update target_config names

codelibs · May 17, 2024 · cdaea45 · cdaea45
1 parent 8506a42
commit cdaea45
Show file tree

Hide file tree

Showing 9 changed files with 39 additions and 37 deletions.
diff --git a/run-chroma.ipynb b/run-chroma.ipynb
@@ -69,7 +69,7 @@
     "\n",
     "def get_dataset_config(target_name):\n",
     "    setting = {\n",
-    "        \"100k-768-m32-ef100-ip\": {\n",
+    "        \"100k-768-m32-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -85,7 +85,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"\",\n",
     "        },\n",
-    "        \"1m-768-m49-ef100-ip\": {\n",
+    "        \"1m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -101,7 +101,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"\",\n",
     "        },\n",
-    "        \"5m-768-m49-ef100-ip\": {\n",
+    "        \"5m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -679,7 +679,7 @@
    "source": [
     "query_data = {\"section_values\": []}\n",
     "results = {}\n",
-    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-ef100-ip\")\n",
+    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-efc200-ef100-ip\")\n",
     "dataset_config = get_dataset_config(target_config)\n",
     "pprint.pprint(dataset_config)"
    ]

diff --git a/run-elasticsearch.ipynb b/run-elasticsearch.ipynb
@@ -56,7 +56,7 @@
     "\n",
     "def get_dataset_config(target_name):\n",
     "    setting = {\n",
-    "        \"100k-768-m32-ef100-ip\": {\n",
+    "        \"100k-768-m32-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -72,7 +72,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"int8\",\n",
     "        },\n",
-    "        \"1m-768-m49-ef100-ip\": {\n",
+    "        \"1m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -88,7 +88,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"int8\",\n",
     "        },\n",
-    "        \"5m-768-m49-ef100-ip\": {\n",
+    "        \"5m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -842,7 +842,7 @@
    "source": [
     "query_data = {\"section_values\": []}\n",
     "results = {}\n",
-    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-ef100-ip\")\n",
+    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-efc200-ef100-ip\")\n",
     "dataset_config = get_dataset_config(target_config)\n",
     "pprint.pprint(dataset_config)"
    ]

diff --git a/run-milvus.ipynb b/run-milvus.ipynb
@@ -69,7 +69,7 @@
     "\n",
     "def get_dataset_config(target_name):\n",
     "    setting = {\n",
-    "        \"100k-768-m32-ef100-ip\": {\n",
+    "        \"100k-768-m32-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -83,7 +83,7 @@
     "            \"hnsw_ef\": 100,\n",
     "            \"update_docs_per_sec\": 0,\n",
     "        },\n",
-    "        \"1m-768-m49-ef100-ip\": {\n",
+    "        \"1m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -97,7 +97,7 @@
     "            \"hnsw_ef\": 100,\n",
     "            \"update_docs_per_sec\": 0,\n",
     "        },\n",
-    "        \"5m-768-m49-ef100-ip\": {\n",
+    "        \"5m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -809,7 +809,7 @@
    "source": [
     "query_data = {\"section_values\": []}\n",
     "results = {}\n",
-    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-ef100-ip\")\n",
+    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-efc200-ef100-ip\")\n",
     "dataset_config = get_dataset_config(target_config)\n",
     "pprint.pprint(dataset_config)"
    ]

diff --git a/run-opensearch.ipynb b/run-opensearch.ipynb
@@ -56,7 +56,7 @@
     "\n",
     "def get_dataset_config(target_name):\n",
     "    setting = {\n",
-    "        \"100k-768-m32-ef100-ip\": {\n",
+    "        \"100k-768-m32-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -71,7 +71,7 @@
     "            \"hnsw_ef\": 100,\n",
     "            \"update_docs_per_sec\": 0,\n",
     "        },\n",
-    "        \"1m-768-m49-ef100-ip\": {\n",
+    "        \"1m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -86,7 +86,7 @@
     "            \"hnsw_ef\": 100,\n",
     "            \"update_docs_per_sec\": 0,\n",
     "        },\n",
-    "        \"5m-768-m49-ef100-ip\": {\n",
+    "        \"5m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -836,7 +836,7 @@
    "source": [
     "query_data = {\"section_values\": []}\n",
     "results = {}\n",
-    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-ef100-ip\")\n",
+    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-efc200-ef100-ip\")\n",
     "dataset_config = get_dataset_config(target_config)\n",
     "pprint.pprint(dataset_config)"
    ]

diff --git a/run-pgvector.ipynb b/run-pgvector.ipynb
@@ -75,7 +75,7 @@
     "\n",
     "def get_dataset_config(target_name):\n",
     "    setting = {\n",
-    "        \"100k-768-m32-ef100-ip\": {\n",
+    "        \"100k-768-m32-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -90,7 +90,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"halfvec\", # \"vector\",\n",
     "        },\n",
-    "        \"1m-768-m49-ef100-ip\": {\n",
+    "        \"1m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -105,7 +105,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"halfvec\", # \"vector\",\n",
     "        },\n",
-    "        \"5m-768-m49-ef100-ip\": {\n",
+    "        \"5m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -723,7 +723,7 @@
    "source": [
     "query_data = {\"section_values\": []}\n",
     "results = {}\n",
-    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-ef100-ip\")\n",
+    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-efc200-ef100-ip\")\n",
     "dataset_config = get_dataset_config(target_config)\n",
     "pprint.pprint(dataset_config)"
    ]

