correctly filter default model

clients/python/src/jamaibase/protocol.py

@@ -1719,46 +1719,27 @@ class ActionTableSchemaCreate(TableSchemaCreate):
 
 
 class AddActionColumnSchema(ActionTableSchemaCreate):
+    # TODO: Deprecate this
     pass
 
 
 class KnowledgeTableSchemaCreate(TableSchemaCreate):
+    # TODO: Maybe deprecate this and use EmbedGenConfig instead ?
     embedding_model: str
 
-    @model_validator(mode="after")
-    def check_cols(self) -> Self:
-        super().check_cols()
-        num_text_cols = sum(c.id.lower() in ("text", "title", "file id") for c in self.cols)
-        if num_text_cols != 0:
-            raise ValueError("Schema cannot contain column names: 'Text', 'Title', 'File ID'.")
-        return self
-
 
 class AddKnowledgeColumnSchema(TableSchemaCreate):
-    @model_validator(mode="after")
-    def check_cols(self) -> Self:
-        super().check_cols()
-        num_text_cols = sum(c.id.lower() in ("text", "title", "file id") for c in self.cols)
-        if num_text_cols != 0:
-            raise ValueError("Schema cannot contain column names: 'Text', 'Title', 'File ID'.")
-        return self
+    # TODO: Deprecate this
+    pass
 
 
 class ChatTableSchemaCreate(TableSchemaCreate):
-    @model_validator(mode="after")
-    def check_cols(self) -> Self:
-        super().check_cols()
-        num_text_cols = sum(c.id.lower() in ("user", "ai") for c in self.cols)
-        if num_text_cols != 2:
-            raise ValueError("Schema must contain column names: 'User' and 'AI'.")
-        return self
+    pass
 
 
 class AddChatColumnSchema(TableSchemaCreate):
-    @model_validator(mode="after")
-    def check_cols(self) -> Self:
-        super().check_cols()
-        return self
+    # TODO: Deprecate this
+    pass
 
 
 class TableMeta(TableBase):

clients/python/tests/oss/gen_table/test_table_ops.py

@@ -21,6 +21,7 @@
     "str": '"Arrival" is a 2016 science fiction film. "Arrival" è un film di fantascienza del 2016. 「Arrival」は2016年のSF映画です。',
 }
 KT_FIXED_COLUMN_IDS = ["Title", "Title Embed", "Text", "Text Embed", "File ID"]
+CT_FIXED_COLUMN_IDS = ["User"]
 
 TABLE_ID_A = "table_a"
 TABLE_ID_B = "table_b"
@@ -1430,6 +1431,21 @@ def test_kt_drop_invalid_columns(client_cls: Type[JamAI]):
                 )
 
 
+@flaky(max_runs=5, min_passes=1, rerun_filter=_rerun_on_fs_error_with_delay)
+@pytest.mark.parametrize("client_cls", CLIENT_CLS)
+def test_ct_drop_invalid_columns(client_cls: Type[JamAI]):
+    table_type = "chat"
+    jamai = client_cls()
+    with _create_table(jamai, table_type) as table:
+        assert isinstance(table, p.TableMetaResponse)
+        for col in CT_FIXED_COLUMN_IDS:
+            with pytest.raises(RuntimeError):
+                jamai.table.drop_columns(
+                    table_type,
+                    p.ColumnDropRequest(table_id=table.id, column_names=[col]),
+                )
+
+
 @flaky(max_runs=5, min_passes=1, rerun_filter=_rerun_on_fs_error_with_delay)
 @pytest.mark.parametrize("client_cls", CLIENT_CLS)
 @pytest.mark.parametrize("table_type", TABLE_TYPES)
@@ -1450,7 +1466,7 @@ def test_rename_columns(
         assert isinstance(table, p.TableMetaResponse)
         assert all(isinstance(c, p.ColumnSchema) for c in table.cols)
         # Test rename on empty table
-        table = jamai.rename_columns(
+        table = jamai.table.rename_columns(
             table_type,
             p.ColumnRenameRequest(table_id=table.id, column_map=dict(y="z")),
         )
@@ -1475,7 +1491,7 @@ def test_rename_columns(
         _add_row(jamai, table_type, False, data=dict(x="True", z="<dummy>"))
         # Test rename table with data
         # Test also auto gen config reference update
-        table = jamai.rename_columns(
+        table = jamai.table.rename_columns(
             table_type,
             p.ColumnRenameRequest(table_id=table.id, column_map=dict(x="a")),
         )
@@ -1503,14 +1519,14 @@ def test_rename_columns(
 
         # Repeated new column names
         with pytest.raises(RuntimeError):
-            jamai.rename_columns(
+            jamai.table.rename_columns(
                 table_type,
                 p.ColumnRenameRequest(table_id=table.id, column_map=dict(a="b", z="b")),
             )
 
