zlei6/ragflow
1
0
Fork 0
mirror of https://github.com/infiniflow/ragflow.git synced 2026-08-05 23:24:05 +08:00
Code Issues Packages Projects Releases Wiki Activity

refa: simplify RAPTOR tree clustering and configuration (#17614)

Browse Source
This commit is contained in:
buua436
2026-08-03 17:46:50 +08:00
committed by GitHub
parent e290a0d476
commit 3e7cfbe052
32 changed files with 318 additions and 1356 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
test/testcases/test_http_api/test_dataset_management/test_create_dataset.py
View File

@@ -439,12 +439,12 @@ class TestDatasetCreate:
("raptor_true", {"raptor": {"use_raptor": True}}),
("raptor_false", {"raptor": {"use_raptor": False}}),
("raptor_prompt", {"raptor": {"prompt": "Who are you?"}}),
("raptor_max_token_min", {"raptor": {"max_token": 1}}),
("raptor_max_token_min", {"raptor": {"max_token": 512}}),
("raptor_max_token_mid", {"raptor": {"max_token": 1024}}),
("raptor_max_token_max", {"raptor": {"max_token": 2048}}),
("raptor_threshold_min", {"raptor": {"threshold": 0.0}}),
("raptor_threshold_mid", {"raptor": {"threshold": 0.5}}),
("raptor_threshold_max", {"raptor": {"threshold": 1.0}}),
("raptor_clustering_threshold_min", {"raptor": {"clustering_threshold": 0.0}}),
("raptor_clustering_threshold_mid", {"raptor": {"clustering_threshold": 0.5}}),
("raptor_clustering_threshold_max", {"raptor": {"clustering_threshold": 1.0}}),
("raptor_max_cluster_min", {"raptor": {"max_cluster": 1}}),
("raptor_max_cluster_mid", {"raptor": {"max_cluster": 512}}),
("raptor_max_cluster_max", {"raptor": {"max_cluster": 1024}}),
@@ -496,9 +496,9 @@ class TestDatasetCreate:
"raptor_max_token_min",
"raptor_max_token_mid",
"raptor_max_token_max",
"raptor_threshold_min",
"raptor_threshold_mid",
"raptor_threshold_max",
"raptor_clustering_threshold_min",
"raptor_clustering_threshold_mid",
"raptor_clustering_threshold_max",
"raptor_max_cluster_min",
"raptor_max_cluster_mid",
"raptor_max_cluster_max",
@@ -567,9 +567,9 @@ class TestDatasetCreate:
("raptor_max_token_max_limit", {"raptor": {"max_token": 2049}}, "Input should be less than or equal to 2048"),
("raptor_max_token_float_not_allowed", {"raptor": {"max_token": 3.14}}, "Input should be a valid integer"),
("raptor_max_token_type_invalid", {"raptor": {"max_token": "string"}}, "Input should be a valid integer"),
("raptor_threshold_min_limit", {"raptor": {"threshold": -0.1}}, "Input should be greater than or equal to 0"),
("raptor_threshold_max_limit", {"raptor": {"threshold": 1.1}}, "Input should be less than or equal to 1"),
("raptor_threshold_type_invalid", {"raptor": {"threshold": "string"}}, "Input should be a valid number"),
("raptor_clustering_threshold_min_limit", {"raptor": {"clustering_threshold": -0.1}}, "Input should be greater than or equal to 0"),
("raptor_clustering_threshold_max_limit", {"raptor": {"clustering_threshold": 1.1}}, "Input should be less than or equal to 1"),
("raptor_clustering_threshold_type_invalid", {"raptor": {"clustering_threshold": "string"}}, "Input should be a valid number"),
("raptor_max_cluster_min_limit", {"raptor": {"max_cluster": 0}}, "Input should be greater than or equal to 1"),
("raptor_max_cluster_max_limit", {"raptor": {"max_cluster": 1025}}, "Input should be less than or equal to 1024"),
("raptor_max_cluster_float_not_allowed", {"raptor": {"max_cluster": 3.14}}, "Input should be a valid integer"),
@@ -625,9 +625,9 @@ class TestDatasetCreate:
"raptor_max_token_max_limit",
"raptor_max_token_float_not_allowed",
"raptor_max_token_type_invalid",
"raptor_threshold_min_limit",
"raptor_threshold_max_limit",
"raptor_threshold_type_invalid",
"raptor_clustering_threshold_min_limit",
"raptor_clustering_threshold_max_limit",
"raptor_clustering_threshold_type_invalid",
"raptor_max_cluster_min_limit",
"raptor_max_cluster_max_limit",
"raptor_max_cluster_float_not_allowed",

42
test/testcases/test_http_api/test_dataset_management/test_update_dataset.py
View File

