_len]. Defaults to None.
+            padding_mask (torch.Tensor, optional): Mask for padding tokens. Shape: [batch_size, seq_len]. Defaults to None.
+            news_emb (torch.Tensor, optional): News embedding tensor. Shape: [batch_size, news_dim]. Defaults to None.
+
+        Returns:
+            Tuple[torch.Tensor, torch.Tensor]:
+                - s1 logits: Logits for s1 token predictions. Shape: [batch_size, seq_len, s1_vocab_size]
+                - context: Context representation from the Transformer. Shape: [batch_size, seq_len, d_model]
+        """
+        x = self.embedding([s1_ids, s2_ids])
+        if stamp is not None:
+            time_embedding = self.time_emb(stamp)
+            x = x + time_embedding
+        x = self.token_drop(x)
+
+        for layer in self.transformer:
+            x = layer(x, key_padding_mask=padding_mask)
+
+        x = self.norm(x)
+
+        if news_emb is not None and self.news_proj is not None:
+            news_bias = self.news_proj(news_emb).unsqueeze(1) # [B, 1, d_model]
+            x = x + news_bias
+
+        s1_logits = self.head(x)
+        return s1_logits, x
+
+    def decode_s2(self, context, s1_ids, padding_mask=None):
+        """
+        Decodes the s2 tokens, conditioned on the context and s1 tokens.
+
+        This method decodes s2 tokens based on a pre-computed context representation (typically from `decode_s1`)
+        and the s1 token IDs. It uses the dependency-aware layer and the conditional s2 head to predict s2 tokens.
+
+        Args:
+            context (torch.Tensor): Context representation from the transformer (output of decode_s1).
+                                     Shape: [batch_size, seq_len, d_model]
+            s1_ids (torch.torch.Tensor): Input tensor of s1 token IDs. Shape: [batch_size, seq_len]
+            padding_mask (torch.Tensor, optional): Mask for padding tokens. Shape: [batch_size, seq_len]. Defaults to None.
+
+        Returns:
+            torch.Tensor: s2 logits. Shape: [batch_size, seq_len, s2_vocab_size]
+        """
+        sibling_embed = self.embedding.emb_s1(s1_ids)
+        x2 = self.dep_layer(context, sibling_embed, key_padding_mask=padding_mask)
+        return self.head.cond_forward(x2)
+
+
+def top_k_top_p_filtering(
+        logits,
+        top_k: int = 0,
+        top_p: float = 1.0,
+        filter_value: float = -float("Inf"),
+        min_tokens_to_keep: int = 1,
+):
+    """Filter a distribution of logits using top-k and/or nucleus (top-p) filtering
+    Args:
+        logits: logits distribution shape (batch size, vocabulary size)
+        if top_k > 0: keep only top k tokens with highest probability (top-k filtering).
+        if top_p < 1.0: keep the top tokens with cumulative probability >= top_p (nucleus filtering).
+            Nucleus filtering is described in Holtzman et al. (http://arxiv.org/abs/1904.09751)
+        Make sure we keep at least min_tokens_to_keep per batch example in the output
+    From: https://gist.github.com/thomwolf/1a5a29f6962089e871b94cbd09daf317
+    """
+    if top_k > 0:
+        top_k = min(max(top_k, min_tokens_to_keep), logits.size(-1))  # Safety check
+        # Remove all tokens with a probability less than the last token of the top-k
+        indices_to_remove = logits < torch.topk(logits, top_k)[0][..., -1, None]
+        logits[indices_to_remove] = filter_value
+        return logits
+
+    if top_p < 1.0:
+        sorted_logits, sorted_indices = torch.sort(logits, descending=True)
+        cumulative_probs = torch.cumsum(F.softmax(sorted_logits, dim=-1), dim=-1)
+
+        # Remove tokens with cumulative probability above the threshold (token with 0 are kept)
+        sorted_indices_to_remove = cumulative_probs > top_p
+        if min_tokens_to_keep > 1:
+            # Keep at least min_tokens_to_keep (set to min_tokens_to_keep-1 because we add the first one below)
+            sorted_indices_to_remove[..., :min_tokens_to_keep] = 0
+        # Shift the indices to the right to keep also the first token above the threshold
+        sorted_indices_to_remove[..., 1:] = sorted_indices_to_remove[..., :-1].clone()
+        sorted_indices_to_remove[..., 0] = 0
+
+        # scatter sorted tensors to original indexing
+        indices_to_remove = sorted_indices_to_remove.scatter(1, sorted_indices, sorted_indices_to_remove)
+        logits[indices_to_remove] = filter_value
+        return logits
+
+
+def sample_from_logits(logits, temperature=1.0, top_k=None, top_p=None, sample_logits=True):
+    logits = logits / temperature
+    if top_k is not None or top_p is not None:
+        if top_k > 0 or top_p < 1.0:
+            logits = top_k_top_p_filtering(logits, top_k=top_k, top_p=top_p)
+
+    probs = F.softmax(logits, dim=-1)
+
+    if not sample_logits:
+        _, x = top_k(probs, k=1, dim=-1)
+    else:
+        x = torch.multinomial(probs, num_samples=1)
+
+    return x
+
+
+def auto_regressive_inference(tokenizer, model, x, x_stamp, y_stamp, max_context, pred_len, clip=5, T=1.0, top_k=0, top_p=0.99, sample_count=5, verbose=False, news_emb=None):
+    with torch.no_grad():
+        x = torch.clip(x, -clip, clip)
+
+        device = x.device
+        x = x.unsqueeze(1).repeat(1, sample_count, 1, 1).reshape(-1, x.size(1), x.size(2)).to(device)
+        x_stamp = x_stamp.unsqueeze(1).repeat(1, sample_count, 1, 1).reshape(-1, x_stamp.size(1), x_stamp.size(2)).to(device)
+        y_stamp = y_stamp.unsqueeze(1).repeat(1, sample_count, 1, 1).reshape(-1, y_stamp.size(1), y_stamp.size(2)).to(device)
+
+        x_token = tokenizer.encode(x, half=True)
+        
+        initial_seq_len = x.size(1)
+        batch_size = x_token[0].size(0)
+        total_seq_len = initial_seq_len + pred_len
+        full_stamp = torch.cat([x_stamp, y_stamp], dim=1)
+
+        generated_pre = x_token[0].new_empty(batch_size, pred_len)
+        generated_post = x_token[1].new_empty(batch_size, pred_len)
+
+        pre_buffer = x_token[0].new_zeros(batch_size, max_context)
+        post_buffer = x_token[1].new_zeros(batch_size, max_context)
+        buffer_len = min(initial_seq_len, max_context)
+        if buffer_len > 0:
+            start_idx = max(0, initial_seq_len - max_context)
+            pre_buffer[:, :buffer_len] = x_token[0][:, start_idx:start_idx + buffer_len]
+            post_buffer[:, :buffer_len] = x_token[1][:, start_idx:start_idx + buffer_len]
+
+        if verbose:
+            ran = trange
+        else:
+            ran = range
+        for i in ran(pred_len):
+            current_seq_len = initial_seq_len + i
+            window_len = min(current_seq_len, max_context)
+
+            if current_seq_len <= max_context:
+                input_tokens = [
+                    pre_buffer[:, :window_len],
+                    post_buffer[:, :window_len]
+                ]
+            else:
+                input_tokens = [pre_buffer, post_buffer]
+
+            context_end = current_seq_len
+            context_start = max(0, context_end - max_context)
+            current_stamp = full_stamp[:, context_start:context_end, :].contiguous()
+
+            s1_logits, context = model.decode_s1(input_tokens[0], input_tokens[1], current_stamp, news_emb=news_emb)
+            s1_logits = s1_logits[:, -1, :]
+            sample_pre = sample_from_logits(s1_logits, temperature=T, top_k=top_k, top_p=top_p, sample_logits=True)
+
+            s2_logits = model.decode_s2(context, sample_pre)
+            s2_logits = s2_logits[:, -1, :]
+            sample_post = sample_from_logits(s2_logits, temperature=T, top_k=top_k, top_p=top_p, sample_logits=True)
