update download url and requirements for text model

gDreamcatcher · gDreamcatcher · commit 3511bd9b720b · 2020-07-28T17:02:01.000+08:00
diff --git a/.gitignore b/.gitignore
@@ -0,0 +1,9 @@
+# Binaries for programs and plugins
+*.swp
+__pycache__/
+
+
+#skip some test file
+model/
+src/log/
+
diff --git a/README.md b/README.md
@@ -45,7 +45,11 @@ Before inserting and searching, a database and space should be creating firstly.
 curl -XPUT -H "content-type:application/json" -d '{"name": "test"}' http://127.0.0.1:4101/db/_create
 
 # create a space in test db which name test too.
-curl -XPUT -H "content-type: application/json" -d' { "name": "test", "dynamic_schema": "strict", "partition_num": 2, "replica_num": 1, "engine": {"name":"gamma","metric_type": "InnerProduct"}, "properties": { "url": { "type": "keyword", "index": true}, "feature1": { "type": "vector", "dimension":512, "format": "normalization" }}} ' http://127.0.0.1:4101/space/test/_create
+## for image
+curl -XPUT -H "content-type: application/json" -d '{"name": "test", "partition_num": 1, "replica_num": 1, "engine": {"name": "gamma", "index_size": 70000, "max_size": 10000000, "id_type": "String", "retrieval_type": "IVFPQ", "retrieval_param": { "metric_type": "InnerProduct", "ncentroids": 256, "nsubvector": 32  } }, "properties": {  "url": { "type": "keyword", "index": true }, "feature1": { "type": "vector", "dimension":512, "format": "normalization"  } }}  }' http://127.0.0.1:4101/space/test/_create
+
+## for text dimension is 768
+curl -XPUT -H "content-type: application/json" -d '{"name": "test", "partition_num": 1, "replica_num": 1, "engine": {"name": "gamma", "index_size": 70000, "max_size": 10000000, "id_type": "String", "retrieval_type": "IVFPQ", "retrieval_param": { "metric_type": "InnerProduct", "ncentroids": 256, "nsubvector": 32  } }, "properties": {  "text": { "type": "keyword", "index": true }, "feature1": { "type": "vector", "dimension":768, "format": "normalization"  } }}  }' http://127.0.0.1:4101/space/test/_create
 ```
 
 A successful response looks like this:
@@ -86,8 +90,12 @@ The method of single import demo:
 
 ```shell
 # single insert
+
+## image
 curl -XPOST -H "content-type: application/json"  -d' { "url": "../images/COCO_val2014_000000123599.jpg", "feature1":{"feature":"../images/COCO_val2014_000000123599.jpg"}} ' http://127.0.0.1:4101/test/test/AW63W9I4JG6WicwQX_RC
 
+## text
+curl -XPOST -H "content-type: application/json"  -d' { "text": "感谢大家", "feature1":{"feature":"感谢大家"} } ' http://127.0.0.1:4101/test/test/AW63W9I4JG6WicwQX_RC'
 ```
 
 A successful response like this:
diff --git a/bin/run.sh b/bin/run.sh
@@ -2,7 +2,8 @@
 
 path=`pwd`
 cd $path/src
-pip install -r requirements.txt
+# pip install -r requirements.txt
+pip install -i https://pypi.tuna.tsinghua.edu.cn/simple -r requirements.txt
 
 run_image_retrieval(){
     python main.py --model_name=image_retrieval
diff --git a/src/text/requirements.txt b/src/text/requirements.txt
@@ -1,4 +1,3 @@
-Keras==2.2.4
-keras-bert==0.80.0
-tensorboard==1.15.0
-tensorflow-gpu==1.15.2
+keras==2.4.3
+keras-bert==0.85.0
+tensorflow-gpu==2.2.0
diff --git a/src/text/text.py b/src/text/text.py
@@ -17,7 +17,7 @@
 from keras_bert import extract_embeddings,get_checkpoint_paths,load_trained_model_from_checkpoint, load_vocabulary
 
 
-MODEL_URL = 'https://storage.googleapis.com/hfl-rc/chinese-bert/chinese_wwm_L-12_H-768_A-12.zip'
+MODEL_URL = 'https://storage.googleapis.com/bert_models/2018_11_03/chinese_L-12_H-768_A-12.zip'
 
 
 class Text(object):
@@ -28,7 +28,7 @@ def __init__(self, config):
             if not os.path.exists(model_dir):
                 os.makedirs(model_dir)
             subprocess.run(
-                f"wget -P {model_dir} {MODEL_URL} && cd {model_dir} && unzip chinese_wwm_L-12_H-768_A-12.zip",
+                f"wget -P {model_dir} {MODEL_URL} && cd {model_dir} && unzip chinese_L-12_H-768_A-12.zip",
                 shell=True)
 
         paths = get_checkpoint_paths(model_path)
