TrustGraph-v0.15.5向けに修正したdocker-compose.yaml

tg-openai-neo4j.yaml

services:
  chunker:
    command:
    - chunker-recursive
    - -p
    - pulsar://pulsar:6650
    - --chunk-size
    - '1000'
    - --chunk-overlap
    - '200'
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  embeddings:
    command:
    - embeddings-hf
    - -p
    - pulsar://pulsar:6650
    - -m
    - all-MiniLM-L6-v2
    deploy:
      resources:
        limits:
          cpus: '1.0'
          memory: 400M
        reservations:
          cpus: '0.5'
          memory: 400M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  grafana:
    deploy:
      resources:
        limits:
          cpus: '1.0'
          memory: 256M
        reservations:
          cpus: '0.5'
          memory: 256M
    environment:
      GF_ORG_NAME: trustgraph.ai
    image: docker.io/grafana/grafana:11.1.4
    ports:
    - 3000:3000
    restart: on-failure:100
    volumes:
    - grafana-storage:/var/lib/grafana
    - ./grafana/provisioning/:/etc/grafana/provisioning/dashboards/
    - ./grafana/provisioning/:/etc/grafana/provisioning/datasources/
    - ./grafana/dashboards/:/var/lib/grafana/dashboards/
  graph-rag:
    command:
    - graph-rag
    - -p
    - pulsar://pulsar:6650
    - --prompt-request-queue
    - non-persistent://tg/request/prompt-rag
    - --prompt-response-queue
    - non-persistent://tg/response/prompt-rag-response
    - --entity-limit
    - '50'
    - --triple-limit
    - '30'
    - --max-subgraph-size
    - '3000'
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  init-pulsar:
    command:
    - tg-init-pulsar
    - -p
    - http://pulsar:8080
    deploy:
      resources:
        limits:
          cpus: '1'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  kg-extract-definitions:
    command:
    - kg-extract-definitions
    - -p
    - pulsar://pulsar:6650
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  kg-extract-relationships:
    command:
    - kg-extract-relationships
    - -p
    - pulsar://pulsar:6650
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  kg-extract-topics:
    command:
    - kg-extract-topics
    - -p
    - pulsar://pulsar:6650
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  metering:
    command:
    - metering
    - -p
    - pulsar://pulsar:6650
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  metering-rag:
    command:
    - metering
    - -p
    - pulsar://pulsar:6650
    - -i
    - non-persistent://tg/response/text-completion-rag-response
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  neo4j:
    deploy:
      resources:
        limits:
          cpus: '1.0'
          memory: 768M
        reservations:
          cpus: '0.5'
          memory: 768M
    environment:
      NEO4J_AUTH: neo4j/password
    image: docker.io/neo4j:5.22.0-community-bullseye
    ports:
    - 7474:7474
    - 7687:7687
    restart: on-failure:100
    volumes:
    - neo4j:/data
  pdf-decoder:
    command:
    - pdf-decoder
    - -p
    - pulsar://pulsar:6650
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  prometheus:
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/prom/prometheus:v2.53.2
    ports:
    - 9090:9090
    restart: on-failure:100
    volumes:
    - ./prometheus:/etc/prometheus/
    - prometheus-data:/prometheus
  prompt:
    command:
    - prompt-template
    - -p
    - pulsar://pulsar:6650
    - --text-completion-request-queue
    - non-persistent://tg/request/text-completion
    - --text-completion-response-queue
    - non-persistent://tg/response/text-completion-response
    - --system-prompt
    - "\u3042\u306A\u305F\u306F\u3001\u89AA\u5207\u306A\u65E5\u672C\u8A9E\u306E\u30A2\
      \u30B7\u30B9\u30BF\u30F3\u30C8\u3067\u3059\u3002\u3042\u306A\u305F\u306E\u4ED5\
      \u4E8B\u306F\u3001NLP\uFF08\u81EA\u7136\u8A00\u8A9E\u51E6\u7406\uFF09\u306E\u30BF\
      \u30B9\u30AF\u3092\u3053\u306A\u3059\u3053\u3068\u3067\u3059\u3002\n"
    - --prompt
    - "agent-react=Answer the following questions as best you can. You have\naccess\
      \ to the following functions:\n\n{% for tool in tools %}{\n    \"function\"\
      : \"{{ tool.name }}\",\n    \"description\": \"{{ tool.description }}\",\n \
      \   \"arguments\": [\n{% for arg in tool.arguments %}        {\n           \
      \ \"name\": \"{{ arg.name }}\",\n            \"type\": \"{{ arg.type }}\",\n\
      \            \"description\": \"{{ arg.description }}\",\n        }\n{% endfor\
      \ %}\n    ]\n}\n{% endfor %}\n\nYou can either choose to call a function to\
      \ get more information, or\nreturn a final answer.\n    \nTo call a function,\
      \ respond with a JSON object of the following format:\n\n{\n    \"thought\"\
      : \"your thought about what to do\",\n    \"action\": \"the action to take,\
      \ should be one of [{{tool_names}}]\",\n    \"arguments\": {\n        \"argument1\"\
      : \"argument_value\",\n        \"argument2\": \"argument_value\"\n    }\n}\n\
      \nTo provide a final answer, response a JSON object of the following format:\n\
      \n{\n  \"thought\": \"I now know the final answer\",\n  \"final-answer\": \"\
      the final answer to the original input question\"\n}\n\nPrevious steps are included\
      \ in the input.  Each step has the following\nformat in your output:\n\n{\n\
      \  \"thought\": \"your thought about what to do\",\n  \"action\": \"the action\
      \ taken\",\n  \"arguments\": {\n      \"argument1\": action argument,\n    \
      \  \"argument2\": action argument2\n  },\n  \"observation\": \"the result of\
      \ the action\",\n}\n\nRespond by describing either one single thought/action/arguments\
      \ or\nthe final-answer.  Pause after providing one action or final-answer.\n\
      \n{% if context %}Additional context has been provided:\n{{context}}{% endif\
      \ %}\n\nQuestion: {{question}}\n\nInput:\n    \n{% for h in history %}\n{\n\
      \    \"action\": \"{{h.action}}\",\n    \"arguments\": [\n{% for k, v in h.arguments.items()\
      \ %}        {\n            \"{{k}}\": \"{{v}}\",\n{%endfor%}        }\n    ],\n\
      \    \"observation\": \"{{h.observation}}\"\n}\n{% endfor %}"
    - 'document-prompt=Study the following context. Use only the information provided
      in the context in your response. Do not speculate if the answer is not found
      in the provided set of knowledge statements.