+
+            generated_pre[:, i] = sample_pre.squeeze(-1)
+            generated_post[:, i] = sample_post.squeeze(-1)
+
+            if current_seq_len < max_context:
+                pre_buffer[:, current_seq_len] = sample_pre.squeeze(-1)
+                post_buffer[:, current_seq_len] = sample_post.squeeze(-1)
+            else:
+                pre_buffer.copy_(torch.roll(pre_buffer, shifts=-1, dims=1))
+                post_buffer.copy_(torch.roll(post_buffer, shifts=-1, dims=1))
+                pre_buffer[:, -1] = sample_pre.squeeze(-1)
+                post_buffer[:, -1] = sample_post.squeeze(-1)
+
+        full_pre = torch.cat([x_token[0], generated_pre], dim=1)
+        full_post = torch.cat([x_token[1], generated_post], dim=1)
+
+        context_start = max(0, total_seq_len - max_context)
+        input_tokens = [
+            full_pre[:, context_start:total_seq_len].contiguous(),
+            full_post[:, context_start:total_seq_len].contiguous()
+        ]
+        z = tokenizer.decode(input_tokens, half=True)
+        z = z.reshape(-1, sample_count, z.size(1), z.size(2))
+        preds = z.cpu().numpy()
+        preds = np.mean(preds, axis=1)
+
+        return preds
+
+
+def calc_time_stamps(x_timestamp):
+    time_df = pd.DataFrame()
+    time_df['minute'] = x_timestamp.dt.minute
+    time_df['hour'] = x_timestamp.dt.hour
+    time_df['weekday'] = x_timestamp.dt.weekday
+    time_df['day'] = x_timestamp.dt.day
+    time_df['month'] = x_timestamp.dt.month
+    return time_df
+
+
+class KronosPredictor:
+
+    def __init__(self, model, tokenizer, device="cuda:0", max_context=512, clip=5):
+        self.tokenizer = tokenizer
+        self.model = model
+        self.max_context = max_context
+        self.clip = clip
+        self.price_cols = ['open', 'high', 'low', 'close']
+        self.vol_col = 'volume'
+        self.amt_vol = 'amount'
+        self.time_cols = ['minute', 'hour', 'weekday', 'day', 'month']
+        self.device = device
+
+        self.tokenizer = self.tokenizer.to(self.device)
+        self.model = self.model.to(self.device)
+
+    def generate(self, x, x_stamp, y_stamp, pred_len, T, top_k, top_p, sample_count, verbose, news_emb=None):
+
+        x_tensor = torch.from_numpy(np.array(x).astype(np.float32)).to(self.device)
+        x_stamp_tensor = torch.from_numpy(np.array(x_stamp).astype(np.float32)).to(self.device)
+        y_stamp_tensor = torch.from_numpy(np.array(y_stamp).astype(np.float32)).to(self.device)
+
+        preds = auto_regressive_inference(self.tokenizer, self.model, x_tensor, x_stamp_tensor, y_stamp_tensor, self.max_context, pred_len,
+                                          self.clip, T, top_k, top_p, sample_count, verbose, news_emb=news_emb)
+        preds = preds[:, -pred_len:, :]
+        return preds
+
+    def predict(self, df, x_timestamp, y_timestamp, pred_len, T=1.0, top_k=0, top_p=0.9, sample_count=1, verbose=True, news_emb=None):
+
+        if not isinstance(df, pd.DataFrame):
+            raise ValueError("Input must be a pandas DataFrame.")
+
+        if not all(col in df.columns for col in self.price_cols):
+            raise ValueError(f"Price columns {self.price_cols} not found in DataFrame.")
+
+        df = df.copy()
+        if self.vol_col not in df.columns:
+            df[self.vol_col] = 0.0  # Fill missing volume with zeros
+            df[self.amt_vol] = 0.0  # Fill missing amount with zeros
+        if self.amt_vol not in df.columns and self.vol_col in df.columns:
+            df[self.amt_vol] = df[self.vol_col] * df[self.price_cols].mean(axis=1)
+
+        if df[self.price_cols + [self.vol_col, self.amt_vol]].isnull().values.any():
+            raise ValueError("Input DataFrame contains NaN values in price or volume columns.")
+
+        x_time_df = calc_time_stamps(x_timestamp)
+        y_time_df = calc_time_stamps(y_timestamp)
+
+        x = df[self.price_cols + [self.vol_col, self.amt_vol]].values.astype(np.float32)
+        x_stamp = x_time_df.values.astype(np.float32)
+        y_stamp = y_time_df.values.astype(np.float32)
+
+        x_mean, x_std = np.mean(x, axis=0), np.std(x, axis=0)
+
+        x = (x - x_mean) / (x_std + 1e-5)
+        x = np.clip(x, -self.clip, self.clip)
+
+        x = x[np.newaxis, :]
+        x_stamp = x_stamp[np.newaxis, :]
+        y_stamp = y_stamp[np.newaxis, :]
+
+        if news_emb is not None:
+            news_emb_tensor = torch.from_numpy(np.array(news_emb).astype(np.float32)).to(self.device)
+            # Ensure batch dimension for news_emb if only one sample
+            if news_emb_tensor.ndim == 1:
+                news_emb_tensor = news_emb_tensor.unsqueeze(0)
+        else:
+            news_emb_tensor = None
+
+        preds = self.generate(x, x_stamp, y_stamp, pred_len, T, top_k, top_p, sample_count, verbose, news_emb=news_emb_tensor)
+
+        preds = preds.squeeze(0)
+        preds = preds * (x_std + 1e-5) + x_mean
+
+        pred_df = pd.DataFrame(preds, columns=self.price_cols + [self.vol_col, self.amt_vol], index=y_timestamp)
+        return pred_df
+
+
+    def predict_batch(self, df_list, x_timestamp_list, y_timestamp_list, pred_len, T=1.0, top_k=0, top_p=0.9, sample_count=1, verbose=True):
+        """
+        Perform parallel (batch) prediction on multiple time series. All series must have the same historical length and prediction length (pred_len).
+
+        Args:
+            df_list (List[pd.DataFrame]): List of input DataFrames, each containing price columns and optional volume/amount columns.
+            x_timestamp_list (List[pd.DatetimeIndex or Series]): List of timestamps corresponding to historical data, length should match the number of rows in each DataFrame.
+            y_timestamp_list (List[pd.DatetimeIndex or Series]): List of future prediction timestamps, length should equal pred_len.
+            pred_len (int): Number of prediction steps.
+            T (float): Sampling temperature.
+            top_k (int): Top-k filtering threshold.
+            top_p (float): Top-p (nucleus sampling) threshold.
+            sample_count (int): Number of parallel samples per series, automatically averaged internally.
+            verbose (bool): Whether to display autoregressive progress.
+
+        Returns:
+            List[pd.DataFrame]: List of prediction results in the same order as input, each DataFrame contains
+                                `open, high, low, close, volume, amount` columns, indexed by corresponding `y_timestamp`.
+        """
+        # Basic validation
+        if not isinstance(df_list, (list, tuple)) or not isinstance(x_timestamp_list, (list, tuple)) or not isinstance(y_timestamp_list, (list, tuple)):
+            raise ValueError("df_list, x_timestamp_list, y_timestamp_list must be list or tuple types.")
+        if not (len(df_list) == len(x_timestamp_list) == len(y_timestamp_list)):
+            raise ValueError("df_list, x_timestamp_list, y_timestamp_list must have consistent lengths.")
+
+        num_series = len(df_list)
+
+        x_list = []
+        x_stamp_list = []
+        y_stamp_list = []
+        means = []
+        stds = []
+        seq_lens = []
+        y_lens = []
+
+        for i in range(num_series):
+            df = df_list[i]
+            if not isinstance(df, pd.DataFrame):
+                raise ValueError(f"Input at index {i} is not a pandas DataFrame.")
+            if not all(col in df.columns for col in self.price_cols):
+                raise ValueError(f"DataFrame at index {i} is missing price columns {self.price_cols}.")
