Add daily growth logs for kindergarten class from September 25 to October 14, 2024, focusing on health and physical activities. Each log details daily themes, activities, observations, and teacher notes to track children's development in self-care, social skills, and motor coordination.

main_chat.py

@@ -31,11 +31,11 @@ from train_chat import clean_corpus
 def get_config(vocab_size: int) -> LanguageConfig:
     return LanguageConfig(
         vocab_size=vocab_size,
-        embed_dim=48, input_dim=24,
-        workspace_dim=96, expert_dim=48,
-        num_experts=10, num_wells=6,
-        ode_steps=3, dt=0.1, tau_w=0.5,
-        jacobian_sparsity=24, noise_std=0.002,
+        embed_dim=64, input_dim=32,
+        workspace_dim=128, expert_dim=64,
+        num_experts=12, num_wells=8,
+        ode_steps=4, dt=0.1, tau_w=0.6,
+        jacobian_sparsity=32, noise_std=0.001,
         use_rk4=True, use_layer_norm=True,
     )
 
@@ -43,20 +43,33 @@ def get_config(vocab_size: int) -> LanguageConfig:
 def load_or_init_model(device: str):
     """加载已保存的模型或初始化新模型"""
     text = clean_corpus()  # 清洗后语料（繁简统一+过滤）
-    tokenizer = CharTokenizer.from_text(text)  # 不截断，保留所有字符
-    config = get_config(tokenizer.vocab_size)
-    model = JSpaceLanguageModel(config).to(device)
 
     mp = Path('outputs/chat_model.pt')
     if mp.exists():
         try:
             ckpt = torch.load(mp, map_location=device, weights_only=False)
+            # 用保存的 tokenizer chars 确保一致
+            saved_chars = ckpt.get('tokenizer_chars', None)
+            if saved_chars:
+                tokenizer = CharTokenizer(
+                    chars=saved_chars,
+                    char_to_idx={c: i for i, c in enumerate(saved_chars)},
+                    idx_to_char={i: c for i, c in enumerate(saved_chars)},
+                )
+            else:
+                tokenizer = CharTokenizer.from_text(text)
+            config = get_config(tokenizer.vocab_size)
+            model = JSpaceLanguageModel(config).to(device)
             model.load_state_dict(ckpt['model'])
-            print(f"已加载模型: {mp}（上次保存的对话状态）")
-        except Exception:
-            print("模型加载失败，全新初始化")
-    else:
-        print("全新初始化（首次对话）")
+            print(f"已加载模型: {mp}（vocab={tokenizer.vocab_size}）")
+            return model, config, tokenizer, text
+        except Exception as e:
+            print(f"模型加载失败: {e}，全新初始化")
+
+    tokenizer = CharTokenizer.from_text(text)
+    config = get_config(tokenizer.vocab_size)
+    model = JSpaceLanguageModel(config).to(device)
+    print("全新初始化（首次对话）")
     return model, config, tokenizer, text