You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@iotdb.apache.org by yc...@apache.org on 2025/03/05 10:54:40 UTC

(iotdb) branch ycy/AINodeTraining updated: adjust training function

This is an automated email from the ASF dual-hosted git repository.

ycycse pushed a commit to branch ycy/AINodeTraining
in repository https://gitbox.apache.org/repos/asf/iotdb.git


The following commit(s) were added to refs/heads/ycy/AINodeTraining by this push:
     new e8de8abb174 adjust training function
e8de8abb174 is described below

commit e8de8abb174dc33375d090f6ed57115721c9e899
Author: YangCaiyin <yc...@gmail.com>
AuthorDate: Tue Mar 4 18:27:52 2025 +0800

    adjust training function
---
 iotdb-core/ainode/.gitignore                       |   4 +-
 .../ainode/ai/ainode/manager/training_manager.py   | 191 -----
 .../ainode/ai/ainode/model/TimerXL/__init__.py     |   0
 .../ai/ainode/model/TimerXL/exp/exp_basic.py       |  26 -
 .../ai/ainode/model/TimerXL/layers/Attn_Bias.py    |  78 --
 .../ainode/model/TimerXL/layers/Attn_Projection.py | 106 ---
 .../ainode/ai/ainode/model/TimerXL/layers/Embed.py | 189 -----
 .../model/TimerXL/layers/SelfAttention_Family.py   | 142 ----
 .../model/TimerXL/layers/Transformer_EncDec.py     | 246 ------
 .../ai/ainode/model/TimerXL/layers/__init__.py     |   0
 .../ai/ainode/model/TimerXL/models/__init__.py     |   0
 .../ai/ainode/model/TimerXL/models/timer_xl.py     |  80 --
 .../ai/ainode/model/built_in_model_factory.py      | 924 ---------------------
 iotdb-core/ainode/ai/ainode/model/model_factory.py | 235 ------
 iotdb-core/ainode/ai/ainode/model/model_storage.py | 123 ---
 iotdb-core/ainode/ai/ainode/util/__init__.py       |  17 -
 iotdb-core/ainode/ai/ainode/util/masking.py        |  35 -
 iotdb-core/ainode/ai/ainode/util/metrics.py        |  44 -
 iotdb-core/ainode/ai/ainode/util/tools.py          |  96 ---
 iotdb-core/ainode/{ai => ainode}/__init__.py       |   0
 .../ainode/{ai/ainode => ainode/core}/__init__.py  |   0
 .../ainode/{ai/ainode => ainode/core}/client.py    |  24 +-
 .../ainode/{ai/ainode => ainode/core}/config.py    |  22 +-
 .../ainode/{ai/ainode => ainode/core}/constant.py  |   2 +-
 .../ainode/{ai/ainode => ainode/core}/exception.py |   2 +-
 .../ainode/{ai/ainode => ainode/core}/handler.py   |  18 +-
 .../ainode/{ai/ainode => ainode/core}/log.py       |   8 +-
 .../{ai/ainode => ainode/core}/manager/__init__.py |   0
 .../core}/manager/cluster_manager.py               |   4 +-
 .../core}/manager/inference_manager.py             |  14 +-
 .../core}/manager/model_manager.py                 |  16 +-
 .../core/manager/training_manager.py}              |  13 +-
 .../ainode/{ai/ainode => ainode/core}/script.py    |  16 +-
 .../ainode/{ai/ainode => ainode/core}/service.py   |   8 +-
 .../ainode/model => ainode/core/util}/__init__.py  |   0
 .../{ai/ainode => ainode/core}/util/decorator.py   |   0
 .../ainode/{ai/ainode => ainode/core}/util/lock.py |   0
 .../{ai/ainode => ainode/core}/util/serde.py       |   2 +-
 .../{ai/ainode => ainode/core}/util/status.py      |   6 +-
 iotdb-core/ainode/pom.xml                          |   8 +-
 iotdb-core/ainode/pyproject.toml                   |  18 +-
 .../ainode/resources/conf/iotdb-ainode.properties  |   6 +-
 iotdb-core/ainode/resources/sbin/remove-ainode.sh  |   4 +-
 .../iotdb/confignode/manager/ModelManager.java     |   8 +-
 .../iotdb/confignode/manager/ProcedureManager.java |   5 +-
 .../impl/model/CreateTrainingProcedure.java        |  11 +-
 .../iotdb/db/queryengine/plan/Coordinator.java     |   4 +-
 .../execution/config/TableConfigTaskVisitor.java   |   7 +
 .../config/executor/ClusterConfigTaskExecutor.java |   1 +
 .../plan/relational/sql/ast/AstVisitor.java        |   4 +
 .../plan/relational/sql/ast/ShowModels.java        |  55 ++
 .../plan/relational/sql/parser/AstBuilder.java     |  10 +
 .../db/relational/grammar/sql/RelationalSql.g4     |  15 +-
 .../src/main/thrift/confignode.thrift              |   5 +-
 54 files changed, 207 insertions(+), 2645 deletions(-)

diff --git a/iotdb-core/ainode/.gitignore b/iotdb-core/ainode/.gitignore
index 0dec9206348..ddf402ed3b3 100644
--- a/iotdb-core/ainode/.gitignore
+++ b/iotdb-core/ainode/.gitignore
@@ -1,8 +1,8 @@
 # generated by Thrift
-/ai/thrift/
+/ainode/thrift/
 
