dream

1
services:
2
  aiml:
3
    build:
4
      args:
5
        skill_endpoint: aiml
6
        skillconfig: skills/aiml/aiml_skill.json
7
        skillhost: 0.0.0.0
8
        skillport: 2080
9
      context: ./
10
      dockerfile: dp/dockerfile_skill_cpu
11
    environment:
12
      - CUDA_VISIBLE_DEVICES=""
13
    ports:
14
      - 2080:2080
15
  transfertransfo:
16
    build:
17
      context: ./skills/transfertransfo/
18
    environment:
19
      DEVICE: cuda
20
    command: gunicorn --workers=1 server:app -b 0.0.0.0:8007 -t 60
21
    volumes: []
22
    deploy:
23
      mode: replicated
24
      replicas: 1
25
      placement:
26
        constraints:
27
          - node.labels.with_gpu == true
28
  retrieval_chitchat:
29
    build:
30
      context: ./skills/retrieval_chitchat/
31
    command: uvicorn server:app --host 0.0.0.0 --port 8015
32
    volumes: []
33
    deploy:
34
      mode: replicated
35
      replicas: 1
36
      placement:
37
        constraints:
38
          - node.labels.with_gpu != true
39
  convert_reddit_with_personality:
40
    build:
41
      context: ./skills/convert_reddit_with_personality/
42
    command: gunicorn --workers=1 server:app -b 0.0.0.0:8048 -t 60
43
    volumes: []
44
    deploy:
45
      mode: replicated
46
      replicas: 2
47
      placement:
48
        constraints:
49
          - node.labels.with_gpu != true
50
  news_skill:
51
    build:
52
      context: .
53
      dockerfile: ./skills/alexa-prize-news/Dockerfile
54
    command: bash -c "python updater.py | gunicorn --workers=1 server:app -b 0.0.0.0:8027 -t 300"
55
    volumes: []
56
    deploy:
57
      mode: replicated
58
      replicas: 1
59
      placement:
60
        constraints:
61
          - node.labels.with_gpu != true
62
  topicalchat_convert_retrieval:
63
    build:
64
      context: ./skills/topicalchat_convert_retrieval/
65
    command: gunicorn --workers=1 server:app -b 0.0.0.0:8060 -t 60
66
    volumes: []
67
    deploy:
68
      mode: replicated
69
      replicas: 2
70
      placement:
71
        constraints:
72
          - node.labels.with_gpu != true
73
          - node.labels.group == 0
74
  reddit_ner_skill:
75
    build:
76
      context: .
77
      dockerfile: ./skills/reddit_ner_skill/Dockerfile
78
    command: gunicorn --workers=2 server:app -b 0.0.0.0:8035
79
    deploy:
80
      mode: replicated
81
      replicas: 2
82
      placement:
83
        constraints:
84
          - node.labels.with_gpu != true
85
  question-generator:
86
    build:
87
      context: ./services/question_generator/
88
      args:
89
        MODEL_URL: http://lnsigo.mipt.ru/export/alexaprize_data/question_generator/model_24_0.94_37.23.pth
90
    command: gunicorn --workers=1 --timeout 300 server:app -b 0.0.0.0:8079
91
    environment:
92
      - CUDA_VISIBLE_DEVICES=0
93
      - DECODING=greedy
94
    deploy:
95
      mode: replicated
96
      replicas: 1
97
      resources:
98
        limits:
99
          memory: 3G
100
        reservations:
101
          memory: 3G
102

103
  speech-function-classifier:
104
    build:
105
      args:
106
        SERVICE_PORT: 8108
107
        SERVICE_NAME: speech_function_classifier # has to be the same with skill dir name
108
      context: ./annotators/speech_function_classifier/
109
    command: uvicorn server:app --reload --host 0.0.0.0 --port 8108
110
    deploy:
111
      mode: replicated
112
      replicas: 2
113
      resources:
114
        limits:
115
          memory: 768M
116
        reservations:
117
          memory: 768M
118

119
  speech-function-predictor:
120
    build:
121
      args:
122
        SERVICE_PORT: 8107
123
        SERVICE_NAME: speech_function_predictor # has to be the same with skill dir name
124
      context: ./annotators/speech_function_predictor/
125
    command: uvicorn server:app --reload --host 0.0.0.0 --port 8107
126
    deploy:
127
      mode: replicated
128
      replicas: 2
129
      resources:
130
        limits:
131
          memory: 768M
132
        reservations:
133
          memory: 768M
134

135
  wikidata-dial-skill:
136
    build:
137
      args:
138
         SERVICE_PORT: 8091
139
      context: skills/wikidata_dial_skill
140
    command: flask run -h 0.0.0.0 -p 8091
141
    environment:
142
      - FLASK_APP=server
143
    deploy:
144
      mode: replicated
145
      replicas: 1
146
      resources:
147
        limits:
148
          memory: 256M
149
        reservations:
150
          memory: 256M
151

152
  wikidata-dial-service:
153
    build:
154
      args:
155
        CONFIG: kg_dial_generator.json
156
        SERVICE_PORT: 8092
157
        COMMIT: 5dac85a06fde0e2983c7569ae058e3bf2d45ce2c
158
      context: services/wikidata_dial_service
159
    command: flask run -h 0.0.0.0 -p 8092
160
    environment:
161
      - CUDA_VISIBLE_DEVICES=0
162
      - FLASK_APP=server
163
    deploy:
164
      mode: replicated
165
      replicas: 1
166
      resources:
167
        limits:
168
          memory: 10G
169
        reservations:
170
          memory: 10G
171

172
version: '3.7'
173