+
+            df = df.copy()
+            if self.vol_col not in df.columns:
+                df[self.vol_col] = 0.0
+                df[self.amt_vol] = 0.0
+            if self.amt_vol not in df.columns and self.vol_col in df.columns:
+                df[self.amt_vol] = df[self.vol_col] * df[self.price_cols].mean(axis=1)
+
+            if df[self.price_cols + [self.vol_col, self.amt_vol]].isnull().values.any():
+                raise ValueError(f"DataFrame at index {i} contains NaN values in price or volume columns.")
+
+            x_timestamp = x_timestamp_list[i]
+            y_timestamp = y_timestamp_list[i]
+
+            x_time_df = calc_time_stamps(x_timestamp)
+            y_time_df = calc_time_stamps(y_timestamp)
+
+            x = df[self.price_cols + [self.vol_col, self.amt_vol]].values.astype(np.float32)
+            x_stamp = x_time_df.values.astype(np.float32)
+            y_stamp = y_time_df.values.astype(np.float32)
+
+            if x.shape[0] != x_stamp.shape[0]:
+                raise ValueError(f"Inconsistent lengths at index {i}: x has {x.shape[0]} vs x_stamp has {x_stamp.shape[0]}.")
+            if y_stamp.shape[0] != pred_len:
+                raise ValueError(f"y_timestamp length at index {i} should equal pred_len={pred_len}, got {y_stamp.shape[0]}.")
+
+            x_mean, x_std = np.mean(x, axis=0), np.std(x, axis=0)
+            x_norm = (x - x_mean) / (x_std + 1e-5)
+            x_norm = np.clip(x_norm, -self.clip, self.clip)
+
+            x_list.append(x_norm)
+            x_stamp_list.append(x_stamp)
+            y_stamp_list.append(y_stamp)
+            means.append(x_mean)
+            stds.append(x_std)
+
+            seq_lens.append(x_norm.shape[0])
+            y_lens.append(y_stamp.shape[0])
+
+        # Require all series to have consistent historical and prediction lengths for batch processing
+        if len(set(seq_lens)) != 1:
+            raise ValueError(f"Parallel prediction requires all series to have consistent historical lengths, got: {seq_lens}")
+        if len(set(y_lens)) != 1:
+            raise ValueError(f"Parallel prediction requires all series to have consistent prediction lengths, got: {y_lens}")
+
+        x_batch = np.stack(x_list, axis=0).astype(np.float32)           # (B, seq_len, feat)
+        x_stamp_batch = np.stack(x_stamp_list, axis=0).astype(np.float32) # (B, seq_len, time_feat)
+        y_stamp_batch = np.stack(y_stamp_list, axis=0).astype(np.float32) # (B, pred_len, time_feat)
+
+        preds = self.generate(x_batch, x_stamp_batch, y_stamp_batch, pred_len, T, top_k, top_p, sample_count, verbose)
+        # preds: (B, pred_len, feat)
+
+        pred_dfs = []
+        for i in range(num_series):
+            preds_i = preds[i] * (stds[i] + 1e-5) + means[i]
+            pred_df = pd.DataFrame(preds_i, columns=self.price_cols + [self.vol_col, self.amt_vol], index=y_timestamp_list[i])
+            pred_dfs.append(pred_df)
+
+        return pred_dfs

skills/alphaear-predictor/scripts/utils/predictor/model/module.py

@@ -0,0 +1,562 @@
+import math
+
+from einops import rearrange, reduce
+import torch
+import torch.nn as nn
+from torch.autograd import Function
+import torch.nn.functional as F
+
+
+class DifferentiableEntropyFunction(Function):
+    @staticmethod
+    def forward(ctx, zq, basis, K, eps):
+        zb = (zq + 1) / 2
+        zi = ((zb * basis).sum(-1)).to(torch.int64)
+        cnt = torch.scatter_reduce(torch.zeros(2 ** K, device=zq.device, dtype=zq.dtype),
+                                   0,
+                                   zi.flatten(),
+                                   torch.ones_like(zi.flatten()).to(zq.dtype),
+                                   'sum')
+        prob = (cnt + eps) / (cnt + eps).sum()
+        H = -(prob * torch.log(prob)).sum()
+        ctx.save_for_backward(zq, zi, prob)
+        ctx.K = K
+        return H
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        zq, zi, prob = ctx.saved_tensors
+        grad_array = -grad_output * (torch.log(prob) + 1) / zi.numel() / ctx.K
+        reord_grad = grad_array[zi.flatten()].reshape(zi.shape)
+        grad_input = reord_grad.unsqueeze(-1) * zq
+        return grad_input, None, None, None, None
+
+
+def codebook_entropy(zq, basis, K, eps=1e-4):
+    return DifferentiableEntropyFunction.apply(zq, basis, K, eps)
+
+
+class BinarySphericalQuantizer(nn.Module):
+    def __init__(self, embed_dim, beta, gamma0, gamma, zeta,
+                 input_format='bchw',
+                 soft_entropy=True, group_size=9,
+                 persample_entropy_compute='analytical',
+                 cb_entropy_compute='group',
+                 l2_norm=True,
+                 inv_temperature=1):
+        """
+        Paper link: https://arxiv.org/pdf/2406.07548.pdf
+        Here we use the official implementation of the BinarySphericalQuantizer.
+        """
+        super().__init__()
+        self.embed_dim = embed_dim
+        self.beta = beta  # loss weight for commit loss
+        self.gamma0 = gamma0  # loss weight for entropy penalty
+        self.gamma = gamma  # loss weight for entropy penalty
+        self.zeta = zeta  # loss weight for entire entropy penalty
+        self.input_format = input_format
+        assert self.embed_dim % group_size == 0, "embed_dim must be divisible by group_size"
+        self.num_groups = self.embed_dim // group_size
+        self.group_size = group_size
+        assert persample_entropy_compute in ['group', 'analytical'], "persample_entropy_compute must be either 'group' or 'analytical'"
+        assert cb_entropy_compute in ['group', 'nce'], "cb_entropy_compute must be either 'group' or 'nce'"
+        self.persample_entropy_compute = persample_entropy_compute
+        self.cb_entropy_compute = cb_entropy_compute
+        self.l2_norm = l2_norm
+        self.inv_temperature = inv_temperature
+
+        self.register_buffer('basis', 2 ** torch.arange(embed_dim - 1, -1, -1))
+        self.register_buffer('group_basis', 2 ** torch.arange(group_size - 1, -1, -1))
+
+        self.num_dimensions = 2 ** embed_dim
+        self.bits_per_index = embed_dim
+
+        # we only need to keep the codebook portion up to the group size
+        # because we approximate the H loss with this subcode
+        group_codes = torch.arange(2 ** self.group_size)
+        group_codebook = self.indexes_to_codes(group_codes).float()[:, -group_size:]
+        self.register_buffer('group_codebook', group_codebook, persistent=False)
+
+        self.soft_entropy = soft_entropy  # soft_entropy: Sec 3.2 of https://arxiv.org/pdf/1911.05894.pdf
+
+    def quantize(self, z):
+        assert z.shape[-1] == self.embed_dim, f"Expected {self.embed_dim} dimensions, got {z.shape[-1]}"
+
+        zhat = torch.where(z > 0,
+                           torch.tensor(1, dtype=z.dtype, device=z.device),
+                           torch.tensor(-1, dtype=z.dtype, device=z.device))
+        return z + (zhat - z).detach()
+
+    def forward(self, z, collect_metrics=True):
+        # if self.input_format == 'bchw':
+        #     z = rearrange(z, 'b c h w -> b h w c')
+        zq = self.quantize(z)
+
+        q_scale = 1. / (self.embed_dim ** 0.5) if self.l2_norm else 1.