 # generated by maven
-/ai/conf/
+/ainode/conf/
 
 # .whl of ainode, generated by Poetry
 /dist/
diff --git a/iotdb-core/ainode/ai/ainode/manager/training_manager.py b/iotdb-core/ainode/ai/ainode/manager/training_manager.py
deleted file mode 100644
index c0e3f1966f1..00000000000
--- a/iotdb-core/ainode/ai/ainode/manager/training_manager.py
+++ /dev/null
@@ -1,191 +0,0 @@
-import argparse
-
-import numpy as np
-import torch
-import time
-from iotdb.table_session import TableSession, TableSessionConfig
-from torch.utils.data import Dataset, DataLoader
-
-from ai.thrift.ainode.ttypes import ITableSchema
-from client import ClientManager
-from config import AINodeDescriptor
-from constant import TSStatusCode
-from log import Logger
-from model.TimerXL.exp.exp_forecast import ExpForecast
-from util.status import get_status
-
-config = TableSessionConfig(
-    node_urls=["localhost:6667"],
-    username="root",
-    password="root",
-    time_zone="UTC+8",
-)
-
-SELECT_SERIES_FORMAT_SQL = "select distinct item_id from %s.%s"
-COUNT_SERIES_LENGTH_SQL = "select count(value) from %s.%s where item_id = '%s'"
-FETCH_SERIES_SQL = "select value from %s.%s where item_id = '%s' offset %s limit %s"
-SERIES_NAME = "%s.%s.%s"
-
-logger = Logger()
-
-class TrainingParameters:
-    def __init__(self):
-        self.seed = 2025
-
-        # forecasting task
-        self.seq_len = 672
-        self.input_token_len = 96
-        self.output_token_len = 96
-
-        # model define
-        self.dropout = 0.1
-        self.e_layers = 1
-        self.d_model = 512
-        self.n_heads = 8
-        self.d_ff = 2048
-        self.activation = 'relu'
-        self.covariate = False
-        self.node_num = 100
-        self.node_list = '23,37,40'
-        self.use_norm = False
-        self.nonautoregressive = False
-        self.output_attention = False
-        self.flash_attention = False
-
-        # adaptation
-        self.adaptation = False
-        self.pretrain_model_path = 'pretrain_model.pth'
-        self.subset_rand_ratio = 1
-
-        # optimization
-        self.num_workers = 10
-        self.itr = 1
-        self.train_epochs = 1
-        self.batch_size = 32
-        self.patience = 3
-        self.learning_rate = 0.0001
-        self.des = 'test'
-        self.loss = 'MSE'
-        self.lradj = 'type1'
-        self.cosine = False
-        self.tmax = 10
-        self.weight_decay = 0
-        self.valid_last = False
-        self.last_token = False
-
-        # GPU
-        self.device = 'cpu'
-        self.ddp = False
-        self.dp = False
-        self.devices = '0,1,2,3'
-
-        # LLM-based model
-        self.gpt_layers = 6
-        self.patch_size = 16
-        self.kernel_size = 25
-        self.stride = 8
-
-        self.context_length = self.seq_len + self.output_token_len
-
-class IoTDBDataset(Dataset):
-
-    def __init__(self, sorted_series:list, start_index:int, end_index:int, output_tokens:int, input_tokens:int, predict_length:int):
-        self.sorted_series = sorted_series
-        self.start_index = start_index
-        self.end_index = end_index
-        self.output_tokens = output_tokens
-        self.input_tokens = input_tokens
-        self.context_length = predict_length + output_tokens
-        self.predict_length = predict_length
-        self.iotdb_session = TableSession(config)
-
-    def __getitem__(self, index):
-        window_index = index + self.start_index
-
-        series_index = 0
-
-        while self.sorted_series[series_index][2] < window_index:
-            series_index += 1
-
-        if series_index != 0:
-            window_index -= self.sorted_series[series_index - 1][2]
-
-        # while self.sorted_series[series_index][1] - real_index < self.context_len:
-        #     series_index += 1
-        if window_index != 0:
-            window_index -= 1
-        series = self.sorted_series[series_index - 1][0]
-        schema = series.split(".")
-
-        start_time = time.time()
-        result = []
-        try:
-            with self.iotdb_session.execute_query_statement(FETCH_SERIES_SQL % (schema[0], schema[1], schema[2], window_index, self.context_length)) as query_result:
-                while query_result.has_next():
-                    result.append(query_result.next().get_fields()[0].get_double_value())
-        except Exception:
-            print("error")
-        result = torch.tensor(result)
-        x_mark = torch.zeros((result.shape[0], 1))
-        y_mark = torch.zeros((result.shape[0], 1))
-        logger.info("fetch data from iotdb cost %s", time.time() - start_time)
-        return result[0:self.predict_length].unsqueeze(-1), result[-self.output_tokens:].unsqueeze(-1), x_mark, y_mark
-
-    def __len__(self):
-        return self.end_index - self.start_index - 1
-
-
-def _fetch_schema(table_schema_list: list[ITableSchema], args:TrainingParameters) -> (int, list):
-    iotdb_session = TableSession(config)
-    series_to_length = {}
-    for table_schema in table_schema_list:
-        series_list = []
-        with iotdb_session.execute_query_statement(SELECT_SERIES_FORMAT_SQL % (table_schema.database, table_schema.tableName)) as show_devices_result:
-            while show_devices_result.has_next():
-                series_list.append(str(show_devices_result.next().get_fields()[0]))
-
-        for series in series_list:
-            with iotdb_session.execute_query_statement(COUNT_SERIES_LENGTH_SQL % (table_schema.database, table_schema.tableName, series)) as count_series_result:
-                length = int(str(count_series_result.next().get_fields()[0]))
-                series_to_length[SERIES_NAME % (table_schema.database, table_schema.tableName, series)] = length
-
-    iotdb_session.close()
-
-    sorted_series = sorted(series_to_length.items(), key=lambda x: x[1])
-    sorted_series_with_prefix_sum = []
-    current_sum = 0
-    window_count = 0
-    for seq_name, seq_length in sorted_series:
-        window_count = seq_length - args.context_length + 1
-        current_sum += window_count
-        sorted_series_with_prefix_sum.append((seq_name, window_count , current_sum))
-
-    return window_count, sorted_series_with_prefix_sum
-
-
-class TrainingManager:
-    def __init__(self):
-        pass
-
-    def create_training_task(self, model_id:str, model_type, table_list: list[ITableSchema], parameters: dict[str, str]):
-        args = get_args()
-        args.model = "timer_xl"
-        total_count, sorted_series = _fetch_schema(table_list, args)
-
-        training_dataset = IoTDBDataset(sorted_series, 0, total_count, args.input_token_len, args.output_token_len, args.seq_len)
-        training_dataloader = DataLoader(training_dataset, batch_size=args.batch_size, shuffle=True, num_workers=1)
-        exp = ExpForecast(args)
-        exp.train(training_dataloader, model_id)
-
-        logger.info(model_id + "finished training")
-        ClientManager().borrow_config_node_client().update_model_info(model_id, 3, "finished", [AINodeDescriptor().get_config().get_ainode_id()],args.seq_len,args.output_token_len)
-        return get_status(TSStatusCode.SUCCESS_STATUS)
-
-
-def get_args():
-    args = TrainingParameters()
-    fix_seed = args.seed
-    torch.manual_seed(fix_seed)
-    np.random.seed(fix_seed)
-
-    return args
diff --git a/iotdb-core/ainode/ai/ainode/model/TimerXL/__init__.py b/iotdb-core/ainode/ai/ainode/model/TimerXL/__init__.py
deleted file mode 100644
index e69de29bb2d..00000000000
diff --git a/iotdb-core/ainode/ai/ainode/model/TimerXL/exp/exp_basic.py b/iotdb-core/ainode/ai/ainode/model/TimerXL/exp/exp_basic.py
deleted file mode 100644
index 3929cf426a9..00000000000
--- a/iotdb-core/ainode/ai/ainode/model/TimerXL/exp/exp_basic.py
+++ /dev/null
@@ -1,26 +0,0 @@
-from ai.ainode.model.TimerXL.models import timer_xl
-
-
-class ExpBasic(object):
-    def __init__(self, args):
-        self.args = args
-        self.model_dict = {
-            # 'timer': timer,
-            'timer_xl': timer_xl,
-        }
-        self.model = self._build_model()
-
-    def _build_model(self):
-        raise NotImplementedError
-
-    def _get_data(self):
-        pass
-
-    def vali(self):
-        pass
-
-    def train(self):
-        pass
-
-    def test(self):
-        pass
diff --git a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Attn_Bias.py b/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Attn_Bias.py
deleted file mode 100644
index 4214e4ba7b0..00000000000
--- a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Attn_Bias.py
+++ /dev/null
@@ -1,78 +0,0 @@
-import abc
-import math
-import torch
-from einops import rearrange
-from torch import nn
-
-
-class AttentionBias(nn.Module, abc.ABC):
-    def __init__(self, dim: int, num_heads: int):
-        super().__init__()
-        assert num_heads > 0 and dim % num_heads == 0
-
-        self.num_heads = num_heads
-        self.head_dim = dim // num_heads
-
-    @abc.abstractmethod
-    def forward(self, query_id, kv_id): ...
-
-
-class BinaryAttentionBias(AttentionBias):
-    def __init__(self, dim: int, num_heads: int):
-        super().__init__(dim, num_heads)
-        self.emb = nn.Embedding(num_embeddings=2, embedding_dim=self.num_heads)
-
-    def forward(self, query_id, kv_id):
-        ind = torch.eq(query_id.unsqueeze(-1), kv_id.unsqueeze(-2))
-        weight = rearrange(
-            self.emb.weight, "two num_heads -> two num_heads 1 1")
-        bias = ~ind * weight[:1] + ind * weight[1:]
-        return bias
-
-
-def _relative_position_bucket(relative_position, bidirectional=True, num_buckets=32, max_distance=128):
-    relative_buckets = 0
-    if bidirectional:
-        num_buckets //= 2
-        relative_buckets += (relative_position >
-                             0).to(torch.long) * num_buckets
-        relative_position = torch.abs(relative_position)
-    else:
-        relative_position = - \
-            torch.min(relative_position, torch.zeros_like(relative_position))
-
-    max_exact = num_buckets // 2
-    is_small = relative_position < max_exact
-    relative_position_if_large = max_exact + (
-        torch.log(relative_position.float() / max_exact)
-        / math.log(max_distance / max_exact)
-        * (num_buckets - max_exact)
-    ).to(torch.long)
-    relative_position_if_large = torch.min(
-        relative_position_if_large, torch.full_like(
-            relative_position_if_large, num_buckets - 1)
-    )
-
-    relative_buckets += torch.where(is_small,
-                                    relative_position, relative_position_if_large)
-    return relative_buckets
-
-
-class T5AttentionBias(AttentionBias):
-    def __init__(self, dim: int, num_heads: int):
-        super().__init__(dim, num_heads)
-        self.num_buckets = 32
-        self.max_distance = 32
-        self.relative_attention_bias = nn.Embedding(self.num_buckets, 1)
-
-    def forward(self, n_vars, n_tokens):
-        context_position = torch.arange(n_tokens, dtype=torch.long,)[:, None]
-        memory_position = torch.arange(n_tokens, dtype=torch.long, )[None, :]
-        relative_position = memory_position - context_position
-        bucket = _relative_position_bucket(relative_position=relative_position, bidirectional=False,
-                                           num_buckets=self.num_buckets, max_distance=self.max_distance).to(self.relative_attention_bias.weight.device)
-        bias = self.relative_attention_bias(bucket).squeeze(-1)
-        bias = bias.reshape(1, 1, bias.shape[0], bias.shape[1])
-        mask1 = torch.ones((n_vars, n_vars), dtype=torch.bool).to(bias.device)
-        final_bias = torch.kron(mask1, bias)
-        return final_bias
diff --git a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Attn_Projection.py b/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Attn_Projection.py
deleted file mode 100644
index 0002744c907..00000000000
--- a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Attn_Projection.py
+++ /dev/null
@@ -1,106 +0,0 @@
-import abc
-import torch
-from functools import cached_property
-from einops import einsum, rearrange, repeat
-from torch import nn
-
-
-class Projection(nn.Module, abc.ABC):
-    def __init__(self, proj_width: int, num_heads: int, **kwargs):
-        super().__init__()
-        self.proj_width = proj_width
-        self.num_heads = num_heads
-
-    @abc.abstractmethod
-    def forward(self, x, seq_id): ...
-
-
-class RotaryProjection(Projection):
-    def __init__(self, *, proj_width: int, num_heads: int, max_len: int = 512, base: int = 10000):
-        super().__init__(proj_width, num_heads)
-        assert (
-            self.proj_width % 2 == 0
-        ), f"proj_width must be even, got {self.proj_width}"
-        self.register_buffer(
-            "theta",
-            1.0
-            / torch.pow(
-                base,
-                torch.arange(0, self.proj_width, 2, dtype=torch.float)
-                / self.proj_width,
-            ),
-            persistent=False,
-        )
-        self.register_buffer("cos", None, persistent=False)
-        self.register_buffer("sin", None, persistent=False)
-        self._init_freq(max_len=max_len)
-
-    def _init_freq(self, max_len: int):
-        if self.cos is None or self.cos.size(-2) < max_len:
-            position = torch.arange(
-                max_len, device=self.theta.device, dtype=self.theta.dtype
-            )
-            m_theta = einsum(position, self.theta,
-                             "length, width -> length width")
-            m_theta = repeat(m_theta, "length width -> length (width 2)")
-            self.register_buffer("cos", torch.cos(m_theta), persistent=False)
-            self.register_buffer("sin", torch.sin(m_theta), persistent=False)
-
-    @staticmethod
-    def _rotate(x):
-        x1, x2 = rearrange(x, "... (dim r) -> r ... dim", r=2)
-        return rearrange([-x2, x1], "r ... dim -> ... (dim r)", r=2)  # noqa
-
-    def forward(self, x, seq_id):
-        self._init_freq(max_len=seq_id.max() + 1)
-        rot_cos = self.cos[seq_id]
-        rot_sin = self.sin[seq_id]
-        return rot_cos * x + rot_sin * self._rotate(x)
-
-
-class QueryKeyProjection(nn.Module):
-    def __init__(self, dim: int, num_heads: int, proj_layer, kwargs=None, partial_factor=None):
-        super().__init__()
-        if partial_factor is not None:
-            assert (
-                0.0 <= partial_factor[0] < partial_factor[1] <= 1.0
-            ), f"got {partial_factor[0]}, {partial_factor[1]}"
-        assert num_heads > 0 and dim % num_heads == 0
-
-        self.head_dim = dim // num_heads
-        self.partial_factor = partial_factor
-        self.query_proj = proj_layer(
-            proj_width=self.proj_width,
-            num_heads=num_heads,
-            **(kwargs or {}),
-        )
-        self.key_proj = self.query_proj
-
-    @cached_property
-    def proj_width(self) -> int:
-        if self.partial_factor is None:
-            return self.head_dim
-        return int(self.head_dim * (self.partial_factor[1] - self.partial_factor[0]))
-
-    @cached_property
-    def split_sizes(self):
-        if self.partial_factor is None:
-            return 0, self.head_dim, 0
-        return (
-            int(self.partial_factor[0] * self.head_dim),
-            self.proj_width,
-            int((1.0 - self.partial_factor[1]) * self.head_dim),
-        )
-
-    def forward(self, query, key, query_id, kv_id):
-        if self.partial_factor is not None:
-            queries = list(query.split(self.split_sizes, dim=-1))
-            keys = list(key.split(self.split_sizes, dim=-1))
-            queries[1] = self.query_proj(queries[1], seq_id=query_id)
-            keys[1] = self.key_proj(keys[1], seq_id=kv_id)
-            query = torch.cat(queries, dim=-1)
-            key = torch.cat(keys, dim=-1)
-        else:
-            query = self.query_proj(query, seq_id=query_id)
-            key = self.key_proj(key, seq_id=kv_id)
-        return query, key
diff --git a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Embed.py b/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Embed.py
deleted file mode 100644
index 6b476be455f..00000000000
--- a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Embed.py
+++ /dev/null
@@ -1,189 +0,0 @@
-import math
-import torch
-import torch.nn as nn
-
-
-class PositionalEmbedding(nn.Module):
-    def __init__(self, d_model, max_len=6500):
-        super(PositionalEmbedding, self).__init__()
-        # Compute the positional encodings once in log space.
-        pe = torch.zeros(max_len, d_model).float()
-        pe.require_grad = False
-
-        position = torch.arange(0, max_len).float().unsqueeze(1)
-        div_term = (torch.arange(0, d_model, 2).float()
-                    * -(math.log(10000.0) / d_model)).exp()
-
-        pe[:, 0::2] = torch.sin(position * div_term)
-        pe[:, 1::2] = torch.cos(position * div_term)
-
-        pe = pe.unsqueeze(0)
-        self.register_buffer('pe', pe)
-
-    def forward(self, x):
-        return self.pe[:, :x.size(1)]
-
-
-class TokenEmbedding(nn.Module):
-    def __init__(self, c_in, d_model):
-        super(TokenEmbedding, self).__init__()
-        padding = 1 if torch.__version__ >= '1.5.0' else 2
-        self.tokenConv = nn.Conv1d(in_channels=c_in, out_channels=d_model,
-                                   kernel_size=3, padding=padding, padding_mode='circular', bias=False)
-        for m in self.modules():
-            if isinstance(m, nn.Conv1d):
-                nn.init.kaiming_normal_(
-                    m.weight, mode='fan_in', nonlinearity='leaky_relu')
-
-    def forward(self, x):
-        x = self.tokenConv(x.permute(0, 2, 1)).transpose(1, 2)
-        return x
-
-
-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, embed_type='fixed', freq='h'):
-        super(TemporalEmbedding, self).__init__()
-
-        minute_size = 4
-        hour_size = 24
-        weekday_size = 7
-        day_size = 32
-        month_size = 13
-
-        Embed = FixedEmbedding if embed_type == 'fixed' else nn.Embedding
-        if freq == 't':
-            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[:, :, 4]) if hasattr(
-            self, 'minute_embed') else 0.
-        hour_x = self.hour_embed(x[:, :, 3])
-        weekday_x = self.weekday_embed(x[:, :, 2])
-        day_x = self.day_embed(x[:, :, 1])
-        month_x = self.month_embed(x[:, :, 0])
-
-        return hour_x + weekday_x + day_x + month_x + minute_x
-
-
-class TimeFeatureEmbedding(nn.Module):
-    def __init__(self, d_model, embed_type='timeF', freq='h'):
-        super(TimeFeatureEmbedding, self).__init__()
-
-        freq_map = {'h': 4, 't': 5, 's': 6,
-                    'm': 1, 'a': 1, 'w': 2, 'd': 3, 'b': 3}
-        d_inp = freq_map[freq]
-        self.embed = nn.Linear(d_inp, d_model, bias=False)
-
-    def forward(self, x):
-        return self.embed(x)
-
-
-class DataEmbedding(nn.Module):
-    def __init__(self, c_in, d_model, embed_type='fixed', freq='h', dropout=0.1):
-        super(DataEmbedding, self).__init__()
-
-        self.value_embedding = TokenEmbedding(c_in=c_in, d_model=d_model)
-        self.position_embedding = PositionalEmbedding(d_model=d_model)
-        self.temporal_embedding = TemporalEmbedding(d_model=d_model, embed_type=embed_type,
-                                                    freq=freq) if embed_type != 'timeF' else TimeFeatureEmbedding(
-            d_model=d_model, embed_type=embed_type, freq=freq)
-        self.dropout = nn.Dropout(p=dropout)
-
-    def forward(self, x, x_mark):
-        if x_mark is None:
-            x = self.value_embedding(x) + self.position_embedding(x)
-        else:
-            x = self.value_embedding(
-                x) + self.temporal_embedding(x_mark) + self.position_embedding(x)
-        return self.dropout(x)
-
-
-class DataEmbedding_inverted(nn.Module):
-    def __init__(self, c_in, d_model, embed_type='fixed', freq='h', dropout=0.1):
-        super(DataEmbedding_inverted, self).__init__()
-        self.value_embedding = nn.Linear(c_in, d_model)
-        self.dropout = nn.Dropout(p=dropout)
-
-    def forward(self, x, x_mark):
-        x = x.permute(0, 2, 1)
-        # x: [Batch Variate Time]
-        if x_mark is None:
-            x = self.value_embedding(x)
-        else:
-            x = self.value_embedding(
-                torch.cat([x, x_mark.permute(0, 2, 1)], 1))
-        # x: [Batch Variate d_model]
-        return self.dropout(x)
-
-
-class DataEmbedding_wo_pos(nn.Module):
-    def __init__(self, c_in, d_model, embed_type='fixed', freq='h', dropout=0.1):
-        super(DataEmbedding_wo_pos, self).__init__()
-
-        self.value_embedding = TokenEmbedding(c_in=c_in, d_model=d_model)
-        self.position_embedding = PositionalEmbedding(d_model=d_model)
-        self.temporal_embedding = TemporalEmbedding(d_model=d_model, embed_type=embed_type,
-                                                    freq=freq) if embed_type != 'timeF' else TimeFeatureEmbedding(
-            d_model=d_model, embed_type=embed_type, freq=freq)
-        self.dropout = nn.Dropout(p=dropout)
-
-    def forward(self, x, x_mark):
-        if x_mark is None:
-            x = self.value_embedding(x)
-        else:
-            x = self.value_embedding(x) + self.temporal_embedding(x_mark)
-        return self.dropout(x)
-
-
-class PatchEmbedding(nn.Module):
-    def __init__(self, d_model, patch_len, stride, padding, dropout):
-        super(PatchEmbedding, self).__init__()
-        # Patching
-        self.patch_len = patch_len
-        self.stride = stride
-        self.padding_patch_layer = nn.ReplicationPad1d((0, padding))
-
-        # Backbone, Input encoding: projection of feature vectors onto a d-dim vector space
-        self.value_embedding = nn.Linear(patch_len, d_model, bias=False)
-
-        # Positional embedding
-        self.position_embedding = PositionalEmbedding(d_model)
-
-        # Residual dropout
-        self.dropout = nn.Dropout(dropout)
-
-    def forward(self, x):
-        # do patching
-        n_vars = x.shape[1]
-        x = self.padding_patch_layer(x)
-        x = x.unfold(dimension=-1, size=self.patch_len, step=self.stride)
-        x = torch.reshape(x, (x.shape[0] * x.shape[1], x.shape[2], x.shape[3]))
-        # Input encoding
-        x = self.value_embedding(x) + self.position_embedding(x)
-        return self.dropout(x), n_vars
diff --git a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/SelfAttention_Family.py b/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/SelfAttention_Family.py
deleted file mode 100644
index 43746baf79d..00000000000
--- a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/SelfAttention_Family.py
+++ /dev/null
@@ -1,142 +0,0 @@
-import torch
-import torch.nn as nn
-import numpy as np
-from math import sqrt
-from einops import repeat
-from ai.ainode.model.TimerXL.layers.Attn_Bias import BinaryAttentionBias
-from ai.ainode.model.TimerXL.layers.Attn_Projection import QueryKeyProjection, RotaryProjection
-from ai.ainode.util.masking import TriangularCausalMask, TimerMultivariateMask, TimerCovariateMask
-
-
-class FullAttention(nn.Module):
-    def __init__(self, mask_flag=True, scale=None, attention_dropout=0.1, output_attention=False):
-        super(FullAttention, self).__init__()
-        self.scale = scale
-        self.mask_flag = mask_flag
-        self.output_attention = output_attention
-        self.dropout = nn.Dropout(attention_dropout)
-
-    def forward(self, queries, keys, values, attn_mask, n_vars=None, n_tokens=None, tau=None, delta=None):
-        B, L, H, E = queries.shape
-        _, S, _, D = values.shape
-        scale = self.scale or 1. / sqrt(E)
-
-        scores = torch.einsum("blhe,bshe->bhls", queries, keys)
-
-        if self.mask_flag:
-            if attn_mask is None:
-                attn_mask = TriangularCausalMask(B, L, device=queries.device)
-
-            scores.masked_fill_(attn_mask.mask, -np.inf)
-
-        A = self.dropout(torch.softmax(scale * scores, dim=-1))
-        V = torch.einsum("bhls,bshd->blhd", A, values)
-
-        if self.output_attention:
-            return V.contiguous(), A
-        else:
-            return V.contiguous(), None
-
-
-class TimeAttention(nn.Module):
-    def __init__(self, mask_flag=True, scale=None, attention_dropout=0.1, output_attention=False, d_model=512, num_heads=8, max_len=100, covariate=False, flash_attention=False):
-        super(TimeAttention, self).__init__()
-        self.scale = scale
-        self.mask_flag = mask_flag
-        self.output_attention = output_attention
-        self.dropout = nn.Dropout(attention_dropout)
-        self.covariate = covariate
-        self.flash_attention = flash_attention
-        self.qk_proj = QueryKeyProjection(dim=d_model, num_heads=num_heads, proj_layer=RotaryProjection, kwargs=dict(max_len=max_len),
-                                          partial_factor=(0.0, 0.5),)
-        self.attn_bias = BinaryAttentionBias(dim=d_model, num_heads=num_heads)
-
-    def forward(self, queries, keys, values, attn_mask, n_vars, n_tokens, tau=None, delta=None):
-        B, L, H, E = queries.shape
-        _, S, _, D = values.shape
-
-        # [B, H, L, E]
-        queries = queries.permute(0, 2, 1, 3)
-        keys = keys.permute(0, 2, 1, 3)
-        if self.flash_attention:
-            values = values.permute(0, 2, 1, 3)
-
-        seq_id = torch.arange(n_tokens * n_vars)
-        seq_id = repeat(seq_id, 'n -> b h n', b=B, h=H)
-
-        queries, keys = self.qk_proj(
-            queries, keys, query_id=seq_id, kv_id=seq_id)
-
-        scale = self.scale or 1. / sqrt(E)
-
-        var_id = repeat(torch.arange(n_vars),
-                        'C -> (C n_tokens)', n_tokens=n_tokens)
-        var_id = repeat(var_id, 'L -> b h L', b=B, h=1).to(queries.device)
-
-        attn_bias = self.attn_bias(var_id, var_id)
-
-        if self.mask_flag:
-            if attn_mask is None:
-                if self.covariate:
-                    attn_mask = TimerCovariateMask(
-                        B, n_vars, n_tokens, device=queries.device)
-                else:
-                    attn_mask = TimerMultivariateMask(
-                        B, n_vars, n_tokens, device=queries.device)
-            attn_mask = attn_bias.masked_fill(attn_mask.mask, float("-inf"))
-        else:
-            attn_mask = attn_bias
-
-        if self.flash_attention:
-            V = torch.nn.functional.scaled_dot_product_attention(
-                queries, keys, values, attn_mask)
-        else:
-            scores = torch.einsum("bhle,bhse->bhls", queries, keys)
-            scores += attn_mask
-            
-            A = self.dropout(torch.softmax(scale * scores, dim=-1))
-            V = torch.einsum("bhls,bshd->blhd", A, values)
-
-        if self.output_attention:
-            return V.contiguous(), None
-        else:
-            return V.contiguous(), None
-
-
-class AttentionLayer(nn.Module):
-    def __init__(self, attention, d_model, n_heads, d_keys=None, d_values=None):
-        super(AttentionLayer, self).__init__()
-
-        d_keys = d_keys or (d_model // n_heads)
-        d_values = d_values or (d_model // n_heads)
-
-        self.inner_attention = attention
-        self.query_projection = nn.Linear(d_model, d_keys * n_heads)
-        self.key_projection = nn.Linear(d_model, d_keys * n_heads)
-        self.value_projection = nn.Linear(d_model, d_values * n_heads)
-        self.out_projection = nn.Linear(d_values * n_heads, d_model)
-        self.n_heads = n_heads
-
-    def forward(self, queries, keys, values, attn_mask, n_vars=None, n_tokens=None, tau=None, delta=None):
-        B, L, _ = queries.shape
-        _, S, _ = keys.shape
-        H = self.n_heads
-
-        queries = self.query_projection(queries).view(B, L, H, -1)
-        keys = self.key_projection(keys).view(B, S, H, -1)
-        values = self.value_projection(values).view(B, S, H, -1)
-
-        out, attn = self.inner_attention(
-            queries,
-            keys,
-            values,
-            attn_mask,
-            n_vars=n_vars,
-            n_tokens=n_tokens,
-            tau=tau,
-            delta=delta
-        )
-        out = out.view(B, L, -1)
-
-        return self.out_projection(out), attn
-
diff --git a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Transformer_EncDec.py b/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Transformer_EncDec.py
deleted file mode 100644
index 963b7984e42..00000000000
--- a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/Transformer_EncDec.py
+++ /dev/null
@@ -1,246 +0,0 @@
-import torch.nn as nn
-import torch.nn.functional as F
-
-
-class EncoderLayer(nn.Module):
-    def __init__(self, attention, d_model, d_ff=None, dropout=0.1, activation="relu"):
-        super(EncoderLayer, self).__init__()
-        d_ff = d_ff or 4 * d_model
-        self.attention = attention
-        self.conv1 = nn.Conv1d(in_channels=d_model,
-                               out_channels=d_ff, kernel_size=1)
-        self.conv2 = nn.Conv1d(
-            in_channels=d_ff, out_channels=d_model, kernel_size=1)
-        self.norm1 = nn.LayerNorm(d_model)
-        self.norm2 = nn.LayerNorm(d_model)
-        self.dropout = nn.Dropout(dropout)
-        self.activation = F.relu if activation == "relu" else F.gelu
-
-    def forward(self, x, attn_mask=None, tau=None, delta=None):
-        new_x, attn = self.attention(
-            x, x, x,
-            attn_mask=attn_mask,
-            tau=tau, delta=delta
-        )
-        x = x + self.dropout(new_x)
-
-        y = x = self.norm1(x)
-        y = self.dropout(self.activation(self.conv1(y.transpose(-1, 1))))
-        y = self.dropout(self.conv2(y).transpose(-1, 1))
-
-        return self.norm2(x + y), attn
-
-
-class DecoderLayer(nn.Module):
-    def __init__(self, self_attention, cross_attention, d_model, d_ff=None,
-                 dropout=0.1, activation="relu"):
-        super(DecoderLayer, self).__init__()
-        d_ff = d_ff or 4 * d_model
-        self.self_attention = self_attention
-        self.cross_attention = cross_attention
-        self.conv1 = nn.Conv1d(in_channels=d_model,
-                               out_channels=d_ff, kernel_size=1)
-        self.conv2 = nn.Conv1d(
-            in_channels=d_ff, out_channels=d_model, kernel_size=1)
-        self.norm1 = nn.LayerNorm(d_model)
-        self.norm2 = nn.LayerNorm(d_model)
-        self.norm3 = nn.LayerNorm(d_model)
-        self.dropout = nn.Dropout(dropout)
-        self.activation = F.relu if activation == "relu" else F.gelu
-
-    def forward(self, x, cross, x_mask=None, cross_mask=None, tau=None, delta=None):
-        x = x + self.dropout(self.self_attention(
-            x, x, x,
-            attn_mask=x_mask,
-            tau=tau, delta=None
-        )[0])
-        x = self.norm1(x)
-
-        x = x + self.dropout(self.cross_attention(
-            x, cross, cross,
-            attn_mask=cross_mask,
-            tau=tau, delta=delta
-        )[0])
-
-        y = x = self.norm2(x)
-        y = self.dropout(self.activation(self.conv1(y.transpose(-1, 1))))
-        y = self.dropout(self.conv2(y).transpose(-1, 1))
-
-        return self.norm3(x + y)
-
-
-class DecoderOnlyLayer(nn.Module):
-    def __init__(self, attention, d_model, d_ff=None, dropout=0.1, activation="relu"):
-        super(DecoderOnlyLayer, self).__init__()
-        d_ff = d_ff or 4 * d_model
-        self.attention = attention
-        self.conv1 = nn.Conv1d(in_channels=d_model,
-                               out_channels=d_ff, kernel_size=1)
-        self.conv2 = nn.Conv1d(
-            in_channels=d_ff, out_channels=d_model, kernel_size=1)
-        self.norm1 = nn.LayerNorm(d_model)
-        self.norm2 = nn.LayerNorm(d_model)
-        self.dropout = nn.Dropout(dropout)
-        self.activation = F.relu if activation == "relu" else F.gelu
-
-    def forward(self, x, attn_mask=None, tau=None, delta=None):
-        new_x, attn = self.attention(
-            x, x, x,
-            attn_mask=attn_mask,
-            tau=tau, delta=delta
-        )
-        x = x + self.dropout(new_x)
-
-        y = x = self.norm1(x)
-        y = self.dropout(self.activation(self.conv1(y.transpose(-1, 1))))
-        y = self.dropout(self.conv2(y).transpose(-1, 1))
-
-        return self.norm2(x + y), attn
-
-
-class TimerLayer(nn.Module):
-    def __init__(self, attention, d_model, d_ff=None, dropout=0.1, activation="relu"):
-        super(TimerLayer, self).__init__()
-        d_ff = d_ff or 4 * d_model
-        self.attention = attention
-        self.conv1 = nn.Conv1d(in_channels=d_model,
-                               out_channels=d_ff, kernel_size=1)
-        self.conv2 = nn.Conv1d(
-            in_channels=d_ff, out_channels=d_model, kernel_size=1)
-        self.norm1 = nn.LayerNorm(d_model)
-        self.norm2 = nn.LayerNorm(d_model)
-        self.dropout = nn.Dropout(dropout)
-        self.activation = F.relu if activation == "relu" else F.gelu