diff --git a/run-qdrant.ipynb b/run-qdrant.ipynb
@@ -55,7 +55,7 @@
     "\n",
     "def get_dataset_config(target_name):\n",
     "    setting = {\n",
-    "        \"100k-768-m32-ef100-ip\": {\n",
+    "        \"100k-768-m32-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -71,7 +71,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"int8\",\n",
     "        },\n",
-    "        \"1m-768-m49-ef100-ip\": {\n",
+    "        \"1m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -87,7 +87,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"int8\",\n",
     "        },\n",
-    "        \"5m-768-m49-ef100-ip\": {\n",
+    "        \"5m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -255,8 +255,12 @@
     "            \"distance\": config.distance,\n",
     "            \"hnsw_config\": {\n",
     "                \"m\": config.hnsw_m,\n",
-    "                \"ef_construction\": config.hnsw_ef_construction,\n",
+    "                \"ef_construct\": config.hnsw_ef_construction,\n",
     "            }\n",
+    "        },\n",
+    "        \"hnsw_config\": {\n",
+    "            \"m\": config.hnsw_m,\n",
+    "            \"ef_construct\": config.hnsw_ef_construction,\n",
     "        }\n",
     "    }\n",
     "    if config.quantization == \"int8\":\n",
@@ -729,7 +733,7 @@
    "source": [
     "query_data = {\"section_values\": []}\n",
     "results = {}\n",
-    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-ef100-ip\")\n",
+    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-efc200-ef100-ip\")\n",
     "dataset_config = get_dataset_config(target_config)\n",
     "pprint.pprint(dataset_config)"
    ]
@@ -776,9 +780,7 @@
    "cell_type": "code",
    "execution_count": null,
    "id": "bcbdfa51-5bdf-494f-a9aa-3f265b8ec20d",
-   "metadata": {
-    "scrolled": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "print_docker_container_stats(dataset_config)\n",

diff --git a/run-vespa.ipynb b/run-vespa.ipynb
@@ -67,7 +67,7 @@
     "\n",
     "def get_dataset_config(target_name):\n",
     "    setting = {\n",
-    "        \"100k-768-m32-ef100-ip\": {\n",
+    "        \"100k-768-m32-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -83,7 +83,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"float\", # \"bfloat16\",\n",
     "        },\n",
-    "        \"1m-768-m49-ef100-ip\": {\n",
+    "        \"1m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -99,7 +99,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"float\", # \"bfloat16\",\n",
     "        },\n",
-    "        \"5m-768-m49-ef100-ip\": {\n",
+    "        \"5m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -819,7 +819,7 @@
    "source": [
     "query_data = {\"section_values\": []}\n",
     "results = {}\n",
-    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-ef100-ip\")\n",
+    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-efc200-ef100-ip\")\n",
     "dataset_config = get_dataset_config(target_config)\n",
     "pprint.pprint(dataset_config)"
    ]

diff --git a/run-weaviate.ipynb b/run-weaviate.ipynb
@@ -54,7 +54,7 @@
     "\n",
     "def get_dataset_config(target_name):\n",
     "    setting = {\n",
-    "        \"100k-768-m32-ef100-ip\": {\n",
+    "        \"100k-768-m32-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -69,7 +69,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"none\", # \"pq\",\n",
     "        },\n",
-    "        \"1m-768-m49-ef100-ip\": {\n",
+    "        \"1m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -84,7 +84,7 @@
     "            \"update_docs_per_sec\": 0,\n",
     "            \"quantization\": \"none\", # \"pq\",\n",
     "        },\n",
-    "        \"5m-768-m49-ef100-ip\": {\n",
+    "        \"5m-768-m48-efc200-ef100-ip\": {\n",
     "            \"content_path\": Path(\"dataset/passages-c400-jawiki-20230403\"),\n",
     "            \"embedding_path\": Path(\"dataset/passages-c400-jawiki-20230403/multilingual-e5-base-passage\"),\n",
     "            \"num_of_docs\": 5555583,\n",
@@ -737,7 +737,7 @@
    "source": [
     "query_data = {\"section_values\": []}\n",
     "results = {}\n",
-    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-ef100-ip\")\n",
+    "target_config = os.getenv(\"TARGET_CONFIG\", \"100k-768-m32-efc200-ef100-ip\")\n",
     "dataset_config = get_dataset_config(target_config)\n",
     "pprint.pprint(dataset_config)"
    ]

diff --git a/scripts/setup.sh b/scripts/setup.sh
@@ -26,7 +26,7 @@ fi
 # wikipedia contents
 data_type=passages-c400-jawiki-20230403
 model_type=multilingual-e5-base-passage
-setting_type=100k-768-m32-ef100-ip
+setting_type=100k-768-m32-efc200-ef100-ip
 
 data_dir="${base_dir}/../dataset/${data_type}"
 output_dir="${base_dir}/../output"