ValueError: Tensor Tensor("dense_2/Softmax:0", shape=(?, 8), dtype=float32) is not an element of thi

import tensorflow as tf import pandas as pd import numpy as np from sklearn.preprocessing import MinMaxScaler from tensorflow.keras.models import Model from tensorflow.keras.layers import Input, LSTM, Dense, Attention, Concatenate # 数据预处理函数 def preprocess_data(file_path, sequence_length=10): """读取并预处理双色球历史数据""" df = pd.read_csv(file_path ,encoding ="gbk") # 确保数据排序正确 df.sort_values('期号', inplace=True) # 提取红球和蓝球数据 red_balls = df[['红1', '红2', '红3', '红4', '红5', '红6']].values blue_ball = df['蓝球'].values.reshape(-1, 1) # 归一化处理 scaler_red = MinMaxScaler() scaler_blue = MinMaxScaler() red_scaled = scaler_red.fit_transform(red_balls) blue_scaled = scaler_blue.fit_transform(blue_ball) # 创建序列数据 X_red, X_blue, y = [], [], [] for i in range(len(red_scaled) - sequence_length): X_red.append(red_scaled[i:i+sequence_length]) X_blue.append(blue_scaled[i:i+sequence_length]) y.append(red_scaled[i+sequence_length]) # 预测下一期红球 return np.array(X_red), np.array(X_blue), np.array(y), scaler_red, scaler_blue # Genie 2世界模型架构 def build_genie2_model(red_seq_length, blue_seq_length, red_features, blue_features): """构建Genie 2世界模型""" # 红球输入分支 red_input = Input(shape=(red_seq_length, red_features)) red_lstm = LSTM(64, return_sequences=True)(red_input) red_lstm2 = LSTM(32)(red_lstm) # 蓝球输入分支 blue_input = Input(shape=(blue_seq_length, blue_features)) blue_lstm = LSTM(32)(blue_input) # 注意力机制融合 attention = Attention()([red_lstm2, blue_lstm]) combined = Concatenate()([red_lstm2, blue_lstm, attention]) # 世界模型核心 world_model = Dense(128, activation='relu')(combined) world_model = Dense(64, activation='relu')(world_model) # 输出层 - 预测6个红球 output = Dense(red_features, activation='sigmoid')(world_model) model = Model(inputs=[red_input, blue_input], outputs=output) model.compile(optimizer='adam', loss='mse') return model # 主执行流程 def main(): # 1. 数据准备 FILE_PATH = "D:\\worker\\lottery_results8.csv" X_red, X_blue, y, scaler_red, scaler_blue = preprocess_data(FILE_PATH, sequence_length=10) # 2. 构建模型 model = build_genie2_model( red_seq_length=X_red.shape[1], blue_seq_length=X_blue.shape[1], red_features=X_red.shape[2], blue_features=X_blue.shape[2] ) # 3. 训练模型 model.fit([X_red, X_blue], y, epochs=100, batch_size=32, validation_split=0.2) # 4. 预测最新一期 last_red = X_red[-1].reshape(1, X_red.shape[1], X_red.shape[2]) last_blue = X_blue[-1].reshape(1, X_blue.shape[1], X_blue.shape[2]) predicted_red = model.predict([last_red, last_blue]) # 5. 反归一化并输出预测结果 predicted_red_actual = scaler_red.inverse_transform(predicted_red) predicted_red_actual = np.round(predicted_red_actual).astype(int) # 蓝球预测简化版（实际应用中可扩展） blue_prediction = np.random.randint(1, 17) # 随机预测作为演示 print("\n预测结果:") print(f"红球: {predicted_red_actual[0]}") print(f"蓝球: {blue_prediction}") if __name__ == "__main__": main() ValueError: Exception encountered when calling Attention.call(). Dimension must be 2 but is 3 for '{{node functional_1/attention_1/transpose}} = Transpose[T=DT_FLOAT, Tperm=DT_INT32](functional_1/lstm_2_1/strided_slice_3, functional_1/attention_1/transpose/perm)' with input shapes: [?,32], [3]. Arguments received by Attention.call(): • inputs=['tf.Tensor(shape=(None, 32), dtype=float32)', 'tf.Tensor(shape=(None, 32), dtype=float32)'] • mask=['None', 'None'] • training=True • return_attention_scores=False • use_causal_mask=False 完善相关代码