-
-    def forward(self, x, n_vars, n_tokens, attn_mask=None, tau=None, delta=None):
-        new_x, attn = self.attention(
-            x, x, x,
-            n_vars=n_vars,
-            n_tokens=n_tokens,
-            attn_mask=attn_mask,
-            tau=tau, delta=delta
-        )
-        x = x + self.dropout(new_x)
-
-        y = x = self.norm1(x)
-        y = self.dropout(self.activation(self.conv1(y.transpose(-1, 1))))
-        y = self.dropout(self.conv2(y).transpose(-1, 1))
-
-        return self.norm2(x + y), attn
-
-
-class Encoder(nn.Module):
-    def __init__(self, attn_layers, conv_layers=None, norm_layer=None):
-        super(Encoder, self).__init__()
-        self.attn_layers = nn.ModuleList(attn_layers)
-        self.conv_layers = nn.ModuleList(
-            conv_layers) if conv_layers is not None else None
-        self.norm = norm_layer
-
-    def forward(self, x, attn_mask=None, tau=None, delta=None):
-        # x [B, L, D]
-        attns = []
-        if self.conv_layers is not None:
-            for i, (attn_layer, conv_layer) in enumerate(zip(self.attn_layers, self.conv_layers)):
-                delta = delta if i == 0 else None
-                x, attn = attn_layer(
-                    x, attn_mask=attn_mask, tau=tau, delta=delta)
-                x = conv_layer(x)
-                attns.append(attn)
-            x, attn = self.attn_layers[-1](x, tau=tau, delta=None)
-            attns.append(attn)
-        else:
-            for attn_layer in self.attn_layers:
-                x, attn = attn_layer(
-                    x, attn_mask=attn_mask, tau=tau, delta=delta)
-                attns.append(attn)
-
-        if self.norm is not None:
-            x = self.norm(x)
-
-        return x, attns
-
-
-class Decoder(nn.Module):
-    def __init__(self, layers, norm_layer=None, projection=None):
-        super(Decoder, self).__init__()
-        self.layers = nn.ModuleList(layers)
-        self.norm = norm_layer
-        self.projection = projection
-
-    def forward(self, x, cross, x_mask=None, cross_mask=None, tau=None, delta=None):
-        for layer in self.layers:
-            x = layer(x, cross, x_mask=x_mask,
-                      cross_mask=cross_mask, tau=tau, delta=delta)
-
-        if self.norm is not None:
-            x = self.norm(x)
-
-        if self.projection is not None:
-            x = self.projection(x)
-        return x
-
-
-class DecoderOnly(nn.Module):
-    def __init__(self, attn_layers, conv_layers=None, norm_layer=None):
-        super(DecoderOnly, self).__init__()
-        self.attn_layers = nn.ModuleList(attn_layers)
-        self.conv_layers = nn.ModuleList(
-            conv_layers) if conv_layers is not None else None
-        self.norm = norm_layer
-
-    def forward(self, x, attn_mask=None, tau=None, delta=None):
-        # x [B, L, D]
-        attns = []
-        if self.conv_layers is not None:
-            for i, (attn_layer, conv_layer) in enumerate(zip(self.attn_layers, self.conv_layers)):
-                delta = delta if i == 0 else None
-                x, attn = attn_layer(
-                    x, attn_mask=attn_mask, tau=tau, delta=delta)
-                x = conv_layer(x)
-                attns.append(attn)
-            x, attn = self.attn_layers[-1](x, tau=tau, delta=None)
-            attns.append(attn)
-        else:
-            for attn_layer in self.attn_layers:
-                x, attn = attn_layer(
-                    x, attn_mask=attn_mask, tau=tau, delta=delta)
-                attns.append(attn)
-
-        if self.norm is not None:
-            x = self.norm(x)
-
-        return x, attns
-
-
-class TimerBlock(nn.Module):
-    def __init__(self, attn_layers, conv_layers=None, norm_layer=None):
-        super(TimerBlock, self).__init__()
-        self.attn_layers = nn.ModuleList(attn_layers)
-        self.conv_layers = nn.ModuleList(
-            conv_layers) if conv_layers is not None else None
-        self.norm = norm_layer
-
-    def forward(self, x, n_vars, n_tokens, attn_mask=None, tau=None, delta=None):
-        # x [B, L, D]
-        attns = []
-        if self.conv_layers is not None:
-            for i, (attn_layer, conv_layer) in enumerate(zip(self.attn_layers, self.conv_layers)):
-                delta = delta if i == 0 else None
-                x, attn = attn_layer(
-                    x, attn_mask=attn_mask, tau=tau, delta=delta)
-                x = conv_layer(x)
-                attns.append(attn)
-            x, attn = self.attn_layers[-1](x, n_vars,
-                                           n_tokens, tau=tau, delta=None)
-            attns.append(attn)
-        else:
-            for attn_layer in self.attn_layers:
-                x, attn = attn_layer(x, n_vars, n_tokens,
-                                     attn_mask=attn_mask, tau=tau, delta=delta)
-                attns.append(attn)
-
-        if self.norm is not None:
-            x = self.norm(x)
-
-        return x, attns
diff --git a/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/__init__.py b/iotdb-core/ainode/ai/ainode/model/TimerXL/layers/__init__.py
deleted file mode 100644
index e69de29bb2d..00000000000
diff --git a/iotdb-core/ainode/ai/ainode/model/TimerXL/models/__init__.py b/iotdb-core/ainode/ai/ainode/model/TimerXL/models/__init__.py
deleted file mode 100644
index e69de29bb2d..00000000000
diff --git a/iotdb-core/ainode/ai/ainode/model/TimerXL/models/timer_xl.py b/iotdb-core/ainode/ai/ainode/model/TimerXL/models/timer_xl.py
deleted file mode 100644
index 39a1f76b888..00000000000
--- a/iotdb-core/ainode/ai/ainode/model/TimerXL/models/timer_xl.py
+++ /dev/null
@@ -1,80 +0,0 @@
-import torch
-from torch import nn
-from ai.ainode.model.TimerXL.layers.Transformer_EncDec import TimerBlock, TimerLayer
-from ai.ainode.model.TimerXL.layers.SelfAttention_Family import AttentionLayer, TimeAttention
-
-
-class Model(nn.Module):
-    """
-    Timer-XL: Long-Context Transformers for Unified Time Series Forecasting 
-
-    Paper: https://arxiv.org/abs/2410.04803
-    
-    GitHub: https://github.com/thuml/Timer-XL
-    
-    Citation: @article{liu2024timer,
-        title={Timer-XL: Long-Context Transformers for Unified Time Series Forecasting},
-        author={Liu, Yong and Qin, Guo and Huang, Xiangdong and Wang, Jianmin and Long, Mingsheng},
-        journal={arXiv preprint arXiv:2410.04803},
-        year={2024}
-    }
-    """
-    def __init__(self, configs):
-        super().__init__()
-        self.input_token_len = configs.input_token_len
-        self.embedding = nn.Linear(self.input_token_len, configs.d_model)
-        self.output_attention = configs.output_attention
-        self.blocks = TimerBlock(
-            [
-                TimerLayer(
-                    AttentionLayer(
-                        TimeAttention(True, attention_dropout=configs.dropout,
-                                    output_attention=self.output_attention, 
-                                    d_model=configs.d_model, num_heads=configs.n_heads,
-                                    covariate=configs.covariate, flash_attention=configs.flash_attention),
-                                    configs.d_model, configs.n_heads),
-                    configs.d_model,
-                    configs.d_ff,
-                    dropout=configs.dropout,
-                    activation=configs.activation
-                ) for l in range(configs.e_layers)
-            ],
-            norm_layer=torch.nn.LayerNorm(configs.d_model)
-        )
-        self.head = nn.Linear(configs.d_model, configs.output_token_len)
-        self.use_norm = configs.use_norm
-
-    def forecast(self, x, x_mark, y_mark):
-        if self.use_norm:
-            means = x.mean(1, keepdim=True).detach()
-            x = x - means
-            stdev = torch.sqrt(
-                torch.var(x, dim=1, keepdim=True, unbiased=False) + 1e-5)
-            x /= stdev
-        B, _, C = x.shape
-        # [B, C, L]
-        x = x.permute(0, 2, 1)
-        # [B, C, N, P]
-        x = x.unfold(
-            dimension=-1, size=self.input_token_len, step=self.input_token_len)
-        N = x.shape[2]
-        # [B, C, N, D]
-        embed_out = self.embedding(x)
-        # [B, C * N, D]
-        embed_out = embed_out.reshape(B, C * N, -1)
-        embed_out, attns = self.blocks(embed_out, n_vars=C, n_tokens=N)
-        # [B, C * N, P]
-        dec_out = self.head(embed_out)
-        # [B, C, N * P]
-        dec_out = dec_out.reshape(B, C, N, -1).reshape(B, C, -1)
-        # [B, L, C]
-        dec_out = dec_out.permute(0, 2, 1)
-
-        if self.use_norm:
-            dec_out = dec_out * stdev + means
-        if self.output_attention:
-            return dec_out, attns
-        return dec_out
-
-    def forward(self, x, x_mark=None, y_mark=None):
-        return self.forecast(x, x_mark, y_mark)
diff --git a/iotdb-core/ainode/ai/ainode/model/built_in_model_factory.py b/iotdb-core/ainode/ai/ainode/model/built_in_model_factory.py
deleted file mode 100644
index 07e46d1932e..00000000000
--- a/iotdb-core/ainode/ai/ainode/model/built_in_model_factory.py
+++ /dev/null
@@ -1,924 +0,0 @@
-# Licensed to the Apache Software Foundation (ASF) under one
-# or more contributor license agreements.  See the NOTICE file
-# distributed with this work for additional information
-# regarding copyright ownership.  The ASF licenses this file
-# to you under the Apache License, Version 2.0 (the
-# "License"); you may not use this file except in compliance
-# with the License.  You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing,
-# software distributed under the License is distributed on an
-# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
-# KIND, either express or implied.  See the License for the
-# specific language governing permissions and limitations
-# under the License.
-#
-from abc import abstractmethod
-from typing import List, Dict
-
-import numpy as np
-from sklearn.preprocessing import MinMaxScaler
-from sktime.annotation.hmm_learn import GaussianHMM, GMMHMM
-from sktime.annotation.stray import STRAY
-from sktime.forecasting.arima import ARIMA
-from sktime.forecasting.exp_smoothing import ExponentialSmoothing
-from sktime.forecasting.naive import NaiveForecaster
-from sktime.forecasting.trend import STLForecaster
-
-from ai.ainode.constant import AttributeName, BuiltInModelType
-from ai.ainode.exception import InferenceModelInternalError, AttributeNotSupportError
-from ai.ainode.exception import WrongAttributeTypeError, NumericalRangeException, StringRangeException, \
-    ListRangeException, BuiltInModelNotSupportError
-from ai.ainode.log import Logger
-
-logger = Logger()
-
-
-def get_model_attributes(model_id: str):
-    if model_id == BuiltInModelType.ARIMA.value:
-        attribute_map = arima_attribute_map
-    elif model_id == BuiltInModelType.NAIVE_FORECASTER.value:
-        attribute_map = naive_forecaster_attribute_map
-    elif model_id == BuiltInModelType.EXPONENTIAL_SMOOTHING.value:
-        attribute_map = exponential_smoothing_attribute_map
-    elif model_id == BuiltInModelType.STL_FORECASTER.value:
-        attribute_map = stl_forecaster_attribute_map
-    elif model_id == BuiltInModelType.GMM_HMM.value:
-        attribute_map = gmmhmm_attribute_map
-    elif model_id == BuiltInModelType.GAUSSIAN_HMM.value:
-        attribute_map = gaussian_hmm_attribute_map
-    elif model_id == BuiltInModelType.STRAY.value:
-        attribute_map = stray_attribute_map
-    else:
-        raise BuiltInModelNotSupportError(model_id)
-    return attribute_map
-
-
-def fetch_built_in_model(model_id, inference_attributes):
-    """
-    Args:
-        model_id: the unique id of the model
-        inference_attributes: a list of attributes to be inferred, in this function, the attributes will include some
-            parameters of the built-in model. Some parameters are optional, and if the parameters are not
-            specified, the default value will be used.
-    Returns:
-        model: the built-in model
-        attributes: a dict of attributes, where the key is the attribute name, the value is the parsed value of the
-            attribute
-    Description:
-        the create_built_in_model function will create the built-in model, which does not require user
-        registration. This module will parse the inference attributes and create the built-in model.
-    """
-    attribute_map = get_model_attributes(model_id)
-
-    # validate the inference attributes
-    for attribute_name in inference_attributes:
-        if attribute_name not in attribute_map:
-            raise AttributeNotSupportError(model_id, attribute_name)
-
-    # parse the inference attributes, attributes is a Dict[str, Any]
-    attributes = parse_attribute(inference_attributes, attribute_map)
-
-    # build the built-in model
-    if model_id == BuiltInModelType.ARIMA.value:
-        model = ArimaModel(attributes)
-    elif model_id == BuiltInModelType.EXPONENTIAL_SMOOTHING.value:
-        model = ExponentialSmoothingModel(attributes)
-    elif model_id == BuiltInModelType.NAIVE_FORECASTER.value:
-        model = NaiveForecasterModel(attributes)
-    elif model_id == BuiltInModelType.STL_FORECASTER.value:
-        model = STLForecasterModel(attributes)
-    elif model_id == BuiltInModelType.GMM_HMM.value:
-        model = GMMHMMModel(attributes)
-    elif model_id == BuiltInModelType.GAUSSIAN_HMM.value:
-        model = GaussianHmmModel(attributes)
-    elif model_id == BuiltInModelType.STRAY.value:
-        model = STRAYModel(attributes)
-    else:
-        raise BuiltInModelNotSupportError(model_id)
-
-    return model
-
-
-class Attribute(object):
-    def __init__(self, name: str):
-        """
-        Args:
-            name: the name of the attribute
-        """
-        self._name = name
-
-    @abstractmethod
-    def get_default_value(self):
-        raise NotImplementedError
-
-    @abstractmethod
-    def validate_value(self, value):
-        raise NotImplementedError
-
-    @abstractmethod
-    def parse(self, string_value: str):
-        raise NotImplementedError
-
-
-class IntAttribute(Attribute):
-    def __init__(self, name: str,
-                 default_value: int,
-                 default_low: int,
-                 default_high: int,
-                 ):
-        super(IntAttribute, self).__init__(name)
-        self.__default_value = default_value
-        self.__default_low = default_low
-        self.__default_high = default_high
-
-    def get_default_value(self):
-        return self.__default_value
-
-    def validate_value(self, value):
-        if self.__default_low <= value <= self.__default_high:
-            return True
-        raise NumericalRangeException(self._name, value, self.__default_low, self.__default_high)
-
-    def parse(self, string_value: str):
-        try:
-            int_value = int(string_value)
-        except:
-            raise WrongAttributeTypeError(self._name, "int")
-        return int_value
-
-
-class FloatAttribute(Attribute):
-    def __init__(self, name: str,
-                 default_value: float,
-                 default_low: float,
-                 default_high: float,
-                 ):
-        super(FloatAttribute, self).__init__(name)
-        self.__default_value = default_value
-        self.__default_low = default_low
-        self.__default_high = default_high
-
-    def get_default_value(self):
-        return self.__default_value
-
-    def validate_value(self, value):
-        if self.__default_low <= value <= self.__default_high:
-            return True
-        raise NumericalRangeException(self._name, value, self.__default_low, self.__default_high)
-
-    def parse(self, string_value: str):
-        try:
-            float_value = float(string_value)
-        except:
-            raise WrongAttributeTypeError(self._name, "float")
-        return float_value
-
-
-class StringAttribute(Attribute):
-    def __init__(self, name: str, default_value: str, value_choices: List[str]):
-        super(StringAttribute, self).__init__(name)
-        self.__default_value = default_value
-        self.__value_choices = value_choices
-
-    def get_default_value(self):
-        return self.__default_value
-
-    def validate_value(self, value):
-        if value in self.__value_choices:
-            return True
-        raise StringRangeException(self._name, value, self.__value_choices)
-
-    def parse(self, string_value: str):
-        return string_value
-
-
-class BooleanAttribute(Attribute):
-    def __init__(self, name: str, default_value: bool):
-        super(BooleanAttribute, self).__init__(name)
-        self.__default_value = default_value
-
-    def get_default_value(self):
-        return self.__default_value
-
-    def validate_value(self, value):
-        if isinstance(value, bool):
-            return True
-        raise WrongAttributeTypeError(self._name, "bool")
-
-    def parse(self, string_value: str):
-        if string_value.lower() == "true":
-            return True
-        elif string_value.lower() == "false":
-            return False
-        else:
-            raise WrongAttributeTypeError(self._name, "bool")
-
-
-class ListAttribute(Attribute):
-    def __init__(self, name: str, default_value: List, value_type):
-        """
-        value_type is the type of the elements in the list, e.g. int, float, str
-        """
-        super(ListAttribute, self).__init__(name)
-        self.__default_value = default_value
-        self.__value_type = value_type
-        self.__type_to_str = {str: "str", int: "int", float: "float"}
-
-    def get_default_value(self):
-        return self.__default_value
-
-    def validate_value(self, value):
-        if not isinstance(value, list):
-            raise WrongAttributeTypeError(self._name, "list")
-        for value_item in value:
-            if not isinstance(value_item, self.__value_type):
-                raise WrongAttributeTypeError(self._name, self.__value_type)
-        return True
-
-    def parse(self, string_value: str):
-        try:
-            list_value = eval(string_value)
-        except:
-            raise WrongAttributeTypeError(self._name, "list")
-        if not isinstance(list_value, list):
-            raise WrongAttributeTypeError(self._name, "list")
-        for i in range(len(list_value)):
-            try:
-                list_value[i] = self.__value_type(list_value[i])
-            except:
-                raise ListRangeException(self._name, list_value, self.__type_to_str[self.__value_type])
-        return list_value
-
-
-class TupleAttribute(Attribute):
-    def __init__(self, name: str, default_value: tuple, value_type):
-        """
-        value_type is the type of the elements in the list, e.g. int, float, str
-        """
-        super(TupleAttribute, self).__init__(name)
-        self.__default_value = default_value
-        self.__value_type = value_type
-        self.__type_to_str = {str: "str", int: "int", float: "float"}
-
-    def get_default_value(self):
-        return self.__default_value
-
-    def validate_value(self, value):
-        if not isinstance(value, tuple):
-            raise WrongAttributeTypeError(self._name, "tuple")
-        for value_item in value:
-            if not isinstance(value_item, self.__value_type):
-                raise WrongAttributeTypeError(self._name, self.__value_type)
-        return True
-
-    def parse(self, string_value: str):
-        try:
-            tuple_value = eval(string_value)
-        except:
-            raise WrongAttributeTypeError(self._name, "tuple")
-        if not isinstance(tuple_value, tuple):
-            raise WrongAttributeTypeError(self._name, "tuple")
-        list_value = list(tuple_value)
-        for i in range(len(list_value)):
-            try:
-                list_value[i] = self.__value_type(list_value[i])
-            except:
-                raise ListRangeException(self._name, list_value, self.__type_to_str[self.__value_type])
-        tuple_value = tuple(list_value)
-        return tuple_value
-
-
-def parse_attribute(input_attributes: Dict[str, str], attribute_map: Dict[str, Attribute]):
-    """
-    Args:
-        input_attributes: a dict of attributes, where the key is the attribute name, the value is the string value of
-            the attribute
-        attribute_map: a dict of hyperparameters, where the key is the attribute name, the value is the Attribute
-            object
-    Returns:
-        a dict of attributes, where the key is the attribute name, the value is the parsed value of the attribute
-    """
-    attributes = {}
-    for attribute_name in attribute_map:
-        # user specified the attribute
-        if attribute_name in input_attributes:
-            attribute = attribute_map[attribute_name]
-            value = attribute.parse(input_attributes[attribute_name])
-            attribute.validate_value(value)
-            attributes[attribute_name] = value
-        # user did not specify the attribute, use the default value
-        else:
-            try:
-                attributes[attribute_name] = attribute_map[attribute_name].get_default_value()
-            except NotImplementedError as e:
-                logger.error(f"attribute {attribute_name} is not implemented.")
-                raise e
-    return attributes
-
-
-# built-in sktime model attributes
-# NaiveForecaster
-naive_forecaster_attribute_map = {
-    AttributeName.PREDICT_LENGTH.value: IntAttribute(
-        name=AttributeName.PREDICT_LENGTH.value,
-        default_value=1,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.STRATEGY.value: StringAttribute(
-        name=AttributeName.STRATEGY.value,
-        default_value="last",
-        value_choices=["last", "mean"],
-    ),
-    AttributeName.SP.value: IntAttribute(
-        name=AttributeName.SP.value,
-        default_value=1,
-        default_low=1,
-        default_high=5000
-    ),
-}
-# ExponentialSmoothing
-exponential_smoothing_attribute_map = {
-    AttributeName.PREDICT_LENGTH.value: IntAttribute(
-        name=AttributeName.PREDICT_LENGTH.value,
-        default_value=1,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.DAMPED_TREND.value: BooleanAttribute(
-        name=AttributeName.DAMPED_TREND.value,
-        default_value=False,
-    ),
-    AttributeName.INITIALIZATION_METHOD.value: StringAttribute(
-        name=AttributeName.INITIALIZATION_METHOD.value,
-        default_value="estimated",
-        value_choices=["estimated", "heuristic", "legacy-heuristic", "known"],
-    ),
-    AttributeName.OPTIMIZED.value: BooleanAttribute(
-        name=AttributeName.OPTIMIZED.value,
-        default_value=True,
-    ),
-    AttributeName.REMOVE_BIAS.value: BooleanAttribute(
-        name=AttributeName.REMOVE_BIAS.value,
-        default_value=False,
-    ),
-    AttributeName.USE_BRUTE.value: BooleanAttribute(
-        name=AttributeName.USE_BRUTE.value,
-        default_value=False,
-    )
-}
-# Arima
-arima_attribute_map = {
-    AttributeName.PREDICT_LENGTH.value: IntAttribute(
-        name=AttributeName.PREDICT_LENGTH.value,
-        default_value=1,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.ORDER.value: TupleAttribute(
-        name=AttributeName.ORDER.value,
-        default_value=(1, 0, 0),
-        value_type=int
-    ),
-    AttributeName.SEASONAL_ORDER.value: TupleAttribute(
-        name=AttributeName.SEASONAL_ORDER.value,
-        default_value=(0, 0, 0, 0),
-        value_type=int
-    ),
-    AttributeName.METHOD.value: StringAttribute(
-        name=AttributeName.METHOD.value,
-        default_value="lbfgs",
-        value_choices=["lbfgs", "bfgs", "newton", "nm", "cg", "ncg", "powell"],
-    ),
-    AttributeName.MAXITER.value: IntAttribute(
-        name=AttributeName.MAXITER.value,
-        default_value=1,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.SUPPRESS_WARNINGS.value: BooleanAttribute(
-        name=AttributeName.SUPPRESS_WARNINGS.value,
-        default_value=True,
-    ),
-    AttributeName.OUT_OF_SAMPLE_SIZE.value: IntAttribute(
-        name=AttributeName.OUT_OF_SAMPLE_SIZE.value,
-        default_value=0,
-        default_low=0,
-        default_high=5000
-    ),
-    AttributeName.SCORING.value: StringAttribute(
-        name=AttributeName.SCORING.value,
-        default_value="mse",
-        value_choices=["mse", "mae", "rmse", "mape", "smape", "rmsle", "r2"],
-    ),
-    AttributeName.WITH_INTERCEPT.value: BooleanAttribute(
-        name=AttributeName.WITH_INTERCEPT.value,
-        default_value=True,
-    ),
-    AttributeName.TIME_VARYING_REGRESSION.value: BooleanAttribute(
-        name=AttributeName.TIME_VARYING_REGRESSION.value,
-        default_value=False,
-    ),
-    AttributeName.ENFORCE_STATIONARITY.value: BooleanAttribute(
-        name=AttributeName.ENFORCE_STATIONARITY.value,
-        default_value=True,
-    ),
-    AttributeName.ENFORCE_INVERTIBILITY.value: BooleanAttribute(
-        name=AttributeName.ENFORCE_INVERTIBILITY.value,
-        default_value=True,
-    ),
-    AttributeName.SIMPLE_DIFFERENCING.value: BooleanAttribute(
-        name=AttributeName.SIMPLE_DIFFERENCING.value,
-        default_value=False,
-    ),
-    AttributeName.MEASUREMENT_ERROR.value: BooleanAttribute(
-        name=AttributeName.MEASUREMENT_ERROR.value,
-        default_value=False,
-    ),
-    AttributeName.MLE_REGRESSION.value: BooleanAttribute(
-        name=AttributeName.MLE_REGRESSION.value,
-        default_value=True,
-    ),
-    AttributeName.HAMILTON_REPRESENTATION.value: BooleanAttribute(
-        name=AttributeName.HAMILTON_REPRESENTATION.value,
-        default_value=False,
-    ),
-    AttributeName.CONCENTRATE_SCALE.value: BooleanAttribute(
-        name=AttributeName.CONCENTRATE_SCALE.value,
-        default_value=False,
-    )
-}
-# STLForecaster
-stl_forecaster_attribute_map = {
-    AttributeName.PREDICT_LENGTH.value: IntAttribute(
-        name=AttributeName.PREDICT_LENGTH.value,
-        default_value=1,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.SP.value: IntAttribute(
-        name=AttributeName.SP.value,
-        default_value=2,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.SEASONAL.value: IntAttribute(
-        name=AttributeName.SEASONAL.value,
-        default_value=7,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.SEASONAL_DEG.value: IntAttribute(
-        name=AttributeName.SEASONAL_DEG.value,
-        default_value=1,
-        default_low=0,
-        default_high=5000
-    ),
-    AttributeName.TREND_DEG.value: IntAttribute(
-        name=AttributeName.TREND_DEG.value,
-        default_value=1,
-        default_low=0,
-        default_high=5000
-    ),
-    AttributeName.LOW_PASS_DEG.value: IntAttribute(
-        name=AttributeName.LOW_PASS_DEG.value,
-        default_value=1,
-        default_low=0,
-        default_high=5000
-    ),
-    AttributeName.SEASONAL_JUMP.value: IntAttribute(
-        name=AttributeName.SEASONAL_JUMP.value,
-        default_value=1,
-        default_low=0,
-        default_high=5000
-    ),
-    AttributeName.TREND_JUMP.value: IntAttribute(
-        name=AttributeName.TREND_JUMP.value,
-        default_value=1,
-        default_low=0,
-        default_high=5000
-    ),
-    AttributeName.LOSS_PASS_JUMP.value: IntAttribute(
-        name=AttributeName.LOSS_PASS_JUMP.value,
-        default_value=1,
-        default_low=0,
-        default_high=5000
-    ),
-}
-
-# GAUSSIAN_HMM
-gaussian_hmm_attribute_map = {
-    AttributeName.N_COMPONENTS.value: IntAttribute(
-        name=AttributeName.N_COMPONENTS.value,
-        default_value=1,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.COVARIANCE_TYPE.value: StringAttribute(
-        name=AttributeName.COVARIANCE_TYPE.value,
-        default_value="diag",
-        value_choices=["spherical", "diag", "full", "tied"],
-    ),
-    AttributeName.MIN_COVAR.value: FloatAttribute(
-        name=AttributeName.MIN_COVAR.value,
-        default_value=1e-3,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.STARTPROB_PRIOR.value: FloatAttribute(
-        name=AttributeName.STARTPROB_PRIOR.value,
-        default_value=1.0,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.TRANSMAT_PRIOR.value: FloatAttribute(
-        name=AttributeName.TRANSMAT_PRIOR.value,
-        default_value=1.0,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.MEANS_PRIOR.value: FloatAttribute(
-        name=AttributeName.MEANS_PRIOR.value,
-        default_value=0.0,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.MEANS_WEIGHT.value: FloatAttribute(
-        name=AttributeName.MEANS_WEIGHT.value,
-        default_value=0.0,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.COVARS_PRIOR.value: FloatAttribute(
-        name=AttributeName.COVARS_PRIOR.value,
-        default_value=1e-2,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.COVARS_WEIGHT.value: FloatAttribute(
-        name=AttributeName.COVARS_WEIGHT.value,
-        default_value=1.0,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.ALGORITHM.value: StringAttribute(
-        name=AttributeName.ALGORITHM.value,
-        default_value="viterbi",
-        value_choices=["viterbi", "map"],
-    ),
-    AttributeName.N_ITER.value: IntAttribute(
-        name=AttributeName.N_ITER.value,
-        default_value=10,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.TOL.value: FloatAttribute(
-        name=AttributeName.TOL.value,
-        default_value=1e-2,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.PARAMS.value: StringAttribute(
-        name=AttributeName.PARAMS.value,
-        default_value="stmc",
-        value_choices=["stmc", "stm"],
-    ),
-    AttributeName.INIT_PARAMS.value: StringAttribute(
-        name=AttributeName.INIT_PARAMS.value,
-        default_value="stmc",
-        value_choices=["stmc", "stm"],
-    ),
-    AttributeName.IMPLEMENTATION.value: StringAttribute(
-        name=AttributeName.IMPLEMENTATION.value,
-        default_value="log",
-        value_choices=["log", "scaling"],
-    )
-}
-
-# GMMHMM
-gmmhmm_attribute_map = {
-    AttributeName.N_COMPONENTS.value: IntAttribute(
-        name=AttributeName.N_COMPONENTS.value,
-        default_value=1,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.N_MIX.value: IntAttribute(
-        name=AttributeName.N_MIX.value,
-        default_value=1,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.MIN_COVAR.value: FloatAttribute(
-        name=AttributeName.MIN_COVAR.value,
-        default_value=1e-3,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.STARTPROB_PRIOR.value: FloatAttribute(
-        name=AttributeName.STARTPROB_PRIOR.value,
-        default_value=1.0,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.TRANSMAT_PRIOR.value: FloatAttribute(
-        name=AttributeName.TRANSMAT_PRIOR.value,
-        default_value=1.0,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.WEIGHTS_PRIOR.value: FloatAttribute(
-        name=AttributeName.WEIGHTS_PRIOR.value,
-        default_value=1.0,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.MEANS_PRIOR.value: FloatAttribute(
-        name=AttributeName.MEANS_PRIOR.value,
-        default_value=0.0,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.MEANS_WEIGHT.value: FloatAttribute(
-        name=AttributeName.MEANS_WEIGHT.value,
-        default_value=0.0,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.ALGORITHM.value: StringAttribute(
-        name=AttributeName.ALGORITHM.value,
-        default_value="viterbi",
-        value_choices=["viterbi", "map"],
-    ),
-    AttributeName.COVARIANCE_TYPE.value: StringAttribute(
-        name=AttributeName.COVARIANCE_TYPE.value,
-        default_value="diag",
-        value_choices=["sperical", "diag", "full", "tied"],
-    ),
-    AttributeName.N_ITER.value: IntAttribute(
-        name=AttributeName.N_ITER.value,
-        default_value=10,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.TOL.value: FloatAttribute(
-        name=AttributeName.TOL.value,
-        default_value=1e-2,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.INIT_PARAMS.value: StringAttribute(
-        name=AttributeName.INIT_PARAMS.value,
-        default_value="stmcw",
-        value_choices=["s", "t", "m", "c", "w", "st", "sm", "sc", "sw", "tm", "tc", "tw", "mc", "mw", "cw", "stm",
-                       "stc", "stw", "smc", "smw", "scw", "tmc", "tmw", "tcw", "mcw", "stmc", "stmw", "stcw", "smcw",
-                       "tmcw", "stmcw"]
-    ),
-    AttributeName.PARAMS.value: StringAttribute(
-        name=AttributeName.PARAMS.value,
-        default_value="stmcw",
-        value_choices=["s", "t", "m", "c", "w", "st", "sm", "sc", "sw", "tm", "tc", "tw", "mc", "mw", "cw", "stm",
-                       "stc", "stw", "smc", "smw", "scw", "tmc", "tmw", "tcw", "mcw", "stmc", "stmw", "stcw", "smcw",
-                       "tmcw", "stmcw"]
-    ),
-    AttributeName.IMPLEMENTATION.value: StringAttribute(
-        name=AttributeName.IMPLEMENTATION.value,