         # Overlapping new and old column names
         with pytest.raises(RuntimeError):
-            jamai.rename_columns(
+            jamai.table.rename_columns(
                 table_type,
                 p.ColumnRenameRequest(table_id=table.id, column_map=dict(a="b", z="a")),
             )
@@ -1525,7 +1541,22 @@ def test_kt_rename_invalid_columns(client_cls: Type[JamAI]):
         assert isinstance(table, p.TableMetaResponse)
         for col in KT_FIXED_COLUMN_IDS:
             with pytest.raises(RuntimeError):
-                jamai.rename_columns(
+                jamai.table.rename_columns(
+                    table_type,
+                    p.ColumnRenameRequest(table_id=table.id, column_map={col: col}),
+                )
+
+
+@flaky(max_runs=5, min_passes=1, rerun_filter=_rerun_on_fs_error_with_delay)
+@pytest.mark.parametrize("client_cls", CLIENT_CLS)
+def test_ct_rename_invalid_columns(client_cls: Type[JamAI]):
+    table_type = "chat"
+    jamai = client_cls()
+    with _create_table(jamai, table_type) as table:
+        assert isinstance(table, p.TableMetaResponse)
+        for col in CT_FIXED_COLUMN_IDS:
+            with pytest.raises(RuntimeError):
+                jamai.table.rename_columns(
                     table_type,
                     p.ColumnRenameRequest(table_id=table.id, column_map={col: col}),
                 )
@@ -1582,7 +1613,7 @@ def test_reorder_columns(
         cols = [c.id for c in table.cols]
         assert cols == expected_order, cols
         # Test reorder empty table
-        table = jamai.reorder_columns(
+        table = jamai.table.reorder_columns(
             table_type,
             p.ColumnReorderRequest(table_id=TABLE_ID_A, column_names=column_names),
         )
@@ -1692,7 +1723,7 @@ def test_reorder_columns_invalid(
         else:
             raise ValueError(f"Invalid table type: {table_type}")
         with pytest.raises(RuntimeError, match="referenced an invalid source column"):
-            jamai.reorder_columns(
+            jamai.table.reorder_columns(
                 table_type,
                 p.ColumnReorderRequest(table_id=TABLE_ID_A, column_names=column_names),
             )

scripts/remove_cloud_modules.ps1

@@ -19,3 +19,4 @@ function quiet_rm($item)
 quiet_rm "services/app/ecosystem.config.cjs"
 quiet_rm "services/appecosystem.json"
 quiet_rm ".github/workflows/trigger-push-gh-image.yml"
+quiet_rm ".github/workflows/ci.cloud.yml"

scripts/remove_cloud_modules.sh

@@ -9,3 +9,4 @@ find . -type d -name "(cloud)" -exec rm -rf {} +
 rm -f services/app/ecosystem.config.cjs
 rm -f services/app/ecosystem.json
 rm -f .github/workflows/trigger-push-gh-image.yml
+rm -f .github/workflows/ci.cloud.yml

services/api/pyproject.toml

@@ -132,7 +132,7 @@ dependencies = [
     "sqlmodel~=0.0.21",
     "srsly~=2.4.8",
     # starlette 0.38.3 and 0.38.4 seem to have issues with background tasks
-    "starlette==0.38.2",
+    "starlette~=0.41.3",
     "stripe~=9.12.0",
     "tantivy~=0.22.0",
     "tenacity~=8.5.0",

services/api/src/owl/configs/models.json

@@ -48,7 +48,7 @@
       "id": "ellm/BAAI/bge-small-en-v1.5",
       "name": "ELLM BAAI BGE Small EN v1.5",
       "context_length": 512,
-      "embedding_size": 1024,
+      "embedding_size": 384,
       "languages": ["mul"],
       "capabilities": ["embed"],
       "deployments": [

services/api/src/owl/configs/models_aipc.json

@@ -132,7 +132,7 @@
       "id": "ellm/BAAI/bge-small-en-v1.5",
       "name": "ELLM BAAI BGE Small EN v1.5",
       "context_length": 512,
-      "embedding_size": 1024,
+      "embedding_size": 384,
       "languages": ["mul"],
       "capabilities": ["embed"],
       "deployments": [

services/api/src/owl/configs/models_ollama.json

@@ -62,7 +62,7 @@
       "id": "ellm/BAAI/bge-small-en-v1.5",
       "name": "ELLM BAAI BGE Small EN v1.5",
       "context_length": 512,
-      "embedding_size": 1024,
+      "embedding_size": 384,
       "languages": ["mul"],
       "capabilities": ["embed"],
       "deployments": [