+
+        zq = zq * q_scale
+
+        if not collect_metrics:
+            return zq, zq.new_zeros(()), {}
+
+        indices = self.codes_to_indexes(zq.detach())
+        group_indices = self.codes_to_group_indexes(zq.detach())
+        if not self.training:
+            used_codes = torch.unique(indices, return_counts=False)
+        else:
+            used_codes = None
+
+        if self.soft_entropy:
+            persample_entropy, cb_entropy, avg_prob = self.soft_entropy_loss(z)
+            entropy_penalty = self.gamma0 * persample_entropy - self.gamma * cb_entropy
+        else:
+            zb_by_sample = ((zq + 1) / 2).reshape(z.shape[0], -1, z.shape[-1]).to(torch.float32)
+            persample_entropy = self.get_hard_per_sample_entropy(zb_by_sample)
+            cb_entropy = codebook_entropy(zq, self.basis, self.embed_dim)
+            entropy_penalty = self.gamma0 * persample_entropy - self.gamma * cb_entropy
+
+        # commit loss
+        commit_loss = self.beta * torch.mean(((zq.detach() - z) ** 2).sum(dim=-1))
+
+        # if self.input_format == 'bchw':
+        #     zq = rearrange(zq, 'b h w c -> b c h w')
+
+        return (
+            zq,
+            commit_loss + self.zeta * entropy_penalty / self.inv_temperature,
+            {"H": cb_entropy, "used_codes": used_codes, "indices": indices, "group_indices": group_indices,
+             "avg_prob": avg_prob}
+        )
+
+    def soft_entropy_loss(self, z):
+        # if we divide the code in subgroups of size group_size, the codebook will be of size 2 ** group_size
+        # the sub-code is the last group_size bits of the full code
+        group_code_book = self.group_codebook / (self.embed_dim ** 0.5 if self.l2_norm else 1)
+        divided_z = rearrange(z, '... (g c) -> ... g c', c=self.group_size)
+
+        # we calculate the distance between the divided_z and the codebook for each subgroup
+        distance = - 2 * torch.einsum('... g c, d c ->... g d', divided_z, group_code_book)
+        prob = (-distance * self.inv_temperature).softmax(dim=-1)
+        if self.persample_entropy_compute == 'analytical':
+            if self.l2_norm:
+                p = torch.sigmoid(-4 * z / (self.embed_dim ** 0.5) * self.inv_temperature)
+            else:
+                p = torch.sigmoid(-4 * z * self.inv_temperature)
+            prob = torch.stack([p, 1 - p], dim=-1)
+            per_sample_entropy = self.get_entropy(prob, dim=-1, normalize=False).sum(dim=-1).mean()
+        else:
+            per_sample_entropy = self.get_entropy(prob, dim=-1, normalize=False).sum(dim=-1).mean()
+
+        # macro average of the probability of each subgroup
+        avg_prob = reduce(prob, '... g d ->g d', 'mean')
+        codebook_entropy = self.get_entropy(avg_prob, dim=-1, normalize=False)
+
+        # the approximation of the entropy is the sum of the entropy of each subgroup
+        return per_sample_entropy, codebook_entropy.sum(), avg_prob
+
+    def get_hard_per_sample_entropy(self, zb_by_sample):
+        probs_per_dim = zb_by_sample.sum(1) / zb_by_sample.shape[1]
+        persample_entropy = - probs_per_dim * torch.log(probs_per_dim + 1e-8) - (1 - probs_per_dim) * torch.log(1 - probs_per_dim + 1e-8)
+        persample_entropy = persample_entropy.sum(-1)
+        return persample_entropy.mean()
+
+    def codes_to_indexes(self, zhat):
+        """Converts a `code` to an index in the codebook.
+        Args:
+            zhat: A tensor of shape (B, ..., C) containing the codes. must be in {-1, 1}
+        """
+        assert zhat.shape[-1] == self.embed_dim, f"Expected {self.embed_dim} dimensions, got {zhat.shape[-1]}"
+        return ((zhat + 1) / 2 * self.basis).sum(axis=-1).to(torch.int64)
+
+    def codes_to_group_indexes(self, zhat):
+        """Converts a `code` to a list of indexes (in groups) in the codebook.
+        Args:
+            zhat: A tensor of shape (B, ..., C) containing the codes. must be in {-1, 1}
+        """
+        zhat_in_group = rearrange(zhat, 'b ... (g c) -> b ... g c', c=self.group_size)
+        return ((zhat_in_group + 1) / 2 * self.group_basis).sum(axis=-1).to(torch.int64)
+
+    def indexes_to_codes(self, indices):
+        """Inverse of `indexes_to_codes`."""
+        indices = indices.unsqueeze(-1)
+        codes_non_centered = torch.remainder(
+            torch.floor_divide(indices, self.basis), 2
+        )
+        return codes_non_centered * 2 - 1
+
+    def group_indexes_to_codes(self, group_indices):
+        """Inverse of `group_indexes_to_codes`."""
+        group_indices = group_indices.unsqueeze(-1)
+        codes_non_centered = torch.remainder(
+            torch.floor_divide(group_indices, self.group_basis), 2
+        )
+        codes_non_centered = rearrange(codes_non_centered, 'b ... g c -> b ... (g c)')
+        return codes_non_centered * 2 - 1
+
+    def get_entropy(self, count, dim=-1, eps=1e-4, normalize=True):
+        if normalize:
+            probs = (count + eps) / (count + eps).sum(dim=dim, keepdim=True)
+        else:
+            probs = count
+        H = -(probs * torch.log(probs + 1e-8)).sum(dim=dim)
+        return H
+
+    def get_group_codebook_entry(self, group_indices):
+        z_q = self.group_indexes_to_codes(group_indices)
+        q_scale = 1. / (self.embed_dim ** 0.5) if self.l2_norm else 1.
+        z_q = z_q * q_scale
+        if self.input_format == 'bchw':
+            h, w = int(z_q.shape[1] ** 0.5)
+            assert h * w == z_q.shape[1], 'Invalid sequence length'
+            z_q = rearrange(z_q, 'b (h w) c -> b c h w', h=h)
+        return z_q
+
+    def get_codebook_entry(self, indices):
+        z_q = self.indexes_to_codes(indices)
+        q_scale = 1. / (self.embed_dim ** 0.5) if self.l2_norm else 1.
+        z_q = z_q * q_scale
+        if self.input_format == 'bchw':
+            h, w = int(z_q.shape[1] ** 0.5)
+            assert h * w == z_q.shape[1], 'Invalid sequence length'
+            z_q = rearrange(z_q, 'b (h w) c -> b c h w', h=h)
+        return z_q
+
+
+class BSQuantizer(nn.Module):
+
+    def __init__(self, s1_bits, s2_bits, beta, gamma0, gamma, zeta, group_size):
+        super().__init__()
+        self.codebook_dim = s1_bits + s2_bits
+        self.s1_bits = s1_bits
+        self.s2_bits = s2_bits
+        self.bsq = BinarySphericalQuantizer(self.codebook_dim, beta, gamma0, gamma, zeta, group_size=group_size)
+
+    def bits_to_indices(self, bits):
+        bits = (bits >= 0).to(torch.long)
+        indices = 2 ** torch.arange(
+            0,
+            bits.shape[-1],
+            1,
+            dtype=torch.long,
+            device=bits.device,
+        )
+        return (bits * indices).sum(-1)
+
+    def forward(self, z, half=False, collect_metrics=True):
+        z = F.normalize(z, dim=-1)
+        quantized, bsq_loss, metrics = self.bsq(z, collect_metrics=collect_metrics)
+        if half:
+            q_pre = quantized[:, :, :self.s1_bits]
+            q_post = quantized[:, :, self.s1_bits:]
+            z_indices = [self.bits_to_indices(q_pre), self.bits_to_indices(q_post)]
+        else:
+            z_indices = self.bits_to_indices(quantized)
