[pull] master from CogStack:master

docker/docker-compose-gpu.yml

@@ -32,7 +32,43 @@ services:
           devices:
             - driver: nvidia
               count: 1
-              device_ids: ['0']
+              #device_ids: ['0']
+              capabilities: ["gpu", "utility", "compute", "video"]
+
+  nlp-medcat-service-production-1:
+    container_name: cogstack-medcat-service-production-1
+    ### Multiple images available:
+    ## default image, only CPU support: cogstacksystems/medcat-service:latest 
+    ## GPU support: cogstacksystems/medcat-service-gpu:latest
+    # image: cogstacksystems/medcat-service:latest
+    platform: linux
+    restart: always
+    ## Default dockerfile: ../Dockerfile
+    ## GPU dockerfile: ../Dockerfile_gpu
+    build:
+      context: ../
+      dockerfile: "Dockerfile_gpu"
+    environment:
+      - http_proxy=$HTTP_PROXY
+      - https_proxy=$HTTPS_PROXY
+      - no_proxy=$no_proxy
+    env_file:
+      - ../envs/env_app_1
+      - ../envs/env_medcat_1
+    volumes:
+    - ../models:/cat/models/:rw
+    ports:
+      - "5556:5000"
+    networks:
+      - cognet
+
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: 1
+              # device_ids: ['0']
               capabilities: ["gpu", "utility", "compute", "video"]
 
 networks:

docker/docker-compose.yml

@@ -25,6 +25,33 @@ services:
       - "5555:5000"
     networks:
       - cognet
+
+  nlp-medcat-service-production-1:
+    container_name: cogstack-medcat-service-production-1
+    ### Multiple images available:
+    ## default image, only CPU support: cogstacksystems/medcat-service:latest 
+    ## GPU support: cogstacksystems/medcat-service-gpu:latest
+    # image: cogstacksystems/medcat-service:latest
+    platform: linux
+    restart: always
+    ## Default dockerfile: ../Dockerfile
+    ## GPU dockerfile: ../Dockerfile_gpu
+    build:
+      context: ../
+      dockerfile: "Dockerfile"
+    environment:
+      - http_proxy=$HTTP_PROXY
+      - https_proxy=$HTTPS_PROXY
+      - no_proxy=$no_proxy
+    env_file:
+      - ../envs/env_app_1
+      - ../envs/env_medcat_1
+    volumes:
+    - ../models:/cat/models/:rw
+    ports:
+      - "5556:5000"
+    networks:
+      - cognet
 
 networks:
   cognet:

envs/env_app

@@ -17,7 +17,7 @@ APP_MODEL_META_PATH_LIST=/cat/models/Status
 # MedCAT Model Pack path
 # IMPORTANT: if this parameter has value IT WILL BE LOADED FIRST OVER EVERYTHING ELSE (CDB, Vocab, MetaCATs, etc.) declared above.
 # Respect the same paths as above : /cat/models/model_pack_name.zip
-APP_MEDCAT_MODEL_PACK=
+APP_MEDCAT_MODEL_PACK=/cat/models/medcat_deid_model_691c3f6a6e5400e7.zip
 
 # optionally, an filter the reported concepts by CUIs
 # APP_MODEL_CUI_FILTER_PATH=/cat/models/cui_filter.txt

envs/env_app_1

@@ -0,0 +1,42 @@
+# Additional settings
+
+# IMPORTANT : log level set 
+# CRITICAL - 50, ERROR - 40, WARNING - 30, INFO - 20, DEBUG - 10, NOTSET - 0
+APP_LOG_LEVEL=INFO
+
+# can include only one model for NER
+APP_NAME=MedCAT
+APP_MODEL_LANGUAGE=en
+APP_MODEL_NAME=MedMen
+APP_MODEL_CDB_PATH=/cat/models/cdb.dat
+APP_MODEL_VOCAB_PATH=/cat/models/vocab.dat
+
+# optionally, can include multiple models for meta tasks, separated using ':'
+APP_MODEL_META_PATH_LIST=/cat/models/Status
+
+# MedCAT Model Pack path
+# IMPORTANT: if this parameter has value IT WILL BE LOADED FIRST OVER EVERYTHING ELSE (CDB, Vocab, MetaCATs, etc.) declared above.
+# Respect the same paths as above : /cat/models/model_pack_name.zip
+APP_MEDCAT_MODEL_PACK=/cat/models/medcat_deid_model_691c3f6a6e5400e7.zip
+
+# optionally, an filter the reported concepts by CUIs
+# APP_MODEL_CUI_FILTER_PATH=/cat/models/cui_filter.txt
+
+# NLP processing
+APP_BULK_NPROC=8
+APP_TRAINING_MODE=False
+
+# Flask server config
+SERVER_HOST=0.0.0.0
+SERVER_PORT=5000
+SERVER_WORKERS=1
+SERVER_WORKER_TIMEOUT=300
+SERVER_THREADS=1
+
+# set the number of torch threads, this should be used ONLY if you are using CPUs and the default image
+# set to -1 or 0 if you are using GPU
+APP_TORCH_THREADS=8
+
+# GPU SETTING
+# CAUTION, use only if you are using the GPU docker image.
+APP_CUDA_DEVICE_COUNT=1

envs/env_medcat

@@ -32,5 +32,5 @@ SPACY_MODEL=
 # POSSIBLE VALUES: [list, dict], if left empty then "dict" is the default.
 ANNOTATIONS_ENTITY_OUTPUT_MODE=dict
 
-DEID_MODE=False
+DEID_MODE=True
 DEID_REDACT=True

envs/env_medcat_1

@@ -0,0 +1,36 @@
+TYPE=NOT_UMLS
+
+# IMPORTANT : log level set 
+# CRITICAL - 50, ERROR - 40, WARNING - 30, INFO - 20, DEBUG - 10, NOTSET - 0
+LOG_LEVEL=40
+
+NESTED_ENTITIES=False
+
+CNTX_SPAN=9
+CNTX_SPAN_SHORT=3
+MIN_CUI_COUNT=30000
+MIN_CUI_COUNT_STRICT=-1
+MIN_ACC=0.2
+MIN_ACC_TH=0.2
+
+LEARNING_RATE=0.1
+ANNEAL=False
+KEEP_PUNCT=:|.
+
+# can be left empty if you want to use the CDB spacy model, otherwise, possible values are : en_core_sci_lg, en_core_sci_md, etc...
+# please note that only SpaCy models that are installed on the docker container are usable (take a look at the .Dockerfile and install any other custom models you may want to use)
+SPACY_MODEL=
+
+# IMPORTANT:
+#   Mode in which annotation entities should be outputted in the JSON response,
+#   by default this is set to "list" of dicts, so the output would be : {"annotations": [{"id": "0", "cui" : "C1X..", ..}, {"id":"1", "cui": "...."}]}
+#   newer versions of MedCAT (1.2+) output entities as a dict, where the id of the entity is a key and the rest of the data is a value, so for "dict",
+#   the output is
+#    {"annotations": {"entities": {"0": {"cui": "C0027361", "id": 0,.....}, "1": {"cui": "C001111", "id": 1......}}
+#   Be mindful of this option as it can affect other services that rely directly on the responses of the service 
+#    (the NiFi groovy scripts and annotation ingester are two such services that process the output, and so they might require further customisation)
+# POSSIBLE VALUES: [list, dict], if left empty then "dict" is the default.
+ANNOTATIONS_ENTITY_OUTPUT_MODE=dict
+
+DEID_MODE=True
+DEID_REDACT=True