      Here is the context:

      {{documents}}


      Use only the provided knowledge statements to respond to the following:

      {{query}}

      '
    - 'extract-definitions=<instructions>

      Study the following text and derive definitions for any discovered entities.

      Do not provide definitions for entities whose definitions are incomplete

      or unknown.

      Output relationships in JSON format as an arary of objects with fields:

      - entity: the name of the entity

      - definition: English text which defines the entity

      </instructions>


      <text>

      {{text}}

      </text>


      <requirements>

      You will respond only with raw JSON format data. Do not provide

      explanations. Do not use special characters in the abstract text. The

      abstract will be written as plain text.  Do not add markdown formatting

      or headers or prefixes.  Do not include null or unknown definitions.

      </requirements>'
    - 'extract-relationships=<instructions>

      Study the following text and derive entity relationships.  For each

      relationship, derive the subject, predicate and object of the relationship.

      Output relationships in JSON format as an arary of objects with fields:

      - subject: the subject of the relationship

      - predicate: the predicate

      - object: the object of the relationship

      - object-entity: false if the object is a simple data type: name, value or date.  true
      if it is an entity.

      </instructions>


      <text>

      {{text}}

      </text>


      <requirements>

      You will respond only with raw JSON format data. Do not provide

      explanations. Do not use special characters in the abstract text. The

      abstract must be written as plain text.  Do not add markdown formatting

      or headers or prefixes.

      </requirements>'
    - 'extract-rows=<instructions>

      Study the following text and derive objects which match the schema provided.


      You must output an array of JSON objects for each object you discover

      which matches the schema.  For each object, output a JSON object whose fields

      carry the name field specified in the schema.

      </instructions>


      <schema>

      {{schema}}

      </schema>


      <text>

      {{text}}

      </text>


      <requirements>

      You will respond only with raw JSON format data. Do not provide

      explanations. Do not add markdown formatting or headers or prefixes.