+        return bsq_loss, quantized, z_indices
+
+
+class RMSNorm(torch.nn.Module):
+    def __init__(self, dim: int, eps: float = 1e-5):
+        super().__init__()
+        self.eps = eps
+        self.weight = nn.Parameter(torch.ones(dim))
+
+    def _norm(self, x):
+        return x * torch.rsqrt(torch.mean(x * x, dim=-1, keepdim=True) + self.eps)
+
+    def forward(self, x):
+        output = self._norm(x.float()).type_as(x)
+        return output * self.weight
+
+
+class FeedForward(nn.Module):
+    def __init__(self, d_model, ff_dim, ffn_dropout_p=0.0):
+        super().__init__()
+
+        self.w1 = nn.Linear(d_model, ff_dim, bias=False)
+        self.w3 = nn.Linear(d_model, ff_dim, bias=False)
+        self.w2 = nn.Linear(ff_dim, d_model, bias=False)
+        self.ffn_dropout = nn.Dropout(ffn_dropout_p)
+
+    def forward(self, x):
+        return self.ffn_dropout(self.w2(F.silu(self.w1(x)) * self.w3(x)))
+
+
+class RotaryPositionalEmbedding(nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+        inv_freq = 1.0 / (10000 ** (torch.arange(0, dim, 2).float() / dim))
+        self.register_buffer("inv_freq", inv_freq)
+        self.seq_len_cached = None
+        self.cos_cached = None
+        self.sin_cached = None
+
+    def _update_cos_sin_cache(self, x, seq_len):
+        if seq_len != self.seq_len_cached:
+            self.seq_len_cached = seq_len
+            t = torch.arange(seq_len, device=x.device).type_as(self.inv_freq)
+            freqs = torch.einsum('i,j->ij', t, self.inv_freq)
+            emb = torch.cat((freqs, freqs), dim=-1).to(x.device)
+            self.cos_cached = emb.cos()[None, None, :, :]
+            self.sin_cached = emb.sin()[None, None, :, :]
+        return self.cos_cached, self.sin_cached
+
+    def forward(self, q, k):
+        cos, sin = self._update_cos_sin_cache(q, q.shape[-2])
+        return (
+            (q * cos) + (self._rotate_half(q) * sin),
+            (k * cos) + (self._rotate_half(k) * sin),
+        )
+
+    def _rotate_half(self, x):
+        x1, x2 = x.chunk(2, dim=-1)
+        return torch.cat((-x2, x1), dim=-1)
+
+
+class MultiHeadAttentionWithRoPE(nn.Module):
+    def __init__(self, d_model, n_heads, attn_dropout_p=0.0, resid_dropout_p=0.0):
+        super().__init__()
+        self.d_model = d_model
+        self.n_heads = n_heads
+        self.head_dim = d_model // n_heads
+
+        self.q_proj = nn.Linear(d_model, d_model)
+        self.k_proj = nn.Linear(d_model, d_model)
+        self.v_proj = nn.Linear(d_model, d_model)
+        self.out_proj = nn.Linear(d_model, d_model)
+        self.rotary = RotaryPositionalEmbedding(self.head_dim)
+        self.attn_dropout_p = attn_dropout_p
+        self.resid_dropout = nn.Dropout(resid_dropout_p)
+
+    def forward(self, x, key_padding_mask=None):
+        batch_size, seq_len, _ = x.shape
+
+        q = self.q_proj(x).view(batch_size, seq_len, self.n_heads, self.head_dim).transpose(1, 2)
+        k = self.k_proj(x).view(batch_size, seq_len, self.n_heads, self.head_dim).transpose(1, 2)
+        v = self.v_proj(x).view(batch_size, seq_len, self.n_heads, self.head_dim).transpose(1, 2)
+
+        q, k = self.rotary(q, k)
+
+        if key_padding_mask is not None:
+            attn_mask = key_padding_mask.unsqueeze(1).unsqueeze(2)  # [batch, 1, 1, seq_len]
+            attn_mask = attn_mask.expand(-1, self.n_heads, seq_len, -1)  # [batch, n_heads, q_len, k_len]
+        else:
+            attn_mask = None
+
+        attn_output = F.scaled_dot_product_attention(
+            q, k, v,
+            attn_mask=attn_mask,
+            dropout_p=self.attn_dropout_p if self.training else 0.0,
+            is_causal=True
+        )
+
+        attn_output = attn_output.transpose(1, 2).contiguous().view(batch_size, seq_len, self.d_model)
+        return self.resid_dropout(self.out_proj(attn_output))
+
+
+class MultiHeadCrossAttentionWithRoPE(nn.Module):
+    def __init__(self, d_model, n_heads, attn_dropout_p=0.0, resid_dropout=0.0):
+        super().__init__()
+        self.d_model = d_model
+        self.n_heads = n_heads
+        self.head_dim = d_model // n_heads
+
+        self.q_proj = nn.Linear(d_model, d_model)
+        self.k_proj = nn.Linear(d_model, d_model)
+        self.v_proj = nn.Linear(d_model, d_model)
+        self.out_proj = nn.Linear(d_model, d_model)
+        self.rotary = RotaryPositionalEmbedding(self.head_dim)
+        self.attn_dropout_p = attn_dropout_p
+        self.resid_dropout = nn.Dropout(resid_dropout)
+
+    def forward(self, query, key, value, key_padding_mask=None):
+        batch_size, q_len, _ = query.shape
+        _, seq_len, _ = key.shape
+
+        q = self.q_proj(query).view(batch_size, q_len, self.n_heads, self.head_dim).transpose(1, 2)
+        k = self.k_proj(key).view(batch_size, seq_len, self.n_heads, self.head_dim).transpose(1, 2)
+        v = self.v_proj(value).view(batch_size, seq_len, self.n_heads, self.head_dim).transpose(1, 2)
+
+        q, k = self.rotary(q, k)
+
+        if key_padding_mask is not None:
+            attn_mask = key_padding_mask.unsqueeze(1).unsqueeze(2)
+            attn_mask = attn_mask.expand(-1, self.n_heads, q_len, -1)
+        else:
+            attn_mask = None
+
+        is_causal_flag = self.training
+
+        attn_output = F.scaled_dot_product_attention(
+            q, k, v,
+            attn_mask=attn_mask,
+            dropout_p=self.attn_dropout_p if self.training else 0.0,
+            is_causal=is_causal_flag
+        )
+
+        attn_output = attn_output.transpose(1, 2).contiguous().view(batch_size, q_len, self.d_model)
+        return self.resid_dropout(self.out_proj(attn_output))
+
+
+class HierarchicalEmbedding(nn.Module):
+    def __init__(self, s1_bits, s2_bits, d_model=256):
+        super().__init__()
+        self.s1_bits = s1_bits
+        self.s2_bits = s2_bits
+
+        vocab_s1 = 2 ** s1_bits
+        vocab_s2 = 2 ** s2_bits
+
+        self.emb_s1 = nn.Embedding(vocab_s1, d_model)
+        self.emb_s2 = nn.Embedding(vocab_s2, d_model)
+        self.d_model = d_model
+        self.fusion_proj = nn.Linear(d_model * 2, d_model)
+
+        nn.init.normal_(self.emb_s1.weight, mean=0, std=d_model ** -0.5)
+        nn.init.normal_(self.emb_s2.weight, mean=0, std=d_model ** -0.5)
+
+    def split_token(self, token_ids: torch.Tensor, s2_bits: int):
+        """Inputs:
+            token_ids (torch.Tensor): Composite token IDs of shape [batch_size, seq_len] or [N], each in range [0, 2^(s1_bits + s2_bits) - 1].
+            s2_bits (int): Number of low bits used for the fine token (s2).
+        """
+        assert isinstance(s2_bits, int) and s2_bits > 0, "s2_bits must be a positive integer"
+
+        t = token_ids.long()
+        mask = (1 << s2_bits) - 1
+        s2_ids = t & mask           # extract low bits
+        s1_ids = t >> s2_bits       # extract high bits
+        return s1_ids, s2_ids
+
+    def forward(self, token_ids):
+        """Inputs:
+        token_ids:
+            - tuple or list: (s1_ids, s2_ids), each of shape [batch_size, seq_len], or
+            - torch.Tensor: composite token IDs of shape [batch_size, seq_len], which will be split into (s1_ids, s2_ids) internally.
