更新训练参数

AITrain/train_npc_dialogue_lora.py

@@ -98,28 +98,32 @@ def create_lora_config():
     return config
 
 def prepare_dataset(data_path, tokenizer):
-    """准备数据集"""
+    """准备数据集（增强健壮性）"""
     print(f"Loading dataset from: {data_path}")
     
     # 加载JSON数据
     with open(data_path, 'r', encoding='utf-8') as f:
         data = json.load(f)
     
-    print(f"Total samples: {len(data)}")
+    print(f"Total samples before filtering: {len(data)}")
     
     # 转换为Dataset格式
     dataset = Dataset.from_list(data)
     
     # 应用预处理函数
-    def tokenize_function(examples):
-        return process_func(examples, tokenizer)
-    
     tokenized_dataset = dataset.map(
-        tokenize_function,
+        lambda example: process_func(example, tokenizer),
         remove_columns=dataset.column_names,
-        batched=False
+        batched=False  # process_func expects single examples
     )
     
+    # 关键步骤：过滤掉预处理后变为空的样本
+    original_size = len(tokenized_dataset)
+    tokenized_dataset = tokenized_dataset.filter(lambda example: len(example.get("input_ids", [])) > 0)
+    filtered_size = len(tokenized_dataset)
+    
+    print(f"Total samples after filtering: {filtered_size} ({original_size - filtered_size} samples removed)")
+    
     return tokenized_dataset
 
 def train_lora_model(model_path, data_path, output_dir):
@@ -152,17 +156,17 @@ def train_lora_model(model_path, data_path, output_dir):
         logging_steps=10,
         num_train_epochs=3,  # 增加训练轮数以充分学习角色特征
         save_steps=50,
-        learning_rate=5e-5,  # 稍微提高学习率
+        learning_rate=2e-5,  # 降低学习率以增加稳定性
         warmup_ratio=0.1,
-        max_grad_norm=1.0,
+        max_grad_norm=1.0,   # 保持梯度裁剪
         save_on_each_node=True,
         gradient_checkpointing=True,
         gradient_checkpointing_kwargs={"use_reentrant": True},
         dataloader_pin_memory=False,  # 减少内存使用
         remove_unused_columns=False,
         report_to="none",
-        #bf16=True,
+        bf16=True,  # 显式启用bf16以匹配模型加载类型
-        #fp16=True,  # 使用混合精度训练
+        #fp16=False, # 确保fp16被禁用
         save_total_limit=3,  # 只保留最新的3个检查点
     )
     
@@ -271,7 +275,7 @@ def test_trained_model(model_path, lora_path):
 
 def main():
     # 配置路径
-    model_path = '/mnt/e/AI/Project02/AITrain/Qwen/Qwen3-8B-AWQ'  # 基础模型路径
+    model_path = '/mnt/g/Project02/AITrain/Qwen/Qwen3-8B-AWQ'  # 基础模型路径
     data_path = './npc_dialogue_dataset.json'  # 训练数据路径
     output_dir = './output/NPC_Dialogue_LoRA'  # 输出目录