      </requirements>'
    - "extract-topics=You are a helpful assistant that performs information extraction\
      \ tasks for a provided text.\nRead the provided text. You will identify topics\
      \ and their definitions in JSON.\n\nReading Instructions:\n- Ignore document\
      \ formatting in the provided text.\n- Study the provided text carefully.\n\n\
      Here is the text:\n{{text}}\n\nResponse Instructions: \n- Do not respond with\
      \ special characters.\n- Return only topics that are concepts and unique to\
      \ the provided text.\n- Respond only with well-formed JSON.\n- The JSON response\
      \ shall be an array of objects with keys \"topic\" and \"definition\". \n- The\
      \ JSON response shall use the following structure:\n\n```json\n[{\"topic\":\
      \ string, \"definition\": string}]\n```\n\n- Do not write any additional text\
      \ or explanations."
    - 'kg-prompt=Study the following set of knowledge statements. The statements are
      written in Cypher format that has been extracted from a knowledge graph. Use
      only the provided set of knowledge statements in your response. Do not speculate
      if the answer is not found in the provided set of knowledge statements.


      Here''s the knowledge statements:

      {% for edge in knowledge %}({{edge.s}})-[{{edge.p}}]->({{edge.o}})

      {%endfor%}


      Use only the provided knowledge statements to respond to the following:

      {{query}}

      '
    - question={{question}}
    - --prompt-response-type
    - agent-react=json
    - document-prompt=text
    - extract-definitions=json
    - extract-relationships=json
    - extract-rows=json
    - extract-topics=json
    - kg-prompt=text
    - --prompt-schema
    - extract-definitions={"items":{"properties":{"definition":{"type":"string"},"entity":{"type":"string"}},"required":["entity","definition"],"type":"object"},"type":"array"}
    - extract-relationships={"items":{"properties":{"object":{"type":"string"},"object-entity":{"type":"boolean"},"predicate":{"type":"string"},"subject":{"type":"string"}},"required":["subject","predicate","object","object-entity"],"type":"object"},"type":"array"}
    - extract-topics={"items":{"properties":{"definition":{"type":"string"},"topic":{"type":"string"}},"required":["topic","definition"],"type":"object"},"type":"array"}
    - --prompt-term
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  prompt-rag:
    command:
    - prompt-template
    - -p
    - pulsar://pulsar:6650
    - -i
    - non-persistent://tg/request/prompt-rag
    - -o
    - non-persistent://tg/response/prompt-rag-response
    - --text-completion-request-queue
    - non-persistent://tg/request/text-completion-rag
    - --text-completion-response-queue
    - non-persistent://tg/response/text-completion-rag-response
    - --system-prompt
    - "\u3042\u306A\u305F\u306F\u3001\u89AA\u5207\u306A\u65E5\u672C\u8A9E\u306E\u30A2\
      \u30B7\u30B9\u30BF\u30F3\u30C8\u3067\u3059\u3002\u3042\u306A\u305F\u306E\u4ED5\
      \u4E8B\u306F\u3001NLP\uFF08\u81EA\u7136\u8A00\u8A9E\u51E6\u7406\uFF09\u306E\u30BF\
      \u30B9\u30AF\u3092\u3053\u306A\u3059\u3053\u3068\u3067\u3059\u3002\n"
    - --prompt
    - "agent-react=Answer the following questions as best you can. You have\naccess\
      \ to the following functions:\n\n{% for tool in tools %}{\n    \"function\"\
      : \"{{ tool.name }}\",\n    \"description\": \"{{ tool.description }}\",\n \
      \   \"arguments\": [\n{% for arg in tool.arguments %}        {\n           \
      \ \"name\": \"{{ arg.name }}\",\n            \"type\": \"{{ arg.type }}\",\n\
      \            \"description\": \"{{ arg.description }}\",\n        }\n{% endfor\
      \ %}\n    ]\n}\n{% endfor %}\n\nYou can either choose to call a function to\
      \ get more information, or\nreturn a final answer.\n    \nTo call a function,\
      \ respond with a JSON object of the following format:\n\n{\n    \"thought\"\
      : \"your thought about what to do\",\n    \"action\": \"the action to take,\
      \ should be one of [{{tool_names}}]\",\n    \"arguments\": {\n        \"argument1\"\
      : \"argument_value\",\n        \"argument2\": \"argument_value\"\n    }\n}\n\
      \nTo provide a final answer, response a JSON object of the following format:\n\
      \n{\n  \"thought\": \"I now know the final answer\",\n  \"final-answer\": \"\
      the final answer to the original input question\"\n}\n\nPrevious steps are included\
      \ in the input.  Each step has the following\nformat in your output:\n\n{\n\
      \  \"thought\": \"your thought about what to do\",\n  \"action\": \"the action\
      \ taken\",\n  \"arguments\": {\n      \"argument1\": action argument,\n    \
      \  \"argument2\": action argument2\n  },\n  \"observation\": \"the result of\
      \ the action\",\n}\n\nRespond by describing either one single thought/action/arguments\
      \ or\nthe final-answer.  Pause after providing one action or final-answer.\n\
      \n{% if context %}Additional context has been provided:\n{{context}}{% endif\
      \ %}\n\nQuestion: {{question}}\n\nInput:\n    \n{% for h in history %}\n{\n\
      \    \"action\": \"{{h.action}}\",\n    \"arguments\": [\n{% for k, v in h.arguments.items()\
      \ %}        {\n            \"{{k}}\": \"{{v}}\",\n{%endfor%}        }\n    ],\n\
      \    \"observation\": \"{{h.observation}}\"\n}\n{% endfor %}"
    - 'document-prompt=Study the following context. Use only the information provided
      in the context in your response. Do not speculate if the answer is not found
      in the provided set of knowledge statements.