+        Output: [batch_size, seq_len, d_model]
+        """
+        if isinstance(token_ids, tuple) or isinstance(token_ids, list):
+            s1_ids, s2_ids = token_ids
+        else:
+            s1_ids, s2_ids = self.split_token(token_ids, self.s2_bits)
+        s1_emb = self.emb_s1(s1_ids) * math.sqrt(self.d_model)
+        s2_emb = self.emb_s2(s2_ids) * math.sqrt(self.d_model)
+        return self.fusion_proj(torch.cat([s1_emb, s2_emb], dim=-1))
+
+
+class DependencyAwareLayer(nn.Module):
+    def __init__(self, d_model, n_heads=4, attn_dropout_p=0.0, resid_dropout=0.0):
+        super().__init__()
+        self.cross_attn = MultiHeadCrossAttentionWithRoPE(d_model, n_heads, attn_dropout_p, resid_dropout)
+        self.norm = RMSNorm(d_model)
+
+    def forward(self, hidden_states, sibling_embed, key_padding_mask=None):
+        """hidden_states: [batch, seq_len, d_model]
+        sibling_embed: Embedding from another subtoken
+        """
+        attn_out = self.cross_attn(
+            query=sibling_embed,
+            key=hidden_states,
+            value=hidden_states,
+            key_padding_mask=key_padding_mask
+        )
+        return self.norm(hidden_states + attn_out)
+
+
+class TransformerBlock(nn.Module):
+    def __init__(self, d_model, n_heads, ff_dim=1024, ffn_dropout_p=0.0, attn_dropout_p=0.0, resid_dropout_p=0.0):
+        super().__init__()
+        self.norm1 = RMSNorm(d_model)
+        self.self_attn = MultiHeadAttentionWithRoPE(d_model, n_heads, attn_dropout_p, resid_dropout_p)
+        self.norm2 = RMSNorm(d_model)
+        self.ffn = FeedForward(d_model, ff_dim, ffn_dropout_p)
+
+    def forward(self, x, key_padding_mask=None):
+        residual = x
+        x = self.norm1(x)
+        attn_out = self.self_attn(x, key_padding_mask=key_padding_mask)
+        x = residual + attn_out
+
+        residual = x
+        x = self.norm2(x)
+        ffn_out = self.ffn(x)
+        x = residual + ffn_out
+        return x
+
+
+class DualHead(nn.Module):
+    def __init__(self, s1_bits, s2_bits, d_model):
+        super().__init__()
+        self.vocab_s1 = 2 ** s1_bits
+        self.vocab_s2 = 2 ** s2_bits
+        self.proj_s1 = nn.Linear(d_model, self.vocab_s1)
+        self.proj_s2 = nn.Linear(d_model, self.vocab_s2)
+
+    def compute_loss(self, s1_logits, s2_logits, s1_targets, s2_targets, padding_mask=None):
+        if padding_mask is not None:
+            valid_mask = (padding_mask == 0)
+            s1_logits = s1_logits[valid_mask]
+            s2_logits = s2_logits[valid_mask]
+            s1_targets = s1_targets[valid_mask]
+            s2_targets = s2_targets[valid_mask]
+            ce_s1 = F.cross_entropy(s1_logits, s1_targets)
+            ce_s2 = F.cross_entropy(s2_logits, s2_targets)
+        else:
+            ce_s1 = F.cross_entropy(s1_logits.reshape(-1, self.vocab_s1), s1_targets.reshape(-1))
+            ce_s2 = F.cross_entropy(s2_logits.reshape(-1, self.vocab_s2), s2_targets.reshape(-1))
+        ce_loss = (ce_s1 + ce_s2) / 2
+        return ce_loss, ce_s1, ce_s2
+
+    def forward(self, x):
+        return self.proj_s1(x)
+
+    def cond_forward(self, x2):
+        return self.proj_s2(x2)
+
+
+class FixedEmbedding(nn.Module):
+    def __init__(self, c_in, d_model):
+        super(FixedEmbedding, self).__init__()
+
+        w = torch.zeros(c_in, d_model).float()
+        w.require_grad = False
+
+        position = torch.arange(0, c_in).float().unsqueeze(1)
+        div_term = (torch.arange(0, d_model, 2).float() * -(math.log(10000.0) / d_model)).exp()
+
+        w[:, 0::2] = torch.sin(position * div_term)
+        w[:, 1::2] = torch.cos(position * div_term)
+
+        self.emb = nn.Embedding(c_in, d_model)
+        self.emb.weight = nn.Parameter(w, requires_grad=False)
+
+    def forward(self, x):
+        return self.emb(x).detach()
+
+
+class TemporalEmbedding(nn.Module):
+    def __init__(self, d_model, learn_pe):
+        super(TemporalEmbedding, self).__init__()
+
+        minute_size = 60
+        hour_size = 24
+        weekday_size = 7
+        day_size = 32
+        month_size = 13
+
+        Embed = FixedEmbedding if not learn_pe else nn.Embedding
+        self.minute_embed = Embed(minute_size, d_model)
+        self.hour_embed = Embed(hour_size, d_model)
+        self.weekday_embed = Embed(weekday_size, d_model)
+        self.day_embed = Embed(day_size, d_model)
+        self.month_embed = Embed(month_size, d_model)
+
+    def forward(self, x):
+        x = x.long()
+
+        minute_x = self.minute_embed(x[:, :, 0])
+        hour_x = self.hour_embed(x[:, :, 1])
+        weekday_x = self.weekday_embed(x[:, :, 2])
+        day_x = self.day_embed(x[:, :, 3])
+        month_x = self.month_embed(x[:, :, 4])
+
+        return hour_x + weekday_x + day_x + month_x + minute_x

skills/alphaear-predictor/scripts/utils/predictor/training.py

@@ -0,0 +1,539 @@
+import os
+import sys
+import time
+import torch
+import torch.nn as nn
+import pandas as pd
+import numpy as np
+import json
+import random
+from loguru import logger
+from datetime import datetime, timedelta
+from sentence_transformers import SentenceTransformer
+from dotenv import load_dotenv
+
+load_dotenv(os.path.expanduser("~/.config/opencode/.env"))
+
+# Setup paths
+KRONOS_DIR = os.path.dirname(os.path.abspath(__file__))
+SRC_DIR = os.path.dirname(os.path.dirname(KRONOS_DIR))
+if SRC_DIR not in sys.path:
+    sys.path.insert(0, SRC_DIR)
+
+from ..kronos.model import Kronos, KronosTokenizer, KronosPredictor
+from ..database_manager import DatabaseManager
+from ..stock_tools import StockTools
+from ..search_tools import SearchTools
+from ..llm.factory import get_model
+from ..visualizer import VisualizerTools
+from ..schema.models import ForecastResult, KLinePoint
+from agno.agent import Agent
+
+
+class AutoSynthesisTrainer:
+    def __init__(self, news_dim=384):
+        self.device = (
+            "cuda"
+            if torch.cuda.is_available()
+            else "mps"
+            if torch.backends.mps.is_available()
+            else "cpu"
+        )
+        self.db = DatabaseManager()
+        self.tools = StockTools(self.db)
+        self.searcher = SearchTools(self.db)
+        # Try loading from local cache first to avoid network timeouts
+        model_name = os.getenv(
+            "EMBEDDING_MODEL", "sentence-transformers/all-MiniLM-L6-v2"
+        )
+        try:
+            logger.info(f"🔄 Attempting to load {model_name} from local cache...")
+            self.embedder = SentenceTransformer(
+                model_name, device=self.device, local_files_only=True
+            )
+            logger.success("✅ Model loaded from local cache.")
+        except Exception:
+            logger.warning(
+                "⚠️ Local cache not found or incomplete. Attempting to download..."
+            )
+            self.embedder = SentenceTransformer(model_name, device=self.device)
+        self.news_dim = news_dim
+
+        # Try loading from local cache first to avoid network timeouts
+        try:
+            logger.info(
+                "🔄 Attempting to load Kronos and Tokenizer from local cache..."
+            )
+            self.tokenizer = KronosTokenizer.from_pretrained(
+                "NeoQuasar/Kronos-Tokenizer-base", local_files_only=True
+            ).to(self.device)
+            base_model = Kronos.from_pretrained(
+                "NeoQuasar/Kronos-base", local_files_only=True
+            )
+            logger.success("✅ Kronos and Tokenizer loaded from local cache.")
+        except Exception:
+            logger.warning(
+                "⚠️ Local Kronos/Tokenizer not found or incomplete. Attempting to download..."