--------------------------------------------------------------------------- ValueError Traceback (most recent call last) /tmp/ipykernel_424986/1460261549.py in <module> 52 # 训练主循环 53 for epoch in range(epochs): ---> 54 total_loss, evm_loss, spacing_loss = train_step() 55 56 # 温度衰减（后期接近硬采样） ~/miniconda3/lib/python3.8/site-packages/tensorflow/python/util/traceback_utils.py in error_handler(*args, **kwargs) 151 except Exception as e: 152 filtered_tb = _process_traceback_frames(e.__traceback__) --> 153 raise e.with_traceback(filtered_tb) from None 154 finally: 155 del filtered_tb /tmp/__autograph_generated_filekm9_7z0u.py in tf__train_step() 10 retval_ = ag__.UndefinedReturnValue() 11 with ag__.ld(tf).GradientTape() as tape: ---> 12 logits = ag__.converted_call(ag__.ld(encoder), (ag__.converted_call(ag__.ld(tf).expand_dims, (ag__.ld(snr_tf), (- 1)), None, fscope),), None, fscope) 13 (bits, _) = ag__.converted_call(ag__.ld(gumbel_softmax_sampler_tf), (ag__.ld(logits), ag__.ld(n), ag__.ld(bit_sym), ag__.ld(temp)), None, fscope) 14 x = ag__.converted_call(ag__.ld(trans_model), (ag__.ld(bits),), None, fscope) ~/miniconda3/lib/python3.8/site-packages/keras/utils/traceback_utils.py in error_handler(*args, **kwargs) 68 # To get the full stack trace, call: 69 # `tf.debugging.disable_traceback_filtering()` ---> 70 raise e.with_traceback(filtered_tb) from None 71 finally: 72 del filtered_tb /tmp/__autograph_generated_filevxxd8fu5.py in tf__call(self, snr) 9 do_return = False 10 retval_ = ag__.UndefinedReturnValue() ---> 11 x = ag__.converted_call(ag__.ld(self).dense1, (ag__.ld(snr),), None, fscope) 12 x = ag__.converted_call(ag__.ld(self).relu, (ag__.ld(x),), None, fscope) 13 try: ValueError: in user code: File "/tmp/ipykernel_424986/1460261549.py", line 20, in train_step * logits = encoder(tf.expand_dims(snr_tf, -1)) # [M] File "/root/miniconda3/lib/python3.8/site-packages/keras/utils/traceback_utils.py", line 70, in error_handler ** raise e.with_traceback(filtered_tb) from None File "/tmp/__autograph_generated_filevxxd8fu5.py", line 11, in tf__call x = ag__.converted_call(ag__.ld(self).dense1, (ag__.ld(snr),), None, fscope) ValueError: Exception encountered when calling layer 'encoder' (type Encoder). in user code: File "/tmp/ipykernel_424986/3358851902.py", line 16, in call * x = self.dense1(snr) File "/root/miniconda3/lib/python3.8/site-packages/keras/utils/traceback_utils.py", line 70, in error_handler ** raise e.with_traceback(filtered_tb) from None File "/root/miniconda3/lib/python3.8/site-packages/keras/engine/input_spec.py", line 250, in assert_input_compatibility raise ValueError( ValueError: Input 0 of layer "dense" is incompatible with the layer: expected min_ndim=2, found ndim=1. Full shape received: (1,) Call arguments received by layer 'encoder' (type Encoder): • snr=tf.Tensor(shape=(1,), dtype=float32)