      Here is the context:

      {{documents}}


      Use only the provided knowledge statements to respond to the following:

      {{query}}

      '
    - 'extract-definitions=<instructions>

      Study the following text and derive definitions for any discovered entities.

      Do not provide definitions for entities whose definitions are incomplete

      or unknown.

      Output relationships in JSON format as an arary of objects with fields:

      - entity: the name of the entity

      - definition: English text which defines the entity

      </instructions>


      <text>

      {{text}}

      </text>


      <requirements>

      You will respond only with raw JSON format data. Do not provide

      explanations. Do not use special characters in the abstract text. The

      abstract will be written as plain text.  Do not add markdown formatting

      or headers or prefixes.  Do not include null or unknown definitions.

      </requirements>'
    - 'extract-relationships=<instructions>

      Study the following text and derive entity relationships.  For each

      relationship, derive the subject, predicate and object of the relationship.

      Output relationships in JSON format as an arary of objects with fields:

      - subject: the subject of the relationship

      - predicate: the predicate

      - object: the object of the relationship

      - object-entity: false if the object is a simple data type: name, value or date.  true
      if it is an entity.

      </instructions>


      <text>

      {{text}}

      </text>


      <requirements>

      You will respond only with raw JSON format data. Do not provide

      explanations. Do not use special characters in the abstract text. The

      abstract must be written as plain text.  Do not add markdown formatting

      or headers or prefixes.

      </requirements>'
    - 'extract-rows=<instructions>

      Study the following text and derive objects which match the schema provided.


      You must output an array of JSON objects for each object you discover

      which matches the schema.  For each object, output a JSON object whose fields

      carry the name field specified in the schema.

      </instructions>


      <schema>

      {{schema}}

      </schema>


      <text>

      {{text}}

      </text>


      <requirements>

      You will respond only with raw JSON format data. Do not provide

      explanations. Do not add markdown formatting or headers or prefixes.

      </requirements>'
    - "extract-topics=You are a helpful assistant that performs information extraction\
      \ tasks for a provided text.\nRead the provided text. You will identify topics\
      \ and their definitions in JSON.\n\nReading Instructions:\n- Ignore document\
      \ formatting in the provided text.\n- Study the provided text carefully.\n\n\
      Here is the text:\n{{text}}\n\nResponse Instructions: \n- Do not respond with\
      \ special characters.\n- Return only topics that are concepts and unique to\
      \ the provided text.\n- Respond only with well-formed JSON.\n- The JSON response\
      \ shall be an array of objects with keys \"topic\" and \"definition\". \n- The\
      \ JSON response shall use the following structure:\n\n```json\n[{\"topic\":\
      \ string, \"definition\": string}]\n```\n\n- Do not write any additional text\
      \ or explanations."
    - 'kg-prompt=Study the following set of knowledge statements. The statements are
      written in Cypher format that has been extracted from a knowledge graph. Use
      only the provided set of knowledge statements in your response. Do not speculate
      if the answer is not found in the provided set of knowledge statements.