+            )
+            self.tokenizer = KronosTokenizer.from_pretrained(
+                "NeoQuasar/Kronos-Tokenizer-base"
+            ).to(self.device)
+            base_model = Kronos.from_pretrained("NeoQuasar/Kronos-base")
+
+        self.model = Kronos(
+            base_model.s1_bits,
+            base_model.s2_bits,
+            base_model.n_layers,
+            base_model.d_model,
+            base_model.n_heads,
+            base_model.ff_dim,
+            base_model.ffn_dropout_p,
+            base_model.attn_dropout_p,
+            base_model.resid_dropout_p,
+            base_model.token_dropout_p,
+            base_model.learn_te,
+            news_dim=self.news_dim,
+        ).to(self.device)
+        self.model.load_state_dict(base_model.state_dict(), strict=False)
+
+        # LLM for causality verification
+        provider = os.getenv("LLM_PROVIDER", "ust")
+        model_id = os.getenv("LLM_MODEL", "Qwen")
+        self.llm_agent = Agent(model=get_model(provider, model_id))
+
+    def discover_shocks(
+        self, ticker_list, threshold=2.0, limit_per_stock=5, days=365, pred_len=5
+    ):
+        """1. Find days with significant price movements (Look back 1 year)"""
+        shocks = []
+        end_date = datetime.now().strftime("%Y-%m-%d")
+        start_date = (datetime.now() - timedelta(days=days)).strftime("%Y-%m-%d")
+
+        for ticker in ticker_list:
+            df = self.tools.get_stock_price(
+                ticker, start_date=start_date, end_date=end_date
+            )
+            if df.empty or len(df) < 60:
+                continue
+
+            # Look for big moves
+            moves = df[df["change_pct"].abs() > threshold].copy()
+            if moves.empty:
+                continue
+
+            count = 0
+            for idx, row in moves.iterrows():
+                # Ensure we have history before this day AND enough future days for eval
+                date_idx = df.index.get_loc(idx)
+                if date_idx < 50 or date_idx + pred_len > len(df):
+                    continue
+
+                shocks.append(
+                    {
+                        "ticker": ticker,
+                        "date": row["date"],
+                        "change": row["change_pct"],
+                        "history": df.iloc[date_idx - 50 : date_idx],
+                        "target": df.iloc[
+                            date_idx : date_idx + pred_len
+                        ],  # Now capturing pred_len days
+                    }
+                )
+                count += 1
+                if count >= limit_per_stock:
+                    break
+
+        logger.info(
+            f"✨ Discovered {len(shocks)} potential price shocks over the last {days} days."
+        )
+        return shocks
+
+    def find_reason_and_verify(self, shock):
+        """2. Search for reasons and verify causality using LLM"""
+        ticker_info = self.db.get_stock_by_code(shock["ticker"])
+        name = ticker_info["name"] if ticker_info else shock["ticker"]
+        date_str = shock["date"]
+
+        # Try multiple query variations and engines
+        queries = [
+            f"{name} ({shock['ticker']}) {date_str} 为什么涨跌 原因",
+            f"{name} {date_str} 异动 原因",
+            f"{shock['ticker']} {date_str} 新闻",
+        ]
+
+        search_results = []
+        for query in queries:
+            logger.info(f"🔍 Searching for reason: {query}")
+            # Try alternate engines
+            for engine in ["baidu"]:
+                try:
+                    results = self.searcher.search_list(
+                        query, engine=engine, max_results=3, enrich=False
+                    )
+                    if results:
+                        search_results = results
+                        break
+                except Exception as e:
+                    logger.warning(f"Search failed for {query} on {engine}: {e}")
+
+            if search_results:
+                break
+            time.sleep(random.uniform(1.0, 2.0))
+
+        if not search_results:
+            logger.warning(
+                f"⚠️ No search results found for {name} on {date_str} after multiple attempts."
+            )
+            return None
+
+        context = "\n".join(
+            [f"- {r['title']}: {r.get('content', '')[:300]}" for r in search_results]
+        )
+
+        prompt = f"""
+        任务：判断以下新闻是否解释了该股票在 {date_str} 的 {shock["change"]:.2f}% 价格变动。
+        
+        股票：{name}
+        日期：{date_str}
+        变动：{shock["change"]:.2f}%
+        
+        搜索结果：
+        {context}
+        
+        要求：
+        1. 该新闻是否在该日期左右发生？
+        2. 该新闻是否能逻辑上解释这种大幅波动（如财报、利好政策、重组、大环境暴跌等）？
+        3. 如果是，请总结一段 100 字以内的“核心推动原因”。
+        4. 返回 JSON: {{"is_causal": true/false, "summary": "原因摘要"}}
+        """
+
+        try:
+            res = self.llm_agent.run(prompt)
+            data = json.loads(
+                res.content.replace("```json", "").replace("```", "").strip()
+            )
+            if data.get("is_causal"):
+                logger.success(
+                    f"✅ Verified cause for {name} on {date_str}: {data['summary']}"
+                )
+                return data["summary"]
+            else:
+                logger.warning(
+                    f"❌ Verified cause for {name} on {date_str}: {data['summary']}"
+                )
+                return None
+        except Exception as e:
+            logger.warning(f"Verification failed: {e}")
+        return None
+
+    def save_model(self, path=None):
+        """Save the news_proj weights"""
+        if path is None:
+            save_dir = os.path.join(SRC_DIR, "exports/models")
+            os.makedirs(save_dir, exist_ok=True)
+            path = os.path.join(
+                save_dir, f"kronos_news_v1_{datetime.now().strftime('%Y%m%d_%H%M')}.pt"
+            )
+
+        # We only really need to save the news_proj part as it's the only one we train
+        torch.save(
+            {
+                "news_proj_state_dict": self.model.news_proj.state_dict(),
+                "news_dim": self.news_dim,
+                "d_model": self.model.d_model,
+            },
+            path,
+        )
+        logger.success(f"💾 Model weights saved to {path}")
+        return path
+
+    def run_synthesis_and_train(self, tickers, pred_len=5):
+        # 1. Discovery
+        shocks = self.discover_shocks(tickers, pred_len=pred_len)
+        print(f"find {len(shocks)} shocks")
+
+        # 2. News Association & Verification
+        dataset = []
+        max_news_items = 200  # Limit to 200 news items per session to avoid search bans
+
+        logger.info(
+            f"🧬 Starting News Association for {len(shocks)} shocks (Max limit: {max_news_items})"
+        )
+
+        for i, shock in enumerate(shocks):
+            if len(dataset) >= max_news_items:
+                logger.info("Reached maximum news items limit for this session.")
+                break
+
+            summary = self.find_reason_and_verify(shock)
+            if summary:
+                # 3. Embedding news
+                emb = self.embedder.encode(summary)
+                dataset.append(
+                    {
+                        "history": shock["history"],
+                        "target": shock["target"],
+                        "news_emb": emb,
+                        "summary": summary,
+                    }
+                )
+
+            # Add delay after search with randomness to avoid being blocked
+            if i < len(shocks) - 1:
+                delay = random.uniform(2.0, 4.0)
+                time.sleep(delay)
+
+        if not dataset:
+            logger.error(
+                "❌ No verified news-price pairs found. Adjust threshold or check if news is available in that period."
+            )
+            return
+
+        # 4. Train/Val Split
+        random.seed(42)
+        random.shuffle(dataset)
+
+        if len(dataset) < 2:
+            train_set = dataset
+            val_set = []
+            logger.warning(
+                f"⚠️ Only {len(dataset)} sample(s) found. Training on all, skipping validation."
+            )
+        else:
+            split_idx = max(1, int(len(dataset) * 0.8))
+            if split_idx >= len(dataset):
+                split_idx = len(dataset) - 1
+
+            train_set = dataset[:split_idx]
+            val_set = dataset[split_idx:]
+            logger.info(
+                f"🏗️ Dataset Split: {len(train_set)} samples for training, {len(val_set)} for validation."
+            )
+
+        if not train_set:
+            logger.error("❌ No samples for training.")
+            return
+
+        # 5. Training (Few-shot)
+        optimizer = torch.optim.Adam(self.model.news_proj.parameters(), lr=1e-3)
+        criterion = nn.CrossEntropyLoss()
+        self.model.train()
+
+        loss_history = []
+        logger.info(f"🚀 Training for 30 epochs...")