-        default_value="log",
-        value_choices=["log", "scaling"],
-    )
-}
-
-# STRAY
-stray_attribute_map = {
-    AttributeName.ALPHA.value: FloatAttribute(
-        name=AttributeName.ALPHA.value,
-        default_value=0.01,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.K.value: IntAttribute(
-        name=AttributeName.K.value,
-        default_value=10,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.KNN_ALGORITHM.value: StringAttribute(
-        name=AttributeName.KNN_ALGORITHM.value,
-        default_value="brute",
-        value_choices=["brute", "kd_tree", "ball_tree", "auto"],
-    ),
-    AttributeName.P.value: FloatAttribute(
-        name=AttributeName.P.value,
-        default_value=0.5,
-        default_low=-1e10,
-        default_high=1e10,
-    ),
-    AttributeName.SIZE_THRESHOLD.value: IntAttribute(
-        name=AttributeName.SIZE_THRESHOLD.value,
-        default_value=50,
-        default_low=1,
-        default_high=5000
-    ),
-    AttributeName.OUTLIER_TAIL.value: StringAttribute(
-        name=AttributeName.OUTLIER_TAIL.value,
-        default_value="max",
-        value_choices=["min", "max"],
-    ),
-}
-
-
-class BuiltInModel(object):
-    def __init__(self, attributes):
-        self._attributes = attributes
-        self._model = None
-
-    @abstractmethod
-    def inference(self, data):
-        raise NotImplementedError
-
-
-class ArimaModel(BuiltInModel):
-    def __init__(self, attributes):
-        super(ArimaModel, self).__init__(attributes)
-        self._model = ARIMA(
-            order=attributes['order'],
-            seasonal_order=attributes['seasonal_order'],
-            method=attributes['method'],
-            suppress_warnings=attributes['suppress_warnings'],
-            maxiter=attributes['maxiter'],
-            out_of_sample_size=attributes['out_of_sample_size'],
-            scoring=attributes['scoring'],
-            with_intercept=attributes['with_intercept'],
-            time_varying_regression=attributes['time_varying_regression'],
-            enforce_stationarity=attributes['enforce_stationarity'],
-            enforce_invertibility=attributes['enforce_invertibility'],
-            simple_differencing=attributes['simple_differencing'],
-            measurement_error=attributes['measurement_error'],
-            mle_regression=attributes['mle_regression'],
-            hamilton_representation=attributes['hamilton_representation'],
-            concentrate_scale=attributes['concentrate_scale']
-        )
-
-    def inference(self, data):
-        try:
-            predict_length = self._attributes['predict_length']
-            self._model.fit(data)
-            output = self._model.predict(fh=range(predict_length))
-            output = np.array(output, dtype=np.float64)
-            return output
-        except Exception as e:
-            raise InferenceModelInternalError(str(e))
-
-
-class ExponentialSmoothingModel(BuiltInModel):
-    def __init__(self, attributes):
-        super(ExponentialSmoothingModel, self).__init__(attributes)
-        self._model = ExponentialSmoothing(
-            damped_trend=attributes['damped_trend'],
-            initialization_method=attributes['initialization_method'],
-            optimized=attributes['optimized'],
-            remove_bias=attributes['remove_bias'],
-            use_brute=attributes['use_brute']
-        )
-
-    def inference(self, data):
-        try:
-            predict_length = self._attributes['predict_length']
-            self._model.fit(data)
-            output = self._model.predict(fh=range(predict_length))
-            output = np.array(output, dtype=np.float64)
-            return output
-        except Exception as e:
-            raise InferenceModelInternalError(str(e))
-
-
-class NaiveForecasterModel(BuiltInModel):
-    def __init__(self, attributes):
-        super(NaiveForecasterModel, self).__init__(attributes)
-        self._model = NaiveForecaster(
-            strategy=attributes['strategy'],
-            sp=attributes['sp']
-        )
-
-    def inference(self, data):
-        try:
-            predict_length = self._attributes['predict_length']
-            self._model.fit(data)
-            output = self._model.predict(fh=range(predict_length))
-            output = np.array(output, dtype=np.float64)
-            return output
-        except Exception as e:
-            raise InferenceModelInternalError(str(e))
-
-
-class STLForecasterModel(BuiltInModel):
-    def __init__(self, attributes):
-        super(STLForecasterModel, self).__init__(attributes)
-        self._model = STLForecaster(
-            sp=attributes['sp'],
-            seasonal=attributes['seasonal'],
-            seasonal_deg=attributes['seasonal_deg'],
-            trend_deg=attributes['trend_deg'],
-            low_pass_deg=attributes['low_pass_deg'],
-            seasonal_jump=attributes['seasonal_jump'],
-            trend_jump=attributes['trend_jump'],
-            low_pass_jump=attributes['low_pass_jump']
-        )
-
-    def inference(self, data):
-        try:
-            predict_length = self._attributes['predict_length']
-            self._model.fit(data)
-            output = self._model.predict(fh=range(predict_length))
-            output = np.array(output, dtype=np.float64)
-            return output
-        except Exception as e:
-            raise InferenceModelInternalError(str(e))
-
-
-class GMMHMMModel(BuiltInModel):
-    def __init__(self, attributes):
-        super(GMMHMMModel, self).__init__(attributes)
-        self._model = GMMHMM(
-            n_components=attributes['n_components'],
-            n_mix=attributes['n_mix'],
-            min_covar=attributes['min_covar'],
-            startprob_prior=attributes['startprob_prior'],
-            transmat_prior=attributes['transmat_prior'],
-            means_prior=attributes['means_prior'],
-            means_weight=attributes['means_weight'],
-            weights_prior=attributes['weights_prior'],
-            algorithm=attributes['algorithm'],
-            covariance_type=attributes['covariance_type'],
-            n_iter=attributes['n_iter'],
-            tol=attributes['tol'],
-            params=attributes['params'],
-            init_params=attributes['init_params'],
-            implementation=attributes['implementation']
-        )
-
-    def inference(self, data):
-        try:
-            self._model.fit(data)
-            output = self._model.predict(data)
-            output = np.array(output, dtype=np.int32)
-            return output
-        except Exception as e:
-            raise InferenceModelInternalError(str(e))
-
-
-class GaussianHmmModel(BuiltInModel):
-    def __init__(self, attributes):
-        super(GaussianHmmModel, self).__init__(attributes)
-        self._model = GaussianHMM(
-            n_components=attributes['n_components'],
-            covariance_type=attributes['covariance_type'],
-            min_covar=attributes['min_covar'],
-            startprob_prior=attributes['startprob_prior'],
-            transmat_prior=attributes['transmat_prior'],
-            means_prior=attributes['means_prior'],
-            means_weight=attributes['means_weight'],
-            covars_prior=attributes['covars_prior'],
-            covars_weight=attributes['covars_weight'],
-            algorithm=attributes['algorithm'],
-            n_iter=attributes['n_iter'],
-            tol=attributes['tol'],
-            params=attributes['params'],
-            init_params=attributes['init_params'],
-            implementation=attributes['implementation']
-        )
-
-    def inference(self, data):
-        try:
-            self._model.fit(data)
-            output = self._model.predict(data)
-            output = np.array(output, dtype=np.int32)
-            return output
-        except Exception as e:
-            raise InferenceModelInternalError(str(e))
-
-
-class STRAYModel(BuiltInModel):
-    def __init__(self, attributes):
-        super(STRAYModel, self).__init__(attributes)
-        self._model = STRAY(
-            alpha=attributes['alpha'],
-            k=attributes['k'],
-            knn_algorithm=attributes['knn_algorithm'],
-            p=attributes['p'],
-            size_threshold=attributes['size_threshold'],
-            outlier_tail=attributes['outlier_tail']
-        )
-
-    def inference(self, data):
-        try:
-            data = MinMaxScaler().fit_transform(data)
-            output = self._model.fit_transform(data)
-            # change the output to int
-            output = np.array(output, dtype=np.int32)
-            return output
-        except Exception as e:
-            raise InferenceModelInternalError(str(e))
diff --git a/iotdb-core/ainode/ai/ainode/model/model_factory.py b/iotdb-core/ainode/ai/ainode/model/model_factory.py
deleted file mode 100644
index 480676b67a6..00000000000
--- a/iotdb-core/ainode/ai/ainode/model/model_factory.py
+++ /dev/null
@@ -1,235 +0,0 @@
-# Licensed to the Apache Software Foundation (ASF) under one
-# or more contributor license agreements.  See the NOTICE file
-# distributed with this work for additional information
-# regarding copyright ownership.  The ASF licenses this file
-# to you under the Apache License, Version 2.0 (the
-# "License"); you may not use this file except in compliance
-# with the License.  You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing,
-# software distributed under the License is distributed on an
-# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
-# KIND, either express or implied.  See the License for the
-# specific language governing permissions and limitations
-# under the License.
-#
-import os
-import shutil
-from urllib.parse import urlparse, urljoin
-
-import yaml
-from requests import Session
-from requests.adapters import HTTPAdapter
-
-from ai.ainode.constant import DEFAULT_RECONNECT_TIMES, DEFAULT_RECONNECT_TIMEOUT, DEFAULT_CHUNK_SIZE, \
-    DEFAULT_CONFIG_FILE_NAME, DEFAULT_MODEL_FILE_NAME
-from ai.ainode.exception import InvalidUriError, BadConfigValueError
-from ai.ainode.log import Logger
-from ai.ainode.util.serde import get_data_type_byte_from_str
-from ai.thrift.ainode.ttypes import TConfigs
-
-HTTP_PREFIX = "http://"
-HTTPS_PREFIX = "https://"
-
-logger = Logger()
-
-
-def _parse_uri(uri):
-    """
-    Args:
-        uri (str): uri to parse
-    Returns:
-        is_network_path (bool): True if the url is a network path, False otherwise
-        parsed_uri (str): parsed uri to get related file
-    """
-
-    parse_result = urlparse(uri)
-    is_network_path = parse_result.scheme in ('http', 'https')
-    if is_network_path:
-        return True, uri
-
-    # handle file:// in uri
-    if parse_result.scheme == 'file':
-        uri = uri[7:]
-
-    # handle ~ in uri
-    uri = os.path.expanduser(uri)
-    return False, uri
-
-
-def _download_file(url: str, storage_path: str) -> None:
-    """
-    Args:
-        url: url of file to download
-        storage_path: path to save the file
-    Returns:
-        None
-    """
-    logger.debug(f"download file from {url} to {storage_path}")
-
-    session = Session()
-    adapter = HTTPAdapter(max_retries=DEFAULT_RECONNECT_TIMES)
-    session.mount(HTTP_PREFIX, adapter)
-    session.mount(HTTPS_PREFIX, adapter)
-
-    response = session.get(url, timeout=DEFAULT_RECONNECT_TIMEOUT, stream=True)
-    response.raise_for_status()
-
-    with open(storage_path, 'wb') as file:
-        for chunk in response.iter_content(chunk_size=DEFAULT_CHUNK_SIZE):
-            if chunk:
-                file.write(chunk)
-
-    logger.debug(f"download file from {url} to {storage_path} success")
-
-
-def _register_model_from_network(uri: str, model_storage_path: str,
-                                 config_storage_path: str) -> [TConfigs, str]:
-    """
-    Args:
-        uri: network dir path of model to register, where model.pt and config.yaml are required,
-            e.g. https://huggingface.co/user/modelname/resolve/main/
-        model_storage_path: path to save model.pt
-        config_storage_path: path to save config.yaml
-    Returns:
-        configs: TConfigs
-        attributes: str
-    """
-    # concat uri to get complete url
-    uri = uri if uri.endswith("/") else uri + "/"
-    target_model_path = urljoin(uri, DEFAULT_MODEL_FILE_NAME)
-    target_config_path = urljoin(uri, DEFAULT_CONFIG_FILE_NAME)
-
-    # download config file
-    _download_file(target_config_path, config_storage_path)
-
-    # read and parse config dict from config.yaml
-    with open(config_storage_path, 'r', encoding='utf-8') as file:
-        config_dict = yaml.safe_load(file)
-    configs, attributes = _parse_inference_config(config_dict)
-
-    # if config.yaml is correct, download model file
-    _download_file(target_model_path, model_storage_path)
-    return configs, attributes
-
-
-def _register_model_from_local(uri: str, model_storage_path: str,
-                               config_storage_path: str) -> [TConfigs, str]:
-    """
-    Args:
-        uri: local dir path of model to register, where model.pt and config.yaml are required,
-            e.g. /Users/admin/Desktop/model
-        model_storage_path: path to save model.pt
-        config_storage_path: path to save config.yaml
-    Returns:
-        configs: TConfigs
-        attributes: str
-    """
-    # concat uri to get complete path
-    target_model_path = os.path.join(uri, DEFAULT_MODEL_FILE_NAME)
-    target_config_path = os.path.join(uri, DEFAULT_CONFIG_FILE_NAME)
-
-    # check if file exist
-    exist_model_file = os.path.exists(target_model_path)
-    exist_config_file = os.path.exists(target_config_path)
-
-    configs = None
-    attributes = None
-    if exist_model_file and exist_config_file:
-        # copy config.yaml
-        logger.debug(f"copy file from {target_config_path} to {config_storage_path}")
-        shutil.copy(target_config_path, config_storage_path)
-        logger.debug(f"copy file from {target_config_path} to {config_storage_path} success")
-
-        # read and parse config dict from config.yaml
-        with open(config_storage_path, 'r', encoding='utf-8') as file:
-            config_dict = yaml.safe_load(file)
-        configs, attributes = _parse_inference_config(config_dict)
-
-        # if config.yaml is correct, copy model file
-        logger.debug(f"copy file from {target_model_path} to {model_storage_path}")
-        shutil.copy(target_model_path, model_storage_path)
-        logger.debug(f"copy file from {target_model_path} to {model_storage_path} success")
-
-    elif not exist_model_file or not exist_config_file:
-        raise InvalidUriError(uri)
-
-    return configs, attributes
-
-
-def _parse_inference_config(config_dict):
-    """
-    Args:
-        config_dict: dict
-            - configs: dict
-                - input_shape (list<i32>): input shape of the model and needs to be two-dimensional array like [96, 2]
-                - output_shape (list<i32>): output shape of the model and needs to be two-dimensional array like [96, 2]
-                - input_type (list<str>): input type of the model and each element needs to be in ['bool', 'int32', 'int64', 'float32', 'float64', 'text'], default float64
-                - output_type (list<str>): output type of the model and each element needs to be in ['bool', 'int32', 'int64', 'float32', 'float64', 'text'], default float64
-            - attributes: dict
-    Returns:
-        configs: TConfigs
-        attributes: str
-    """
-    configs = config_dict['configs']
-
-    # check if input_shape and output_shape are two-dimensional array
-    if not (isinstance(configs['input_shape'], list) and len(configs['input_shape']) == 2):
-        raise BadConfigValueError('input_shape', configs['input_shape'],
-                                  'input_shape should be a two-dimensional array.')
-    if not (isinstance(configs['output_shape'], list) and len(configs['output_shape']) == 2):
-        raise BadConfigValueError('output_shape', configs['output_shape'],
-                                  'output_shape should be a two-dimensional array.')
-
-    # check if input_shape and output_shape are positive integer
-    input_shape_is_positive_number = isinstance(configs['input_shape'][0], int) and isinstance(
-        configs['input_shape'][1], int) and configs['input_shape'][0] > 0 and configs['input_shape'][1] > 0
-    if not input_shape_is_positive_number:
-        raise BadConfigValueError('input_shape', configs['input_shape'],
-                                  'element in input_shape should be positive integer.')
-
-    output_shape_is_positive_number = isinstance(configs['output_shape'][0], int) and isinstance(
-        configs['output_shape'][1], int) and configs['output_shape'][0] > 0 and configs['output_shape'][1] > 0
-    if not output_shape_is_positive_number:
-        raise BadConfigValueError('output_shape', configs['output_shape'],
-                                  'element in output_shape should be positive integer.')
-
-    # check if input_type and output_type are one-dimensional array with right length
-    if 'input_type' in configs and not (
-            isinstance(configs['input_type'], list) and len(configs['input_type']) == configs['input_shape'][1]):
-        raise BadConfigValueError('input_type', configs['input_type'],
-                                  'input_type should be a one-dimensional array and length of it should be equal to input_shape[1].')
-
-    if 'output_type' in configs and not (
-            isinstance(configs['output_type'], list) and len(configs['output_type']) == configs['output_shape'][1]):
-        raise BadConfigValueError('output_type', configs['output_type'],
-                                  'output_type should be a one-dimensional array and length of it should be equal to output_shape[1].')
-
-    # parse input_type and output_type to byte
-    if 'input_type' in configs:
-        input_type = [get_data_type_byte_from_str(x) for x in configs['input_type']]
-    else:
-        input_type = [get_data_type_byte_from_str('float32')] * configs['input_shape'][1]
-
-    if 'output_type' in configs:
-        output_type = [get_data_type_byte_from_str(x) for x in configs['output_type']]
-    else:
-        output_type = [get_data_type_byte_from_str('float32')] * configs['output_shape'][1]
-
-    # parse attributes
-    attributes = ""
-    if 'attributes' in config_dict:
-        attributes = str(config_dict['attributes'])
-
-    return TConfigs(configs['input_shape'], configs['output_shape'], input_type, output_type), attributes
-
-
-def fetch_model_by_uri(uri: str, model_storage_path: str, config_storage_path: str):
-    is_network_path, uri = _parse_uri(uri)
-
-    if is_network_path:
-        return _register_model_from_network(uri, model_storage_path, config_storage_path)
-    else:
-        return _register_model_from_local(uri, model_storage_path, config_storage_path)
diff --git a/iotdb-core/ainode/ai/ainode/model/model_storage.py b/iotdb-core/ainode/ai/ainode/model/model_storage.py
deleted file mode 100644
index de5227d0d83..00000000000
--- a/iotdb-core/ainode/ai/ainode/model/model_storage.py
+++ /dev/null
@@ -1,123 +0,0 @@
-# Licensed to the Apache Software Foundation (ASF) under one
-# or more contributor license agreements.  See the NOTICE file
-# distributed with this work for additional information
-# regarding copyright ownership.  The ASF licenses this file
-# to you under the Apache License, Version 2.0 (the
-# "License"); you may not use this file except in compliance
-# with the License.  You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing,
-# software distributed under the License is distributed on an
-# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
-# KIND, either express or implied.  See the License for the
-# specific language governing permissions and limitations
-# under the License.
-#
-
-import os
-import shutil
-from collections.abc import Callable
-
-import torch
-import torch._dynamo
-from pylru import lrucache
-
-from ai.ainode.config import AINodeDescriptor
-from ai.ainode.constant import (DEFAULT_MODEL_FILE_NAME,
-                                DEFAULT_CONFIG_FILE_NAME)
-from ai.ainode.exception import ModelNotExistError
-from ai.ainode.log import Logger
-from ai.ainode.model.model_factory import fetch_model_by_uri
-from ai.ainode.util.lock import ModelLockPool
-from manager.training_manager import get_args
-from model.TimerXL.models.timer_xl import Model
-
-logger = Logger()
-
-
-class ModelStorage(object):
-    def __init__(self):
-        self._model_dir = os.path.join(os.getcwd(), AINodeDescriptor().get_config().get_ain_models_dir())
-        if not os.path.exists(self._model_dir):
-            try:
-                os.makedirs(self._model_dir)
-            except PermissionError as e:
-                logger.error(e)
-                raise e
-        self._lock_pool = ModelLockPool()
-        self._model_cache = lrucache(AINodeDescriptor().get_config().get_ain_model_storage_cache_size())
-
-    def register_model(self, model_id: str, uri: str):
-        """
-        Args:
-            model_id: id of model to register
-            uri: network dir path or local dir path of model to register, where model.pt and config.yaml are required,
-                e.g. https://huggingface.co/user/modelname/resolve/main/ or /Users/admin/Desktop/model
-        Returns:
-            configs: TConfigs
-            attributes: str
-        """
-        storage_path = os.path.join(self._model_dir, f'{model_id}')
-        # create storage dir if not exist
-        if not os.path.exists(storage_path):
-            os.makedirs(storage_path)
-        model_storage_path = os.path.join(storage_path, DEFAULT_MODEL_FILE_NAME)
-        config_storage_path = os.path.join(storage_path, DEFAULT_CONFIG_FILE_NAME)
-        return fetch_model_by_uri(uri, model_storage_path, config_storage_path)
-
-    def load_model(self, model_id: str, acceleration: bool) -> Callable:
-        """
-        Returns:
-            model: a ScriptModule contains model architecture and parameters, which can be deployed cross-platform
-        """
-        ain_models_dir = os.path.join(self._model_dir, f'{model_id}')
-        model_path = os.path.join(ain_models_dir, DEFAULT_MODEL_FILE_NAME)
-        with self._lock_pool.get_lock(model_id).read_lock():
-            if model_path in self._model_cache:
-                model = self._model_cache[model_path]
-                if isinstance(model, torch._dynamo.eval_frame.OptimizedModule) or not acceleration:
-                    return model
-                else:
-                    model = torch.compile(model)
-                    self._model_cache[model_path] = model
-                    return model
-            else:
-                # todo: use modelType instead
-                if 'timer' in model_id:
-                    model_file_path = os.path.join(ain_models_dir, 'checkpoint.pth')
-                    state_dict = torch.load(model_file_path)
-                    model = Model(get_args())
-                    model.load_state_dict(state_dict)
-                    self._model_cache[model_path] = model
-                    return model
-                elif not os.path.exists(model_path):
-                    raise ModelNotExistError(model_path)
-                else:
-                    model = torch.jit.load(model_path)
-                    if acceleration:
-                        try:
-                            model = torch.compile(model)
-                        except Exception as e:
-                            logger.warning(f"acceleration failed, fallback to normal mode: {str(e)}")
-                    self._model_cache[model_path] = model
-                    return model
-
-    def delete_model(self, model_id: str) -> None:
-        """
-        Args:
-            model_id: id of model to delete
-        Returns:
-            None
-        """
-        storage_path = os.path.join(self._model_dir, f'{model_id}')
-        with self._lock_pool.get_lock(model_id).write_lock():
-            if os.path.exists(storage_path):
-                for file_name in os.listdir(storage_path):
-                    self._remove_from_cache(os.path.join(storage_path, file_name))
-                shutil.rmtree(storage_path)
-
-    def _remove_from_cache(self, file_path: str) -> None:
-        if file_path in self._model_cache:
-            del self._model_cache[file_path]
diff --git a/iotdb-core/ainode/ai/ainode/util/__init__.py b/iotdb-core/ainode/ai/ainode/util/__init__.py
deleted file mode 100644
index 2a1e720805f..00000000000
--- a/iotdb-core/ainode/ai/ainode/util/__init__.py
+++ /dev/null
@@ -1,17 +0,0 @@
-# Licensed to the Apache Software Foundation (ASF) under one
-# or more contributor license agreements.  See the NOTICE file
-# distributed with this work for additional information
-# regarding copyright ownership.  The ASF licenses this file
-# to you under the Apache License, Version 2.0 (the
-# "License"); you may not use this file except in compliance
-# with the License.  You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing,
-# software distributed under the License is distributed on an
-# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
-# KIND, either express or implied.  See the License for the
-# specific language governing permissions and limitations
-# under the License.
-#
diff --git a/iotdb-core/ainode/ai/ainode/util/masking.py b/iotdb-core/ainode/ai/ainode/util/masking.py
deleted file mode 100644
index 63a0bf5f280..00000000000
--- a/iotdb-core/ainode/ai/ainode/util/masking.py
+++ /dev/null
@@ -1,35 +0,0 @@
-import torch
-
-class TriangularCausalMask():
-    def __init__(self, B, L, device="cpu"):
-        mask_shape = [B, 1, L, L]
-        with torch.no_grad():
-            self._mask = torch.triu(torch.ones(mask_shape, dtype=torch.bool), diagonal=1).to(device)
-
-    @property
-    def mask(self):
-        return self._mask
-
-class TimerMultivariateMask():
-    def __init__(self, B, n_vars, n_tokens, device="cpu"):
-        mask_shape = [B, 1, n_tokens, n_tokens]
-        with torch.no_grad():
-            self._mask1 = torch.ones((n_vars, n_vars), dtype=torch.bool).to(device)
-            self._mask2 = torch.triu(torch.ones(mask_shape, dtype=torch.bool), diagonal=1).to(device)
-            self._mask = torch.kron(self._mask1, self._mask2)
-    @property
-    def mask(self):
-        return self._mask
-
-class TimerCovariateMask():
-    def __init__(self, B, n_vars, n_tokens, device="cpu"):
-        mask_shape = [B, 1, n_tokens, n_tokens]
-        with torch.no_grad():
-            self._mask1 = torch.eye(n_vars, dtype=torch.bool).to(device)
-            self._mask2 = torch.tril(torch.ones(mask_shape, dtype=torch.bool)).to(device)
-            self._mask = ~torch.kron(self._mask1, self._mask2)
-            self._mask[:, :, -n_tokens:, :-n_tokens] = False
-            
-    @property
-    def mask(self):
-        return self._mask
\ No newline at end of file
diff --git a/iotdb-core/ainode/ai/ainode/util/metrics.py b/iotdb-core/ainode/ai/ainode/util/metrics.py
deleted file mode 100644
index e39dcd4285c..00000000000
--- a/iotdb-core/ainode/ai/ainode/util/metrics.py
+++ /dev/null
@@ -1,44 +0,0 @@
-import numpy as np
-
-
-def RSE(pred, true):
-    return np.sqrt(np.sum((true - pred) ** 2)) / np.sqrt(np.sum((true - true.mean()) ** 2))
-
-
-def CORR(pred, true):
-    u = ((true - true.mean(0)) * (pred - pred.mean(0))).sum(0)
-    d = np.sqrt(((true - true.mean(0)) ** 2 *
-                (pred - pred.mean(0)) ** 2).sum(0))
-    return (u / d).mean(-1)
-
-
-def MAE(pred, true):
-    return np.mean(np.abs(pred - true))
-
-
-def MSE(pred, true):
-    return np.mean((pred - true) ** 2)
-
-
-def RMSE(pred, true):
-    return np.sqrt(MSE(pred, true))
-
-
-def MAPE(pred, true):
-    return np.mean(np.abs((pred - true) / true))
-
-
-def MSPE(pred, true):
-    return np.mean(np.square((pred - true) / true))
-
-def SMAPE(pred, true):
-    return np.mean(200 * np.abs(pred - true) / (np.abs(pred) + np.abs(true) + 1e-8))
-
-def metric(pred, true):
-    mae = MAE(pred, true)
-    mse = MSE(pred, true)
-    rmse = RMSE(pred, true)
-    mape = MAPE(pred, true)
-    mspe = MSPE(pred, true)
-    smape = SMAPE(pred, true)
-    return mae, mse, rmse, mape, mspe, smape
diff --git a/iotdb-core/ainode/ai/ainode/util/tools.py b/iotdb-core/ainode/ai/ainode/util/tools.py
deleted file mode 100644
index 6e0dd6595d5..00000000000
--- a/iotdb-core/ainode/ai/ainode/util/tools.py
+++ /dev/null
@@ -1,96 +0,0 @@
-import numpy as np
-import torch
-import matplotlib.pyplot as plt
-import torch.distributed as dist
-plt.switch_backend('agg')
-
-
-def adjust_learning_rate(optimizer, epoch, args):
-    if args.lradj == 'type1':
-        lr_adjust = {epoch: args.learning_rate * (0.1 ** epoch)}
-    elif args.lradj == 'type2':
-        lr_adjust = {epoch: args.learning_rate * (0.5 ** epoch)}
-    elif args.lradj == 'type3':
-        lr_adjust = {epoch: args.learning_rate * (0.9 ** epoch)}
-
-
-    if epoch in lr_adjust.keys():
-        lr = lr_adjust[epoch]
-        for param_group in optimizer.param_groups:
-            param_group['lr'] = lr
-        print('Updating learning rate to {}'.format(lr))
-
-
-class EarlyStopping:
-    def __init__(self, args, verbose=False, delta=0):
-        self.patience = args.patience
-        self.verbose = verbose
-        self.counter = 0
-        self.best_score = None
-        self.early_stop = False
-        self.val_loss_min = np.inf
-        self.delta = delta
-        self.dp = args.dp
-        self.ddp = args.ddp
-        if self.ddp:
-            self.local_rank = args.local_rank
-        else:
-            self.local_rank = None
-
-    def __call__(self, val_loss, model, path):
-        score = -val_loss
-        if self.best_score is None:
-            self.best_score = score
-            if self.verbose:
-                print(
-                    f'Validation loss decreased ({self.val_loss_min:.6f} --> {val_loss:.6f}).')
-            self.val_loss_min = val_loss
-            if self.ddp:
-                if self.local_rank == 0:
-                    self.save_checkpoint(val_loss, model, path)
-                dist.barrier()
-            else:
-                self.save_checkpoint(val_loss, model, path)
-        elif score < self.best_score + self.delta:
-            self.counter += 1
-            print(
-                f'EarlyStopping counter: {self.counter} out of {self.patience}')
-            if self.counter >= self.patience:
-                self.early_stop = True
-        else:
-            self.best_score = score
-            if self.ddp:
-                if self.local_rank == 0:
-                    self.save_checkpoint(val_loss, model, path)
-                dist.barrier()
-            else:
-                self.save_checkpoint(val_loss, model, path)
-            if self.verbose:
-                print(
-                    f'Validation loss decreased ({self.val_loss_min:.6f} --> {val_loss:.6f}).')
-            self.val_loss_min = val_loss
-            self.counter = 0
-
-    def save_checkpoint(self, val_loss, model, path):
-        if self.dp:
-            model = model.module
-        param_grad_dic = {
-            k: v.requires_grad for (k, v) in model.named_parameters()
-        }
-        state_dict = model.state_dict()
-        for k in list(state_dict.keys()):
-            if k in param_grad_dic.keys() and not param_grad_dic[k]:
-                # delete parameters that do not require gradient
-                del state_dict[k]
-        torch.save(state_dict, path + '/' + f'checkpoint.pth')
-
-def visual(true, preds=None, name='./pic/test.pdf'):
-    """
-    Results visualization
-    """
-    plt.figure()
-    plt.plot(true, label='GroundTruth', linewidth=2)
-    if preds is not None:
-        plt.plot(preds, label='Prediction', linewidth=2)
-    plt.legend()
-    plt.savefig(name, bbox_inches='tight')
\ No newline at end of file
diff --git a/iotdb-core/ainode/ai/__init__.py b/iotdb-core/ainode/ainode/__init__.py
similarity index 100%
rename from iotdb-core/ainode/ai/__init__.py
rename to iotdb-core/ainode/ainode/__init__.py
diff --git a/iotdb-core/ainode/ai/ainode/__init__.py b/iotdb-core/ainode/ainode/core/__init__.py
similarity index 100%
rename from iotdb-core/ainode/ai/ainode/__init__.py
rename to iotdb-core/ainode/ainode/core/__init__.py
diff --git a/iotdb-core/ainode/ai/ainode/client.py b/iotdb-core/ainode/ainode/core/client.py
similarity index 92%
rename from iotdb-core/ainode/ai/ainode/client.py
rename to iotdb-core/ainode/ainode/core/client.py
index d7408917724..2796d46ab93 100644
--- a/iotdb-core/ainode/ai/ainode/client.py
+++ b/iotdb-core/ainode/ainode/core/client.py
@@ -21,16 +21,16 @@ from thrift.Thrift import TException
 from thrift.protocol import TCompactProtocol, TBinaryProtocol
 from thrift.transport import TSocket, TTransport
 