@@ -573,20 +573,16 @@ class TestDatasetUpdate:
{"raptor": {"use_raptor": True}},
{"raptor": {"use_raptor": False}},
{"raptor": {"prompt": "Who are you?"}},
{"raptor": {"max_token": 1}},
{"raptor": {"max_token": 512}},
{"raptor": {"max_token": 1024}},
{"raptor": {"max_token": 2048}},
{"raptor": {"threshold": 0.0}},
{"raptor": {"threshold": 0.5}},
{"raptor": {"threshold": 1.0}},
{"raptor": {"clustering_threshold": 0.0}},
{"raptor": {"clustering_threshold": 0.5}},
{"raptor": {"clustering_threshold": 1.0}},
{"raptor": {"max_cluster": 1}},
{"raptor": {"max_cluster": 512}},
{"raptor": {"max_cluster": 1024}},
{"raptor": {"random_seed": 0}},
{"raptor": {"clustering_method": "gmm"}},
{"raptor": {"clustering_method": "ahc"}},
{"raptor": {"tree_builder": "raptor"}},
{"raptor": {"tree_builder": "psi"}},
],
ids=[
"auto_keywords_min",
@@ -630,17 +626,13 @@ class TestDatasetUpdate:
"raptor_max_token_min",
"raptor_max_token_mid",
"raptor_max_token_max",
"raptor_threshold_min",
"raptor_threshold_mid",
"raptor_threshold_max",
"raptor_clustering_threshold_min",
"raptor_clustering_threshold_mid",
"raptor_clustering_threshold_max",
"raptor_max_cluster_min",
"raptor_max_cluster_mid",
"raptor_max_cluster_max",
"raptor_random_seed_min",
"raptor_clustering_method_gmm",
"raptor_clustering_method_ahc",
"raptor_tree_builder_raptor",
"raptor_tree_builder_psi",
],
)
def test_parser_config(self, HttpApiAuth, add_dataset_func, parser_config):
@@ -705,9 +697,9 @@ class TestDatasetUpdate:
({"raptor": {"max_token": 2049}}, "Input should be less than or equal to 2048"),
({"raptor": {"max_token": 3.14}}, "Input should be a valid integer"),
({"raptor": {"max_token": "string"}}, "Input should be a valid integer"),
({"raptor": {"threshold": -0.1}}, "Input should be greater than or equal to 0"),
({"raptor": {"threshold": 1.1}}, "Input should be less than or equal to 1"),
({"raptor": {"threshold": "string"}}, "Input should be a valid number"),
({"raptor": {"clustering_threshold": -0.1}}, "Input should be greater than or equal to 0"),
({"raptor": {"clustering_threshold": 1.1}}, "Input should be less than or equal to 1"),
({"raptor": {"clustering_threshold": "string"}}, "Input should be a valid number"),
({"raptor": {"max_cluster": 0}}, "Input should be greater than or equal to 1"),
({"raptor": {"max_cluster": 1025}}, "Input should be less than or equal to 1024"),
({"raptor": {"max_cluster": 3.14}}, "Input should be a valid integer"),
@@ -715,10 +707,6 @@ class TestDatasetUpdate:
({"raptor": {"random_seed": -1}}, "Input should be greater than or equal to 0"),
({"raptor": {"random_seed": 3.14}}, "Input should be a valid integer"),
({"raptor": {"random_seed": "string"}}, "Input should be a valid integer"),
({"raptor": {"clustering_method": "unknown"}}, "Input should be 'gmm' or 'ahc'"),
({"raptor": {"clustering_method": None}}, "Input should be 'gmm' or 'ahc'"),
({"raptor": {"tree_builder": "ahc"}}, "Input should be 'raptor' or 'psi'"),
({"raptor": {"tree_builder": None}}, "Input should be 'raptor' or 'psi'"),
({"delimiter": "a" * 65536}, "Parser config exceeds size limit (max 65,535 characters)"),
],
ids=[
@@ -765,9 +753,9 @@ class TestDatasetUpdate:
"raptor_max_token_max_limit",
"raptor_max_token_float_not_allowed",
"raptor_max_token_type_invalid",
"raptor_threshold_min_limit",
"raptor_threshold_max_limit",
"raptor_threshold_type_invalid",
"raptor_clustering_threshold_min_limit",
"raptor_clustering_threshold_max_limit",
"raptor_clustering_threshold_type_invalid",
"raptor_max_cluster_min_limit",
"raptor_max_cluster_max_limit",
"raptor_max_cluster_float_not_allowed",
@@ -775,10 +763,6 @@ class TestDatasetUpdate:
"raptor_random_seed_min_limit",
"raptor_random_seed_float_not_allowed",
"raptor_random_seed_type_invalid",
"raptor_clustering_method_invalid",
"raptor_clustering_method_none_invalid",
"raptor_tree_builder_invalid",
"raptor_tree_builder_none_invalid",
"parser_config_type_invalid",
],
)