Catalog Tables

Catalog tables provide metadata about the machine learning models and Python modules created within AISQream, allowing users to query and understand the properties of these stored objects.

Catalog Table: ml_models

Contains all ML models with general common information:

–# DDL

SELECT get_ddl(sqream_catalog.ml_models);

create table "master"."sqream_catalog"."ml_models" (
   "model_id" bigint not null check ('CS ""flat""'),
   "database" text(32) not null check ('CS ""flat""'),
   "schema" text(32) not null check ('CS ""flat""'),
   "name" text(32) not null check ('CS ""flat""'),
   "ml_alg" text(32) not null check ('CS ""flat""')
);

–# ml_models

SELECT * FROM sqream_catalog.ml_models;

10,master,public,m1,LINEAR_REG
11,master,public,ml_xgb_catalog_test,XGBOOST
8,master,s,m,LINEAR_REG
9,master,public,m,PY_MODEL

Catalog table: ml_linear_reg

Contains all Linear Regression models, with more detailed information specifically for Linear Regression:

–# DDL

SELECT get_ddl(sqream_catalog.ml_linear_reg);

create table "master"."sqream_catalog"."ml_linear_reg" (
   "model_id" bigint not null check ('CS ""flat""'),
   "database" text(32) not null check ('CS ""flat""'),
   "schema" text(32) not null check ('CS ""flat""'),
   "name" text(32) not null check ('CS ""flat""'),
   "ml_alg" text(32) not null check ('CS ""flat""'),
   "initial_algorithm" text(32) not null check ('CS ""flat""'),
   "gd_optimizer" text(32) not null check ('CS ""flat""'),
   "coefficients" text(32) not null check ('CS ""flat""'),
   "is_standardized" bool not null check ('CS ""flat""'),
   "learning_rate" real not null check ('CS ""flat""'),
   "epoch_count" int not null check ('CS ""flat""'),
   "loss_function" text(32) not null check ('CS ""flat""'),
   "tolerance" real not null check ('CS ""flat""'),
   "time" text(32) not null check ('CS ""flat""')
);

–# ml_linear_reg

SELECT * FROM sqream_catalog.ml_linear_reg;

10,master,public,m1,LINEAR_REG,SVD,ADAM,{0.500000|0.500000},1,0.0000,1,MSE,0.0000,2025-02-06 15:10:55.150
8,master,s,m,LINEAR_REG,SVD,ADAM,{0.500000|0.500000},1,0.0000,1,MSE,0.0000,2025-02-06 15:10:42.121

Catalog table: ml_xgboost

Contains all XGboost models, with more detailed information specifically for XGboost:

–# DDL

SELECT get_ddl(sqream_catalog.ml_xgboost);

create table "master"."sqream_catalog"."ml_xgboost" (
   "model_id" bigint not null check ('CS ""flat""'),
   "database" text not null check ('CS ""flat""'),
   "schema" text not null check ('CS ""flat""'),
   "name" text not null check ('CS ""flat""'),
   "ml_alg" text not null check ('CS ""flat""'),
   "verbosity" text not null check ('CS ""flat""'),
   "booster" text not null check ('CS ""flat""'),
   "disable_default_eval_metric" text not null check ('CS ""flat""'),
   "tree_booster_parameters_eta" text not null check ('CS ""flat""'),
   "tree_booster_parameters_gamma" text not null check ('CS ""flat""'),
   "tree_booster_parameters_max_depth" text not null check ('CS ""flat""'),
   "tree_booster_parameters_min_child_weight" text not null check ('CS ""flat""'),
   "tree_booster_parameters_max_delta_step" text not null check ('CS ""flat""'),
   "tree_booster_parameters_subsample" text not null check ('CS ""flat""'),
   "tree_booster_parameters_sampling_method" text not null check ('CS ""flat""'),
   "tree_booster_parameters_colsample_bytree" text not null check ('CS ""flat""'),
   "tree_booster_parameters_colsample_bylevel" text not null check ('CS ""flat""'),
   "tree_booster_parameters_colsample_bynode" text not null check ('CS ""flat""'),
   "tree_booster_parameters_lambda" text not null check ('CS ""flat""'),
   "tree_booster_parameters_reg_lambda" text not null check ('CS ""flat""'),
   "tree_booster_parameters_alpha" text not null check ('CS ""flat""'),
   "tree_booster_parameters_reg_alpha" text not null check ('CS ""flat""'),
   "tree_booster_parameters_tree_method" text not null check ('CS ""flat""'),
   "tree_booster_parameters_scale_pos_weight" text not null check ('CS ""flat""'),
   "tree_booster_parameters_process_type" text not null check ('CS ""flat""'),
   "tree_booster_parameters_grow_policy" text not null check ('CS ""flat""'),
   "tree_booster_parameters_max_leaves" text not null check ('CS ""flat""'),
   "tree_booster_parameters_max_bin" text not null check ('CS ""flat""'),
   "tree_booster_parameters_num_parallel_tree" text not null check ('CS ""flat""'),
   "dart_booster_params_sample_type" text not null check ('CS ""flat""'),
   "dart_booster_params_normalize_type" text not null check ('CS ""flat""'),
   "dart_booster_params_rate_drop" text not null check ('CS ""flat""'),
   "dart_booster_params_one_drop" text not null check ('CS ""flat""'),
   "dart_booster_params_skip_drop" text not null check ('CS ""flat""'),
   "linear_booster_params_updater" text not null check ('CS ""flat""'),
   "linear_booster_params_feature_selector" text not null check ('CS ""flat""'),
   "linear_booster_params_top_k" text not null check ('CS ""flat""'),
   "objective" text not null check ('CS ""flat""'),
   "base_score" text not null check ('CS ""flat""'),
   "eval_metric" text not null check ('CS ""flat""'),
   "cut_off_top_positions" text not null check ('CS ""flat""'),
   "seed" text not null check ('CS ""flat""'),
   "seed_per_iteration" text not null check ('CS ""flat""'),
   "tweedie_variance_power" text not null check ('CS ""flat""'),
   "huber_slope" text not null check ('CS ""flat""'),
   "quantile_alpha" text not null check ('CS ""flat""'),
   "aft_loss_distribution" text not null check ('CS ""flat""'),
   "lambdarank_pair_method" text not null check ('CS ""flat""'),
   "lambdarank_num_pair_per_sample" text not null check ('CS ""flat""'),
   "lambdarank_normalization" text not null check ('CS ""flat""'),
   "lambdarank_bias_norm" text not null check ('CS ""flat""'),
   "lambdarank_unbiased" text not null check ('CS ""flat""'),
   "ndcg_exp_gain" text not null check ('CS ""flat""'),
   "epoch_count" text not null check ('CS ""flat""')
);