-from ai.ainode.config import AINodeDescriptor
-from ai.ainode.constant import TSStatusCode
-from ai.ainode.log import Logger
-from ai.ainode.util.decorator import singleton
-from ai.ainode.util.status import verify_success
-from ai.thrift.common.ttypes import TEndPoint, TSStatus, TAINodeLocation, TAINodeConfiguration
-from ai.thrift.confignode import IConfigNodeRPCService
-from ai.thrift.confignode.ttypes import (TAINodeRemoveReq, TNodeVersionInfo,
-                                         TAINodeRegisterReq, TAINodeRestartReq)
-from ai.thrift.confignode.ttypes import TUpdateModelInfoReq
+from ainode.core.config import AINodeDescriptor
+from ainode.core.constant import TSStatusCode
+from ainode.core.log import Logger
+from ainode.core.util.decorator import singleton
+from ainode.core.util.status import verify_success
+from ainode.thrift.common.ttypes import TEndPoint, TSStatus, TAINodeLocation, TAINodeConfiguration
+from ainode.thrift.confignode import IConfigNodeRPCService
+from ainode.thrift.confignode.ttypes import (TAINodeRemoveReq, TNodeVersionInfo,
+                                             TAINodeRegisterReq, TAINodeRestartReq)
+from ainode.thrift.confignode.ttypes import TUpdateModelInfoReq
 