      Here''s the knowledge statements:

      {% for edge in knowledge %}({{edge.s}})-[{{edge.p}}]->({{edge.o}})

      {%endfor%}


      Use only the provided knowledge statements to respond to the following:

      {{query}}

      '
    - question={{question}}
    - --prompt-response-type
    - agent-react=json
    - document-prompt=text
    - extract-definitions=json
    - extract-relationships=json
    - extract-rows=json
    - extract-topics=json
    - kg-prompt=text
    - --prompt-schema
    - extract-definitions={"items":{"properties":{"definition":{"type":"string"},"entity":{"type":"string"}},"required":["entity","definition"],"type":"object"},"type":"array"}
    - extract-relationships={"items":{"properties":{"object":{"type":"string"},"object-entity":{"type":"boolean"},"predicate":{"type":"string"},"subject":{"type":"string"}},"required":["subject","predicate","object","object-entity"],"type":"object"},"type":"array"}
    - extract-topics={"items":{"properties":{"definition":{"type":"string"},"topic":{"type":"string"}},"required":["topic","definition"],"type":"object"},"type":"array"}
    - --prompt-term
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  pulsar:
    command:
    - bin/pulsar
    - standalone
    deploy:
      resources:
        limits:
          cpus: '2.0'
          memory: 1500M
        reservations:
          cpus: '1.0'
          memory: 1500M
    environment:
      PULSAR_MEM: -Xms600M -Xmx600M
    image: docker.io/apachepulsar/pulsar:3.3.1
    ports:
    - 6650:6650
    - 8080:8080
    restart: on-failure:100
    volumes:
    - pulsar-data:/pulsar/data
  qdrant:
    deploy:
      resources:
        limits:
          cpus: '1.0'
          memory: 1024M
        reservations:
          cpus: '0.5'
          memory: 1024M
    image: docker.io/qdrant/qdrant:v1.11.1
    ports:
    - 6333:6333
    - 6334:6334
    restart: on-failure:100
    volumes:
    - qdrant:/qdrant/storage
  query-doc-embeddings:
    command:
    - de-query-qdrant
    - -p
    - pulsar://pulsar:6650
    - -t
    - http://qdrant:6333
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  query-graph-embeddings:
    command:
    - ge-query-qdrant
    - -p
    - pulsar://pulsar:6650
    - -t
    - http://qdrant:6333
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  query-triples:
    command:
    - triples-query-neo4j
    - -p
    - pulsar://pulsar:6650
    - -g
    - bolt://neo4j:7687
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  store-doc-embeddings:
    command:
    - de-write-qdrant
    - -p
    - pulsar://pulsar:6650
    - -t
    - http://qdrant:6333
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  store-graph-embeddings:
    command:
    - ge-write-qdrant
    - -p
    - pulsar://pulsar:6650
    - -t
    - http://qdrant:6333
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  store-triples:
    command:
    - triples-write-neo4j
    - -p
    - pulsar://pulsar:6650
    - -g
    - bolt://neo4j:7687
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  text-completion:
    command:
    - text-completion-openai
    - -p
    - pulsar://pulsar:6650
    - -x
    - '4096'
    - -t
    - '0.300'
    - -m
    - gpt-4o-mini
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    environment:
      OPENAI_TOKEN: ${OPENAI_TOKEN}
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  text-completion-rag:
    command:
    - text-completion-openai
    - -p
    - pulsar://pulsar:6650
    - -x
    - '4096'
    - -t
    - '0.000'
    - -m
    - gpt-4o-mini
    - -i
    - non-persistent://tg/request/text-completion-rag
    - -o
    - non-persistent://tg/response/text-completion-rag-response
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 128M
        reservations:
          cpus: '0.1'
          memory: 128M
    environment:
      OPENAI_TOKEN: ${OPENAI_TOKEN}
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
  vectorize:
    command:
    - embeddings-vectorize
    - -p
    - pulsar://pulsar:6650
    deploy:
      resources:
        limits:
          cpus: '1.0'
          memory: 512M
        reservations:
          cpus: '0.5'
          memory: 512M
    image: docker.io/trustgraph/trustgraph-flow:0.15.5
    restart: on-failure:100
volumes:
  grafana-storage: {}
  neo4j: {}
  prometheus-data: {}
  pulsar-data: {}
  qdrant: {}