–# ml_xgboost

SELECT * FROM sqream_catalog.ml_xgboost;

1,master,public,m2,XGBOOST,VERBOSITY_SILENT,BOOSTER_GBTREE,0,0.300000,0.000000,6,1,0,1.000000,SAMPLING_METHOD_UNIFORM,1.000000,1.000000,1.000000,0.000000,0.000000,0.000000,0.000000,TREE_METHOD_AUTO,1.000000,PROCESS_TYPE_DEFAULT,GROW_POLICY_DEPTHWISE,0,0,1,SAMPLE_TYPE_UNSPECIFIED,NORMALIZE_TYPE_UNSPECIFIED,0.000000,0,0.000000,UPDATER_UNSPECIFIED,FEATURE_SELECTOR_UNSPECIFIED,0,OBJECTIVE_BINARY_LOGISTIC,0.000000,EVAL_METRIC_UNSPECIFIED,"""""",0,0,0.000000,0.000000,,AFT_LOSS_DISTRIBUTION_UNSPECIFIED,LAMBDARANK_PAIR_METHOD_UNSPECIFIED,0,0,0.000000,0,0,100

Catalog table: py_modules

Contains all Python modules:

–# DDL

SELECT get_ddl(sqream_catalog.py_modules);

"module_id" bigint not null check ('CS ""flat""'),   "database_name" text(32) not null check ('CS ""flat""'),   "module_name" text(32) not null check ('CS ""flat""'),   "path" text(32) not null check ('CS ""flat""'),   "entry_points" text(32) not null check ('CS ""flat""')

–# ml_linear_reg

SELECT * FROM sqream_catalog.py_modules;

22,master,pm,/home/sagib/py_udf.py,"{[name:func_name, args:[TEXT], ret_type:DATE, gpu:true]}"
36,master,m1,/home/sagib/py_udf.py,"{[name:return_input2, args:[TEXT], ret_type:TEXT, gpu:false]}"

Catalog table: py_module_permissions

Contains all Python modules permissions:

–# DDL

SELECT get_ddl(sqream_catalog.py_module_permissions);

database_name" text(32) not null check ('CS ""flat""'),   "module_id" bigint not null check ('CS ""flat""'),   "role_id" bigint not null check ('CS ""flat""'),   "permission_type" int not null check ('CS ""flat""')

–# py_module_permissions

SELECT * FROM sqream_catalog.py_module_permissions;

master,0,8,11000
Master,0,8,11001

Catalog table: registered_algorithms

Contains all Python based AI models registered:

–# DDL

SELECT get_ddl(sqream_catalog.registered_algorithms);

"id" bigint not null check ('CS ""flat""'),
  "database_name" text(32) not null check ('CS ""flat""'),
  "alg_name" text(32) not null check ('CS ""flat""'),
  "alg_path" text(32) not null check ('CS ""flat""'),
  "train_method" text(32) not null check ('CS ""flat""'),
  "predict_method" text(32) not null check ('CS ""flat""')

–# registered_algorithms

SELECT * FROM sqream_catalog.registered_algorithms;

6,master,logistic_regression,/tmp/sagib.py,train,predict