 logger = Logger()
 
@@ -214,8 +214,8 @@ class ConfigNodeClient(object):
                 )
                 if ainode_id is not None:
                     req.aiNodeIds = ainode_id
-                req.inputShapes = [input_length]
-                req.outputShapes = [output_length]
+                req.inputLength = input_length
+                req.outputLength = output_length
                 status = self._client.updateModelInfo(req)
                 if not self._update_config_node_leader(status):
                     verify_success(status, "An error occurs when calling update model info")
diff --git a/iotdb-core/ainode/ai/ainode/config.py b/iotdb-core/ainode/ainode/core/config.py
similarity index 91%
rename from iotdb-core/ainode/ai/ainode/config.py
rename to iotdb-core/ainode/ainode/core/config.py
index a68b978d43e..036dc20b9c2 100644
--- a/iotdb-core/ainode/ai/ainode/config.py
+++ b/iotdb-core/ainode/ainode/core/config.py
@@ -17,17 +17,17 @@
 #
 import os
 
-from ai.ainode.constant import (AINODE_CONF_DIRECTORY_NAME,
-                                AINODE_CONF_FILE_NAME,
-                                AINODE_MODELS_DIR, AINODE_LOG_DIR, AINODE_SYSTEM_DIR, AINODE_INFERENCE_RPC_ADDRESS,
-                                AINODE_INFERENCE_RPC_PORT, AINODE_THRIFT_COMPRESSION_ENABLED,
-                                AINODE_SYSTEM_FILE_NAME, AINODE_CLUSTER_NAME, AINODE_VERSION_INFO, AINODE_BUILD_INFO,
-                                AINODE_CONF_GIT_FILE_NAME, AINODE_CONF_POM_FILE_NAME, AINODE_ROOT_DIR,
-                                AINODE_ROOT_CONF_DIRECTORY_NAME)
-from ai.ainode.exception import BadNodeUrlError
-from ai.ainode.log import Logger
-from ai.ainode.util.decorator import singleton
-from ai.thrift.common.ttypes import TEndPoint
+from ainode.core.constant import (AINODE_CONF_DIRECTORY_NAME,
+                                  AINODE_CONF_FILE_NAME,
+                                  AINODE_MODELS_DIR, AINODE_LOG_DIR, AINODE_SYSTEM_DIR, AINODE_INFERENCE_RPC_ADDRESS,
+                                  AINODE_INFERENCE_RPC_PORT, AINODE_THRIFT_COMPRESSION_ENABLED,
+                                  AINODE_SYSTEM_FILE_NAME, AINODE_CLUSTER_NAME, AINODE_VERSION_INFO, AINODE_BUILD_INFO,
+                                  AINODE_CONF_GIT_FILE_NAME, AINODE_CONF_POM_FILE_NAME, AINODE_ROOT_DIR,
+                                  AINODE_ROOT_CONF_DIRECTORY_NAME)
+from ainode.core.exception import BadNodeUrlError
+from ainode.core.log import Logger
+from ainode.core.util.decorator import singleton
+from ainode.thrift.common.ttypes import TEndPoint
 
 logger = Logger()
 
diff --git a/iotdb-core/ainode/ai/ainode/constant.py b/iotdb-core/ainode/ainode/core/constant.py
similarity index 99%
rename from iotdb-core/ainode/ai/ainode/constant.py
rename to iotdb-core/ainode/ainode/core/constant.py
index a50ac382e65..596cde0b224 100644
--- a/iotdb-core/ainode/ai/ainode/constant.py
+++ b/iotdb-core/ainode/ainode/core/constant.py
@@ -23,7 +23,7 @@ from typing import List
 
 AINODE_CONF_DIRECTORY_NAME = "conf"
 AINODE_ROOT_CONF_DIRECTORY_NAME = "conf"
-AINODE_CONF_FILE_NAME = "ai-ainode.properties"
+AINODE_CONF_FILE_NAME = "iotdb-ainode.properties"
 AINODE_CONF_GIT_FILE_NAME = "git.properties"
 AINODE_CONF_POM_FILE_NAME = "pom.properties"
 AINODE_SYSTEM_FILE_NAME = "system.properties"
diff --git a/iotdb-core/ainode/ai/ainode/exception.py b/iotdb-core/ainode/ainode/core/exception.py
similarity index 98%
rename from iotdb-core/ainode/ai/ainode/exception.py
rename to iotdb-core/ainode/ainode/core/exception.py
index cb8e2f0b1ba..a9b8c496d65 100644
--- a/iotdb-core/ainode/ai/ainode/exception.py
+++ b/iotdb-core/ainode/ainode/core/exception.py
@@ -17,7 +17,7 @@
 #
 import re
 