+        for epoch in range(30):
+            total_loss = 0
+            for item in train_set:
+                optimizer.zero_grad()
+
+                # Prep Data
+                hist_df = item["history"]
+                # For training, we still focus on the immediate next point (teacher forcing)
+                target_df = item["target"].iloc[:1]
+
+                hist_raw = hist_df[
+                    ["open", "high", "low", "close", "volume"]
+                ].values.astype(np.float32)
+                hist_raw = np.column_stack([hist_raw, hist_raw[:, 3] * hist_raw[:, 4]])
+
+                mean, std = hist_raw.mean(axis=0), hist_raw.std(axis=0) + 1e-5
+                hist_norm = (
+                    torch.from_numpy((hist_raw - mean) / std)
+                    .unsqueeze(0)
+                    .to(self.device)
+                )
+
+                target_raw = target_df[
+                    ["open", "high", "low", "close", "volume"]
+                ].values.astype(np.float32)
+                target_raw = np.column_stack(
+                    [target_raw, target_raw[:, 3] * target_raw[:, 4]]
+                )
+                target_norm = (
+                    torch.from_numpy((target_raw - mean) / std)
+                    .unsqueeze(0)
+                    .to(self.device)
+                )
+
+                with torch.no_grad():
+                    z_indices = self.tokenizer.encode(hist_norm, half=True)
+                    t_indices = self.tokenizer.encode(target_norm, half=True)
+                    s1_ids, s2_ids = z_indices[0], z_indices[1]
+                    t_s1, t_s2 = t_indices[0], t_indices[1]
+
+                news_t = torch.from_numpy(item["news_emb"]).unsqueeze(0).to(self.device)
+                s1_logits, s2_logits = self.model(
+                    s1_ids,
+                    s2_ids,
+                    news_emb=news_t,
+                    use_teacher_forcing=True,
+                    s1_targets=t_s1,
+                )
+
+                loss = (
+                    criterion(s1_logits[:, -1, :], t_s1[:, 0])
+                    + criterion(s2_logits[:, -1, :], t_s2[:, 0])
+                ) / 2
+                loss.backward()
+                optimizer.step()
+                total_loss += loss.item()
+
+            avg_epoch_loss = total_loss / max(1, len(train_set))
+            loss_history.append(avg_epoch_loss)
+
+            if (epoch + 1) % 10 == 0:
+                logger.info(f"Epoch {epoch + 1} Loss: {avg_epoch_loss:.4f}")
+
+        # 5.1 Visualize Loss Curve
+        loss_chart = VisualizerTools.generate_loss_chart(loss_history)
+        VisualizerTools.render_chart_to_file(
+            loss_chart,
+            os.path.join(SRC_DIR, "exports/training_results/loss_curve.html"),
+        )
+
+        # 5.2 Save final model
+        self.save_model()
+
+        # 6. Final Evaluation on Validation Set
+        if not val_set:
+            logger.warning("⚠️ Validation set is empty. Skipping statistical analysis.")
+            return
+
+        logger.info(
+            f"🧪 Final Evaluation: Base vs News-Integrated ({pred_len}-day Window)"
+        )
+        self.model.eval()
+        predictor = KronosPredictor(self.model, self.tokenizer, device=self.device)
+
+        base_maes = []
+        news_maes = []
+
+        print("\n" + "=" * 90)
+        print(
+            f"{'Date':<12} | {'Ticker':<8} | {'Base MAE':<15} | {'News MAE':<15} | {'Improvement'}"
+        )
+        print("-" * 90)
+
+        for item in val_set:
+            h = item["history"]
+            t = item["target"]
+            actuals = t["close"].values[:pred_len]
+
+            x_ts = pd.to_datetime(h["date"])
+            # Future timestamps: handle business days if possible, or just simple offset
+            future_dates = pd.date_range(
+                start=x_ts.iloc[-1] + timedelta(days=1), periods=pred_len, freq="B"
+            )
+            y_ts = pd.Series(future_dates)
+
+            # A. Base Prediction
+            p_base = predictor.predict(
+                h, x_ts, y_ts, pred_len=pred_len, news_emb=None, verbose=False
+            )
+            b_preds = p_base["close"].values[: len(actuals)]
+
+            # B. News-Aware Prediction
+            p_news = predictor.predict(
+                h,
+                x_ts,
+                y_ts,
+                pred_len=pred_len,
+                news_emb=item["news_emb"],
+                verbose=False,
+            )
+            n_preds = p_news["close"].values[: len(actuals)]
+
+            # Calculate MAE over the window
+            b_mae = np.mean(np.abs(b_preds - actuals))
+            n_mae = np.mean(np.abs(n_preds - actuals))
+
+            base_maes.append(b_mae)
+            news_maes.append(n_mae)
+
+            improvement = (b_mae - n_mae) / (b_mae + 1e-6) * 100
+
+            date_str = str(t["date"].values[0])[:10]
+            ticker = h.iloc[-1]["ticker"] if "ticker" in h.columns else "Stock"
+            print(
+                f"{date_str:<12} | {ticker:<8} | {b_mae:<15.4f} | {n_mae:<15.4f} | {improvement:>+7.1f}%"
+            )
+
+            # C. Generate Visualization for this case
+            try:
+                # Helper to convert DF to KLinePoints
+                def to_kp_list(preds_df):
+                    points = []
+                    for idx, row in preds_df.iterrows():
+                        points.append(
+                            KLinePoint(
+                                date=str(idx)[:10],
+                                open=row["open"],
+                                high=row["high"],
+                                low=row["low"],
+                                close=row["close"],
+                                volume=row["volume"] if "volume" in row else 0,
+                            )
+                        )
+                    return points
+
+                forecast_obj = ForecastResult(
+                    ticker=ticker,
+                    base_forecast=to_kp_list(p_base),
+                    adjusted_forecast=to_kp_list(p_news),
+                    rationale=item["summary"],
+                )
+
+                # Ground truth for visualizer expects a DataFrame with 'date' and 'close'
+                gt_df = t[["date", "open", "high", "low", "close", "volume"]]
+
+                chart = VisualizerTools.generate_stock_chart(
+                    df=h,
+                    ticker=ticker,
+                    title=f"Training Eval: {ticker} ({date_str}) Improvement: {improvement:.1f}%",
+                    forecast=forecast_obj,
+                    ground_truth=gt_df,
+                )
+
+                safe_date = date_str.replace("-", "")
+                filename = f"eval_{ticker}_{safe_date}.html"
+                VisualizerTools.render_chart_to_file(
+                    chart, os.path.join(SRC_DIR, f"exports/training_results/{filename}")
+                )
+            except Exception as e:
+                logger.error(f"Failed to generate eval chart for {ticker}: {e}")
+
+        # Summary Statistics
+        avg_base_err = sum(base_maes) / max(1, len(base_maes))
+        avg_news_err = sum(news_maes) / max(1, len(news_maes))
+        overall_imp = (avg_base_err - avg_news_err) / (avg_base_err + 1e-6) * 100
+
+        print("-" * 90)
+        print(
+            f"{'AVERAGE':<12} | {'-':<8} | {avg_base_err:<15.4f} | {avg_news_err:<15.4f} | {overall_imp:>+7.1f}%"
+        )
+        print("=" * 90 + "\n")
+
+        logger.success(
+            f"🏁 Statistical Analysis Complete. Avg Error Reduction ({pred_len}-day): {overall_imp:.2f}%"
+        )
+        logger.info(
+            f"📊 Visualization results saved to: {os.path.join(SRC_DIR, 'exports/training_results/')}"
+        )
+
+
+if __name__ == "__main__":
+    trainer = AutoSynthesisTrainer()
+
+    logger.info("📂 Fetching all stock codes from database...")
+    res = trainer.db.execute_query("SELECT code FROM stock_list")
+    all_tickers = [row["code"] for row in res]
+
+    if not all_tickers:
+        logger.warning("⚠️ No tickers found in stock_list table. Trying to sync...")
+        trainer.tools._check_and_update_stock_list(force=True)
+        res = trainer.db.execute_query("SELECT code FROM stock_list")
+        all_tickers = [row["code"] for row in res]
+
+    logger.info(f"🚀 Starting training on potential stocks (1-year scan)...")
+    # 为了演示，我们扫描前 100 个股票，寻找最近一年的冲击点
+    trainer.run_synthesis_and_train(all_tickers[:100], pred_len=1)