-from ai.ainode.constant import DEFAULT_MODEL_FILE_NAME, DEFAULT_CONFIG_FILE_NAME
+from ainode.core.constant import DEFAULT_MODEL_FILE_NAME, DEFAULT_CONFIG_FILE_NAME
 
 
 class _BaseError(Exception):
diff --git a/iotdb-core/ainode/ai/ainode/handler.py b/iotdb-core/ainode/ainode/core/handler.py
similarity index 73%
rename from iotdb-core/ainode/ai/ainode/handler.py
rename to iotdb-core/ainode/ainode/core/handler.py
index 3b7e47cd505..292fd2d4b5f 100644
--- a/iotdb-core/ainode/ai/ainode/handler.py
+++ b/iotdb-core/ainode/ainode/core/handler.py
@@ -16,15 +16,15 @@
 # under the License.
 #
 
-from ai.ainode.manager.cluster_manager import ClusterManager
-from ai.ainode.manager.inference_manager import InferenceManager
-from ai.ainode.manager.model_manager import ModelManager
-from ai.thrift.ainode import IAINodeRPCService
-from ai.thrift.ainode.ttypes import (TDeleteModelReq, TRegisterModelReq,
-                                     TAIHeartbeatReq, TInferenceReq, TRegisterModelResp, TInferenceResp,
-                                     TAIHeartbeatResp, TTrainingReq)
-from ai.thrift.common.ttypes import TSStatus
-from manager.training_manager import TrainingManager
+from ainode.core.manager.cluster_manager import ClusterManager
+from ainode.core.manager.inference_manager import InferenceManager
+from ainode.core.manager.model_manager import ModelManager
+from ainode.thrift.ainode import IAINodeRPCService
+from ainode.thrift.ainode.ttypes import (TDeleteModelReq, TRegisterModelReq,
+                                         TAIHeartbeatReq, TInferenceReq, TRegisterModelResp, TInferenceResp,
+                                         TAIHeartbeatResp, TTrainingReq)
+from ainode.thrift.common.ttypes import TSStatus
+from ainode.core.manager.training_manager import TrainingManager
 
 
 class AINodeRPCServiceHandler(IAINodeRPCService.Iface):
diff --git a/iotdb-core/ainode/ai/ainode/log.py b/iotdb-core/ainode/ainode/core/log.py
similarity index 94%
rename from iotdb-core/ainode/ai/ainode/log.py
rename to iotdb-core/ainode/ainode/core/log.py
index 04e5c2f1679..34a24d370c4 100644
--- a/iotdb-core/ainode/ai/ainode/log.py
+++ b/iotdb-core/ainode/ainode/core/log.py
@@ -23,8 +23,8 @@ import random
 import sys
 import threading
 
-from ai.ainode.constant import STD_LEVEL, AINODE_LOG_FILE_NAMES, AINODE_LOG_FILE_LEVELS
-from ai.ainode.util.decorator import singleton
+from ainode.core.constant import STD_LEVEL, AINODE_LOG_FILE_NAMES, AINODE_LOG_FILE_LEVELS
+from ainode.core.util.decorator import singleton
 
 
 class LoggerFilter(logging.Filter):
@@ -43,10 +43,10 @@ class LoggerFilter(logging.Filter):
         stack_info = ""
         frame_info = stack_trace[7]
         file_name = frame_info.filename
-        # if file_name is not in current working directory, find the first "ai" in the path
+        # if file_name is not in current working directory, find the first "core" in the path
         for l in range(len(file_name)):
             i = len(file_name) - l - 1
-            if file_name[i:].startswith("ai/") or file_name[i:].startswith("ai\\"):
+            if file_name[i:].startswith("core/") or file_name[i:].startswith("core\\"):
                 file_name = file_name[i:]
                 break
 
diff --git a/iotdb-core/ainode/ai/ainode/manager/__init__.py b/iotdb-core/ainode/ainode/core/manager/__init__.py
similarity index 100%
rename from iotdb-core/ainode/ai/ainode/manager/__init__.py
rename to iotdb-core/ainode/ainode/core/manager/__init__.py
diff --git a/iotdb-core/ainode/ai/ainode/manager/cluster_manager.py b/iotdb-core/ainode/ainode/core/manager/cluster_manager.py
similarity index 93%
rename from iotdb-core/ainode/ai/ainode/manager/cluster_manager.py
rename to iotdb-core/ainode/ainode/core/manager/cluster_manager.py
index 7948f95b125..da7008b7762 100644
--- a/iotdb-core/ainode/ai/ainode/manager/cluster_manager.py
+++ b/iotdb-core/ainode/ainode/core/manager/cluster_manager.py
@@ -17,8 +17,8 @@
 #
 import psutil
 
-from ai.thrift.ainode.ttypes import TAIHeartbeatResp, TAIHeartbeatReq
-from ai.thrift.common.ttypes import TLoadSample
+from ainode.thrift.ainode.ttypes import TAIHeartbeatResp, TAIHeartbeatReq
+from ainode.thrift.common.ttypes import TLoadSample
 
 
 class ClusterManager:
diff --git a/iotdb-core/ainode/ai/ainode/manager/inference_manager.py b/iotdb-core/ainode/ainode/core/manager/inference_manager.py
similarity index 95%
rename from iotdb-core/ainode/ai/ainode/manager/inference_manager.py
rename to iotdb-core/ainode/ainode/core/manager/inference_manager.py
index c5480575adb..ebfc6d41c9e 100644
--- a/iotdb-core/ainode/ai/ainode/manager/inference_manager.py
+++ b/iotdb-core/ainode/ainode/core/manager/inference_manager.py
@@ -18,13 +18,13 @@
 import pandas as pd
 from torch import tensor
 
-from ai.ainode.constant import TSStatusCode
-from ai.ainode.exception import InvalidWindowArgumentError, InferenceModelInternalError, runtime_error_extractor
-from ai.ainode.log import Logger
-from ai.ainode.manager.model_manager import ModelManager
-from ai.ainode.util.serde import convert_to_binary, convert_to_df
-from ai.ainode.util.status import get_status
-from ai.thrift.ainode.ttypes import TInferenceReq, TInferenceResp
+from ainode.core.constant import TSStatusCode
+from ainode.core.exception import InvalidWindowArgumentError, InferenceModelInternalError, runtime_error_extractor
+from ainode.core.log import Logger
+from ainode.core.manager.model_manager import ModelManager
+from ainode.core.util.serde import convert_to_binary, convert_to_df
+from ainode.core.util.status import get_status
+from ainode.thrift.ainode.ttypes import TInferenceReq, TInferenceResp
 
 logger = Logger()
 
diff --git a/iotdb-core/ainode/ai/ainode/manager/model_manager.py b/iotdb-core/ainode/ainode/core/manager/model_manager.py
similarity index 86%
rename from iotdb-core/ainode/ai/ainode/manager/model_manager.py
rename to iotdb-core/ainode/ainode/core/manager/model_manager.py
index c4119d16d8b..ead833a5839 100644
--- a/iotdb-core/ainode/ai/ainode/manager/model_manager.py
+++ b/iotdb-core/ainode/ainode/core/manager/model_manager.py
@@ -19,14 +19,14 @@ from typing import Callable
 
 from yaml import YAMLError
 
-from ai.ainode.constant import TSStatusCode, BuiltInModelType
-from ai.ainode.exception import InvalidUriError, BadConfigValueError, BuiltInModelNotSupportError
-from ai.ainode.log import Logger
-from ai.ainode.model.built_in_model_factory import fetch_built_in_model
-from ai.ainode.model.model_storage import ModelStorage
-from ai.ainode.util.status import get_status
-from ai.thrift.ainode.ttypes import TRegisterModelReq, TRegisterModelResp, TDeleteModelReq
-from ai.thrift.common.ttypes import TSStatus
+from ainode.core.constant import TSStatusCode, BuiltInModelType
+from ainode.core.exception import InvalidUriError, BadConfigValueError, BuiltInModelNotSupportError
+from ainode.core.log import Logger
+from ainode.core.model.built_in_model_factory import fetch_built_in_model
+from ainode.core.model.model_storage import ModelStorage
+from ainode.core.util.status import get_status
+from ainode.thrift.ainode.ttypes import TRegisterModelReq, TRegisterModelResp, TDeleteModelReq
+from ainode.thrift.common.ttypes import TSStatus
 
 logger = Logger()
 
diff --git a/iotdb-core/ainode/ai/ainode/util/decorator.py b/iotdb-core/ainode/ainode/core/manager/training_manager.py
similarity index 78%
copy from iotdb-core/ainode/ai/ainode/util/decorator.py
copy to iotdb-core/ainode/ainode/core/manager/training_manager.py
index 33b9f4835ac..0233d270469 100644
--- a/iotdb-core/ainode/ai/ainode/util/decorator.py
+++ b/iotdb-core/ainode/ainode/core/manager/training_manager.py
@@ -15,12 +15,9 @@
 # specific language governing permissions and limitations
 # under the License.
 #
-def singleton(cls):
-    instances = {}
+class TrainingManager:
+    def __init__(self):
+        pass
 
-    def get_instance(*args, **kwargs):
-        if cls not in instances:
-            instances[cls] = cls(*args, **kwargs)
-        return instances[cls]
-
-    return get_instance
+    def create_training_task(self, model_id:str, model_type, table_list: list[ITableSchema], parameters: dict[str, str]):
+        raise NotImplementedError
diff --git a/iotdb-core/ainode/ai/ainode/script.py b/iotdb-core/ainode/ainode/core/script.py
similarity index 93%
rename from iotdb-core/ainode/ai/ainode/script.py
rename to iotdb-core/ainode/ainode/core/script.py
index 6c74fc5f26f..b27bb6ab61b 100644
--- a/iotdb-core/ainode/ai/ainode/script.py
+++ b/iotdb-core/ainode/ainode/core/script.py
@@ -22,14 +22,14 @@ from datetime import datetime
 
 import psutil
 
-from ai.ainode.client import ClientManager
-from ai.ainode.config import AINodeDescriptor
-from ai.ainode.constant import TSStatusCode, AINODE_SYSTEM_FILE_NAME
-from ai.ainode.exception import MissingConfigError
-from ai.ainode.log import Logger
-from ai.ainode.service import RPCService
-from ai.thrift.common.ttypes import TAINodeLocation, TEndPoint, TAINodeConfiguration, TNodeResource
-from ai.thrift.confignode.ttypes import TNodeVersionInfo
+from ainode.core.client import ClientManager
+from ainode.core.config import AINodeDescriptor
+from ainode.core.constant import TSStatusCode, AINODE_SYSTEM_FILE_NAME
+from ainode.core.exception import MissingConfigError
+from ainode.core.log import Logger
+from ainode.core.service import RPCService
+from ainode.thrift.common.ttypes import TAINodeLocation, TEndPoint, TAINodeConfiguration, TNodeResource
+from ainode.thrift.confignode.ttypes import TNodeVersionInfo
 
 logger = Logger()
 
diff --git a/iotdb-core/ainode/ai/ainode/service.py b/iotdb-core/ainode/ainode/core/service.py
similarity index 91%
rename from iotdb-core/ainode/ai/ainode/service.py
rename to iotdb-core/ainode/ainode/core/service.py
index 5c21ce5cd7c..9532093d1da 100644
--- a/iotdb-core/ainode/ai/ainode/service.py
+++ b/iotdb-core/ainode/ainode/core/service.py
@@ -21,10 +21,10 @@ from thrift.protocol import TCompactProtocol, TBinaryProtocol
 from thrift.server import TServer
 from thrift.transport import TSocket, TTransport
 
-from ai.ainode.config import AINodeDescriptor
-from ai.ainode.handler import AINodeRPCServiceHandler
-from ai.ainode.log import Logger
-from ai.thrift.ainode import IAINodeRPCService
+from ainode.core.config import AINodeDescriptor
+from ainode.core.handler import AINodeRPCServiceHandler
+from ainode.core.log import Logger
+from ainode.thrift.ainode import IAINodeRPCService
 
 logger = Logger()
 
diff --git a/iotdb-core/ainode/ai/ainode/model/__init__.py b/iotdb-core/ainode/ainode/core/util/__init__.py
similarity index 100%
rename from iotdb-core/ainode/ai/ainode/model/__init__.py
rename to iotdb-core/ainode/ainode/core/util/__init__.py
diff --git a/iotdb-core/ainode/ai/ainode/util/decorator.py b/iotdb-core/ainode/ainode/core/util/decorator.py
similarity index 100%
rename from iotdb-core/ainode/ai/ainode/util/decorator.py
rename to iotdb-core/ainode/ainode/core/util/decorator.py
diff --git a/iotdb-core/ainode/ai/ainode/util/lock.py b/iotdb-core/ainode/ainode/core/util/lock.py
similarity index 100%
rename from iotdb-core/ainode/ai/ainode/util/lock.py
rename to iotdb-core/ainode/ainode/core/util/lock.py
diff --git a/iotdb-core/ainode/ai/ainode/util/serde.py b/iotdb-core/ainode/ainode/core/util/serde.py
similarity index 99%
rename from iotdb-core/ainode/ai/ainode/util/serde.py
rename to iotdb-core/ainode/ainode/core/util/serde.py
index c6b677896c8..b9edccfd03e 100644
--- a/iotdb-core/ainode/ai/ainode/util/serde.py
+++ b/iotdb-core/ainode/ainode/core/util/serde.py
@@ -21,7 +21,7 @@ from enum import Enum
 import numpy as np
 import pandas as pd
 
-from ai.ainode.exception import BadConfigValueError
+from ainode.core.exception import BadConfigValueError
 
 
 class TSDataType(Enum):
diff --git a/iotdb-core/ainode/ai/ainode/util/status.py b/iotdb-core/ainode/ainode/core/util/status.py
similarity index 90%
rename from iotdb-core/ainode/ai/ainode/util/status.py
rename to iotdb-core/ainode/ainode/core/util/status.py
index 8cb983b5b44..37368b0068b 100644
--- a/iotdb-core/ainode/ai/ainode/util/status.py
+++ b/iotdb-core/ainode/ainode/core/util/status.py
@@ -16,9 +16,9 @@
 # under the License.
 #
 
-from ai.ainode.constant import TSStatusCode
-from ai.ainode.log import Logger
-from ai.thrift.common.ttypes import TSStatus
+from ainode.core.constant import TSStatusCode
+from ainode.core.log import Logger
+from ainode.thrift.common.ttypes import TSStatus
 
 
 def get_status(status_code: TSStatusCode, message: str = None) -> TSStatus:
diff --git a/iotdb-core/ainode/pom.xml b/iotdb-core/ainode/pom.xml
index ad6679c7701..ee3137cf664 100644
--- a/iotdb-core/ainode/pom.xml
+++ b/iotdb-core/ainode/pom.xml
@@ -67,7 +67,7 @@
                             <directory>dist</directory>
                         </fileset>
                         <fileset>
-                            <directory>iotdb</directory>
+                            <directory>ainode</directory>
                             <includes>
                                 <include>conf/</include>
                                 <include>thrift/</include>
@@ -138,7 +138,7 @@
                             <goal>copy-resources</goal>
                         </goals>
                         <configuration>
-                            <outputDirectory>${basedir}/iotdb/thrift/</outputDirectory>
+                            <outputDirectory>${basedir}/ainode/thrift/</outputDirectory>
                             <resources>
                                 <resource>
                                     <directory>${basedir}/../../iotdb-protocol/thrift-commons/target/generated-sources-python/iotdb/thrift/</directory>
@@ -167,7 +167,7 @@
                             <goal>copy-resources</goal>
                         </goals>
                         <configuration>
-                            <outputDirectory>${basedir}/iotdb/conf/</outputDirectory>
+                            <outputDirectory>${basedir}/ainode/conf/</outputDirectory>
                             <resources>
                                 <resource>
                                     <directory>${basedir}/resources/</directory>
@@ -192,7 +192,7 @@
                         </goals>
                         <configuration>
                             <generateGitPropertiesFile>true</generateGitPropertiesFile>
-                            <generateGitPropertiesFilename>${project.basedir}/iotdb/conf/git.properties</generateGitPropertiesFilename>
+                            <generateGitPropertiesFilename>${project.basedir}/ainode/conf/git.properties</generateGitPropertiesFilename>
                             <includeOnlyProperties>
                                 <includeOnlyProperty>^git.commit.id.abbrev$</includeOnlyProperty>
                                 <includeOnlyProperty>^git.dirty$</includeOnlyProperty>
diff --git a/iotdb-core/ainode/pyproject.toml b/iotdb-core/ainode/pyproject.toml
index 161757d21c7..4f80ca6b03b 100644
--- a/iotdb-core/ainode/pyproject.toml
+++ b/iotdb-core/ainode/pyproject.toml
@@ -34,15 +34,15 @@ classifiers = [
     "Topic :: Software Development :: Libraries :: Python Modules",
 ]
 include = [
-    {path = "iotdb/thrift/*", format = "wheel"},
-    {path = "iotdb/thrift/common/*", format = "wheel"},
-    {path = "iotdb/thrift/confignode/*", format = "wheel"},
-    {path = "iotdb/thrift/datanode/*", format = "wheel"},
-    {path = "iotdb/thrift/ainode/*", format = "wheel"},
-    {path = "iotdb/conf/*", format = "wheel"},
+    {path = "ainode/thrift/*", format = "wheel"},
+    {path = "ainode/thrift/common/*", format = "wheel"},
+    {path = "ainode/thrift/confignode/*", format = "wheel"},
+    {path = "ainode/thrift/datanode/*", format = "wheel"},
+    {path = "ainode/thrift/ainode/*", format = "wheel"},
+    {path = "ainode/conf/*", format = "wheel"},
 ]
 packages = [
-    { include = "iotdb" }
+    { include = "ainode" }
 ]
 
 [tool.poetry.dependencies]
@@ -53,7 +53,7 @@ pandas = "^1.3.5"
 torch = ">=2.2.0"
 pylru = "^1.2.1"
 
-thrift = "^0.13.0"
+thrift = ">=0.14.0"
 dynaconf = "^3.1.11"
 requests = "^2.31.0"
 optuna = "^3.2.0"
@@ -63,4 +63,4 @@ pmdarima = "^2.0.4"
 hmmlearn = "^0.3.0"
 
 [tool.poetry.scripts]
-ainode = "iotdb.ainode.script:main"
\ No newline at end of file
+ainode = "ainode.core.script:main"
\ No newline at end of file
diff --git a/iotdb-core/ainode/resources/conf/iotdb-ainode.properties b/iotdb-core/ainode/resources/conf/iotdb-ainode.properties
index 2b208e72125..207bfcf9bef 100644
--- a/iotdb-core/ainode/resources/conf/iotdb-ainode.properties
+++ b/iotdb-core/ainode/resources/conf/iotdb-ainode.properties
@@ -40,19 +40,19 @@
 # The starting directory of the relative path is related to the operating system.
 # It is recommended to use an absolute path.
 # Datatype: String
-# ain_system_dir=data/ainode/system
+# ain_system_dir=data/core/system
 
 # The path where AINode stores model files
 # The starting directory of the relative path is related to the operating system.
 # It is recommended to use an absolute path.
 # Datatype: String
-# ain_models_dir=data/ainode/models
+# ain_models_dir=data/core/models
 
 # The path where AINode stores logs
 # The starting directory of the relative path is related to the operating system.
 # It is recommended to use an absolute path.
 # Datatype: String
-# ain_logs_dir=logs/ainode
+# ain_logs_dir=logs/core
 
 # Whether to use compression in Thrift
 # Please use 0 or 1
diff --git a/iotdb-core/ainode/resources/sbin/remove-ainode.sh b/iotdb-core/ainode/resources/sbin/remove-ainode.sh
index 2a27661460d..24074f7064a 100755
--- a/iotdb-core/ainode/resources/sbin/remove-ainode.sh
+++ b/iotdb-core/ainode/resources/sbin/remove-ainode.sh
@@ -74,7 +74,7 @@ else
   ain_interpreter_dir="$p_ain_interpreter_dir"
 fi
 
-# If ain_system_dir is empty, set default value to ../data/ainode/system
+# If ain_system_dir is empty, set default value to ../data/core/system
 if [ -z "$ain_system_dir" ]
 then
   ain_system_dir="$SCRIPT_DIR/../data/ainode/system"
@@ -89,7 +89,7 @@ fi
 
 # Change the working directory to the parent directory
 cd "$SCRIPT_DIR/.."
-ain_ainode_dir=$(dirname "$ain_interpreter_dir")/ainode
+ain_ainode_dir=$(dirname "$ain_interpreter_dir")/core
 
 
 if [ -z "$p_ain_remove_target" ]; then
diff --git a/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/manager/ModelManager.java b/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/manager/ModelManager.java
index ad19be00041..8a98d473a86 100644
--- a/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/manager/ModelManager.java
+++ b/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/manager/ModelManager.java
@@ -153,11 +153,11 @@ public class ModelManager {
       modelInformation.updateStatus(ModelStatus.values()[req.getModelStatus()]);
       modelInformation.setAttribute(req.getAttributes());
       modelInformation.setInputColumnSize(1);
-      if (req.getOutputShapesSize() > 0) {
-        modelInformation.setOutputLength(req.getOutputShapes().get(0));
+      if (req.isSetOutputLength()) {
+        modelInformation.setOutputLength(req.getOutputLength());
       }
-      if (req.getInputShapesSize() > 0) {
-        modelInformation.setInputLength(req.getInputShapes().get(0));
+      if (req.isSetInputLength()) {
+        modelInformation.setInputLength(req.getInputLength());
       }
       UpdateModelInfoPlan updateModelInfoPlan =
           new UpdateModelInfoPlan(req.getModelId(), modelInformation);
diff --git a/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/manager/ProcedureManager.java b/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/manager/ProcedureManager.java
index 9cdb6af607b..71ca36612c2 100644
--- a/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/manager/ProcedureManager.java
+++ b/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/manager/ProcedureManager.java
@@ -1292,8 +1292,9 @@ public class ProcedureManager {
             new CreateTrainingProcedure(
                 req.modelId,
                 req.curDatabase,
-                req.targetTables == null ? Collections.emptyList() : req.targetTables,
-                req.targetDbs == null ? Collections.emptyList() : req.targetDbs,
+                req.isSetTargetTables() ? req.getTargetTables() : Collections.emptyList(),
+                req.isSetTargetDbs() ? req.targetDbs : Collections.emptyList(),
+                req.isSetParameters() ? req.getParameters() : Collections.emptyMap(),
                 req.useAllData));
     LOGGER.info("CreateModelProcedure was submitted, procedureId: {}.", procedureId);
     return RpcUtils.SUCCESS_STATUS;
diff --git a/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/procedure/impl/model/CreateTrainingProcedure.java b/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/procedure/impl/model/CreateTrainingProcedure.java
index 5257a86ffe5..dbd84ddf1b7 100644
--- a/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/procedure/impl/model/CreateTrainingProcedure.java
+++ b/iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/procedure/impl/model/CreateTrainingProcedure.java
@@ -34,6 +34,7 @@ import org.slf4j.LoggerFactory;
 import java.io.IOException;
 import java.util.ArrayList;
 import java.util.List;
+import java.util.Map;
 
 public class CreateTrainingProcedure extends AbstractNodeProcedure<CreateTrainingState> {
 
@@ -44,6 +45,7 @@ public class CreateTrainingProcedure extends AbstractNodeProcedure<CreateTrainin
   private List<String> targetTables;
   private List<String> targetDbs;
   private boolean useAllData;
+  private Map<String, String> parameters;
 
   private static final IClientManager<ConfigRegionId, ConfigNodeClient> CONFIG_NODE_CLIENT_MANAGER =
       ConfigNodeClientManager.getInstance();
@@ -53,12 +55,14 @@ public class CreateTrainingProcedure extends AbstractNodeProcedure<CreateTrainin
       String curDatabase,
       List<String> targetTables,
       List<String> targetDbs,
+      Map<String, String> parameters,
       boolean useAllData) {
     this.modelId = modelId;
     this.curDatabase = curDatabase;
     this.targetTables = targetTables;
     this.targetDbs = targetDbs;
     this.useAllData = useAllData;
+    this.parameters = parameters;
   }
 
   public CreateTrainingProcedure() {
@@ -77,9 +81,9 @@ public class CreateTrainingProcedure extends AbstractNodeProcedure<CreateTrainin
       //                trainingReq.setExistingModelId(createTraining.getExistingModelId());
       //            }
 
-      //            if (createTraining.getParameters() != null) {
-      //                trainingReq.setParameters(createTraining.getParameters());
-      //            }
+      if (!parameters.isEmpty()) {
+        trainingReq.setParameters(parameters);
+      }
 
       List<ITableSchema> tableSchemaList = new ArrayList<>();
 
@@ -111,7 +115,6 @@ public class CreateTrainingProcedure extends AbstractNodeProcedure<CreateTrainin
         tableSchemaList.add(new ITableSchema(curDatabase, tableName));
       }
       trainingReq.setTargetTables(tableSchemaList);
-
       try (AINodeClient client =
           AINodeClientManager.getInstance().borrowClient(AINodeInfo.endPoint)) {
         status = client.createTrainingTask(trainingReq);
diff --git a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/Coordinator.java b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/Coordinator.java
index cbabf929375..d8e5e23f6dd 100644
--- a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/Coordinator.java
+++ b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/Coordinator.java
@@ -63,8 +63,8 @@ import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.AlterDB;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ClearCache;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.CreateDB;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.CreateFunction;
-import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.CreateTraining;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.CreateTable;
+import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.CreateTraining;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.DeleteDevice;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.DescribeTable;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.DropColumn;
@@ -97,6 +97,7 @@ import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowCurrentUser;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowDB;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowDataNodes;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowFunctions;
+import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowModels;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowRegions;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowTables;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowVariables;
@@ -441,6 +442,7 @@ public class Coordinator {
         || statement instanceof ReconstructRegion
         || statement instanceof ExtendRegion
         || statement instanceof CreateTraining
+        || statement instanceof ShowModels
         || statement instanceof RemoveRegion) {
       return new ConfigExecution(
           queryContext,
diff --git a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/execution/config/TableConfigTaskVisitor.java b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/execution/config/TableConfigTaskVisitor.java
index cb0c2a53166..655bb77ad16 100644
--- a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/execution/config/TableConfigTaskVisitor.java
+++ b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/execution/config/TableConfigTaskVisitor.java
@@ -46,6 +46,7 @@ import org.apache.iotdb.db.queryengine.plan.execution.config.metadata.ShowPipePl
 import org.apache.iotdb.db.queryengine.plan.execution.config.metadata.ShowRegionTask;
 import org.apache.iotdb.db.queryengine.plan.execution.config.metadata.ShowVariablesTask;
 import org.apache.iotdb.db.queryengine.plan.execution.config.metadata.ai.CreateTrainingTask;
+import org.apache.iotdb.db.queryengine.plan.execution.config.metadata.ai.ShowModelsTask;
 import org.apache.iotdb.db.queryengine.plan.execution.config.metadata.region.ExtendRegionTask;
 import org.apache.iotdb.db.queryengine.plan.execution.config.metadata.region.MigrateRegionTask;
 import org.apache.iotdb.db.queryengine.plan.execution.config.metadata.region.ReconstructRegionTask;
@@ -158,6 +159,7 @@ import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowCurrentUser;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowDB;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowDataNodes;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowFunctions;
+import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowModels;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowPipePlugins;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowPipes;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowRegions;
@@ -1095,4 +1097,9 @@ public class TableConfigTaskVisitor extends AstVisitor<IConfigTask, MPPQueryCont
     node.setCurDatabase(curDatabase);
     return new CreateTrainingTask(node);
   }
+
+  @Override
+  protected IConfigTask visitShowModels(ShowModels node, MPPQueryContext context) {
+    return new ShowModelsTask(node.getModelId());
+  }
 }
diff --git a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/execution/config/executor/ClusterConfigTaskExecutor.java b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/execution/config/executor/ClusterConfigTaskExecutor.java
index 42010e3f33c..4583ee9c525 100644
--- a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/execution/config/executor/ClusterConfigTaskExecutor.java
+++ b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/execution/config/executor/ClusterConfigTaskExecutor.java
@@ -3147,6 +3147,7 @@ public class ClusterConfigTaskExecutor implements IConfigTaskExecutor {
         CONFIG_NODE_CLIENT_MANAGER.borrowClient(ConfigNodeInfo.CONFIG_REGION_ID)) {
       final TCreateTrainingReq req =
           new TCreateTrainingReq(createTraining.getModelId(), createTraining.getCurDatabase());
+      req.setParameters(createTraining.getParameters());
       req.setTargetDbs(createTraining.getTargetDbs());
       req.setUseAllData(createTraining.isUseAllData());
       req.setTargetTables(
diff --git a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/ast/AstVisitor.java b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/ast/AstVisitor.java
index 3f3e2562fca..03456444001 100644
--- a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/ast/AstVisitor.java
+++ b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/ast/AstVisitor.java
@@ -672,4 +672,8 @@ public abstract class AstVisitor<R, C> {
   protected R visitCreateTraining(CreateTraining node, C context) {
     return visitStatement(node, context);
   }
+
+  protected R visitShowModels(ShowModels node, C context) {
+    return visitStatement(node, context);
+  }
 }
diff --git a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/ast/ShowModels.java b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/ast/ShowModels.java
new file mode 100644
index 00000000000..a3b803636e0
--- /dev/null
+++ b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/ast/ShowModels.java
@@ -0,0 +1,55 @@
+package org.apache.iotdb.db.queryengine.plan.relational.sql.ast;
+
+import com.google.common.collect.ImmutableList;
+
+import java.util.List;
+
+import static com.google.common.base.MoreObjects.toStringHelper;
+
+public class ShowModels extends Statement {
+
+  private String modelId = null;
+
+  public ShowModels() {
+    super(null);
+  }
+
+  @Override
+  public List<? extends Node> getChildren() {
+    return ImmutableList.of();
+  }
+
+  public void setModelId(String modelId) {
+    this.modelId = modelId;
+  }
+
+  public String getModelId() {
+    return modelId;
+  }
+
+  @Override
+  public <R, C> R accept(AstVisitor<R, C> visitor, C context) {
+    return visitor.visitShowModels(this, context);
+  }
+
+  @Override
+  public int hashCode() {
+    return 0;
+  }
+
+  @Override
+  public boolean equals(Object obj) {
+    if (!(obj instanceof ShowModels)) {
+      return false;
+    }
+    if (modelId == null) {
+      return ((ShowModels) obj).getModelId() == null;
+    }
+    return modelId.equals(((ShowModels) obj).getModelId());
+  }
+
+  @Override
+  public String toString() {
+    return toStringHelper(this).toString();
+  }
+}
diff --git a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/parser/AstBuilder.java b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/parser/AstBuilder.java
index 820c54c3bdf..408a1192104 100644
--- a/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/parser/AstBuilder.java
+++ b/iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/relational/sql/parser/AstBuilder.java
@@ -158,6 +158,7 @@ import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowDataNodes;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowDevice;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowFunctions;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowIndex;
+import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowModels;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowPipePlugins;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowPipes;
 import org.apache.iotdb.db.queryengine.plan.relational.sql.ast.ShowRegions;
@@ -2809,6 +2810,15 @@ public class AstBuilder extends RelationalSqlBaseVisitor<Node> {
     return createTraining;
   }
 
+  @Override
+  public Node visitShowModelsStatement(RelationalSqlParser.ShowModelsStatementContext ctx) {
+    ShowModels showModels = new ShowModels();
+    if (ctx.modelId != null) {
+      showModels.setModelId(ctx.modelId.getText());
+    }
+    return showModels;
+  }
+
   // ***************** arguments *****************
   @Override
   public Node visitGenericType(RelationalSqlParser.GenericTypeContext ctx) {
diff --git a/iotdb-core/relational-grammar/src/main/antlr4/org/apache/iotdb/db/relational/grammar/sql/RelationalSql.g4 b/iotdb-core/relational-grammar/src/main/antlr4/org/apache/iotdb/db/relational/grammar/sql/RelationalSql.g4
index 91a8456f925..272f8cd303b 100644
--- a/iotdb-core/relational-grammar/src/main/antlr4/org/apache/iotdb/db/relational/grammar/sql/RelationalSql.g4
+++ b/iotdb-core/relational-grammar/src/main/antlr4/org/apache/iotdb/db/relational/grammar/sql/RelationalSql.g4
@@ -149,6 +149,8 @@ statement
 
     // AI
     | createModelStatement
+    | showModelsStatement
+    | callInferenceStatement
 
     // View, Trigger, pipe, CQ, Quota are not supported yet
     ;
@@ -734,6 +736,15 @@ hparamPair
     : hparamKey=identifier '=' hyparamValue=primaryExpression
     ;
 
+showModelsStatement
+    : SHOW MODELS
+    | SHOW MODELS modelId=identifier
+    ;
+
+callInferenceStatement
+    : CALL INFERENCE '(' modelId=identifier COMMA inputSql=string (COMMA hparamPair)* ')'
+    ;
+
 // ------------------------------------------- Query Statement ---------------------------------------------------------
 queryStatement
     : query                                                        #statementDefault
@@ -1135,7 +1146,7 @@ nonReserved
     | JSON
     | KEEP | KEY | KEYS | KILL
     | LANGUAGE | LAST | LATERAL | LEADING | LEAVE | LEVEL | LIMIT | LINEAR | LOAD | LOCAL | LOGICAL | LOOP
-    | MAP | MATCH | MATCHED | MATCHES | MATCH_RECOGNIZE | MATERIALIZED | MEASURES | METHOD | MERGE | MICROSECOND | MIGRATE | MILLISECOND | MINUTE | MODEL | MODIFY | MONTH
+    | MAP | MATCH | MATCHED | MATCHES | MATCH_RECOGNIZE | MATERIALIZED | MEASURES | METHOD | MERGE | MICROSECOND | MIGRATE | MILLISECOND | MINUTE | MODEL | MODELS | MODIFY | MONTH
     | NANOSECOND | NESTED | NEXT | NFC | NFD | NFKC | NFKD | NO | NODEID | NONE | NULLIF | NULLS
     | OBJECT | OF | OFFSET | OMIT | ONE | ONLY | OPTION | ORDINALITY | OUTPUT | OVER | OVERFLOW
     | PARTITION | PARTITIONS | PASSING | PAST | PATH | PATTERN | PER | PERIOD | PERMUTE | PIPE | PIPEPLUGIN | PIPEPLUGINS | PIPES | PLAN | POSITION | PRECEDING | PRECISION | PRIVILEGES | PREVIOUS | PROCESSLIST | PROCESSOR | PROPERTIES | PRUNE
@@ -1293,6 +1304,7 @@ IGNORE: 'IGNORE';
 IMMEDIATE: 'IMMEDIATE';
 IN: 'IN';
 INCLUDING: 'INCLUDING';
+INFERENCE: 'INFERENCE';
 INITIAL: 'INITIAL';
 INNER: 'INNER';
 INPUT: 'INPUT';
@@ -1352,6 +1364,7 @@ MIGRATE: 'MIGRATE';
 MILLISECOND: 'MS';
 MINUTE: 'MINUTE' | 'M';
 MODEL: 'MODEL';
+MODELS: 'MODELS';
 MODIFY: 'MODIFY';
 MONTH: 'MONTH' | 'MO';
 NANOSECOND: 'NS';
diff --git a/iotdb-protocol/thrift-confignode/src/main/thrift/confignode.thrift b/iotdb-protocol/thrift-confignode/src/main/thrift/confignode.thrift
index 9b7e225fb84..8eca91b428d 100644
--- a/iotdb-protocol/thrift-confignode/src/main/thrift/confignode.thrift
+++ b/iotdb-protocol/thrift-confignode/src/main/thrift/confignode.thrift
@@ -1040,8 +1040,8 @@ struct TUpdateModelInfoReq {
     2: required i32 modelStatus
     3: optional string attributes
     4: optional list<i32> aiNodeIds
-    5: optional list<i32> inputShapes
-    6: optional list<i32> outputShapes
+    5: optional i32 inputLength
+    6: optional i32 outputLength
 }
 
 struct TCreateTrainingReq {
@@ -1050,6 +1050,7 @@ struct TCreateTrainingReq {
     3: optional list<string> targetTables
     4: optional list<string> targetDbs
     5: optional bool useAllData
+    6: optional map<string, string> parameters
 }
 
 // ====================================================