mirror of
https://github.com/trustgraph-ai/trustgraph.git
synced 2026-06-10 15:25:14 +02:00
Merge branch 'release/v0.15'
This commit is contained in:
Cyber MacGeddon
commit
fd6abdc4c1
149 changed files with 4856 additions and 1274 deletions
20
.github/workflows/pull-request.yaml
vendored
Normal file
20
.github/workflows/pull-request.yaml
vendored
Normal file
|
|
@ -0,0 +1,20 @@
|
|||
|
||||
name: Test pull request
|
||||
|
||||
on:
|
||||
pull_request:
|
||||
|
||||
permissions:
|
||||
contents: read
|
||||
|
||||
jobs:
|
||||
|
||||
container-push:
|
||||
|
||||
name: Do nothing
|
||||
runs-on: ubuntu-latest
|
||||
|
||||
steps:
|
||||
- name: Checkout
|
||||
uses: actions/checkout@v3
|
||||
|
||||
86
.github/workflows/release.yaml
vendored
Normal file
86
.github/workflows/release.yaml
vendored
Normal file
|
|
@ -0,0 +1,86 @@
|
|||
|
||||
name: Build
|
||||
|
||||
on:
|
||||
workflow_dispatch:
|
||||
push:
|
||||
tags:
|
||||
- v0.15.*
|
||||
|
||||
permissions:
|
||||
contents: read
|
||||
|
||||
jobs:
|
||||
|
||||
deploy:
|
||||
|
||||
name: Build everything
|
||||
runs-on: ubuntu-latest
|
||||
permissions:
|
||||
contents: write
|
||||
id-token: write
|
||||
environment:
|
||||
name: release
|
||||
|
||||
steps:
|
||||
|
||||
- name: Checkout
|
||||
uses: actions/checkout@v3
|
||||
|
||||
- name: Log in to Docker Hub
|
||||
uses: docker/login-action@f4ef78c080cd8ba55a85445d5b36e214a81df20a
|
||||
with:
|
||||
username: ${{ secrets.DOCKER_USERNAME }}
|
||||
password: ${{ secrets.DOCKER_SECRET }}
|
||||
|
||||
- name: Install build dependencies
|
||||
run: pip3 install jsonnet
|
||||
|
||||
- name: Get version
|
||||
id: version
|
||||
run: echo VERSION=$(git describe --exact-match --tags | sed 's/^v//') >> $GITHUB_OUTPUT
|
||||
|
||||
- run: echo ${{ steps.version.outputs.VERSION }}
|
||||
|
||||
- name: Build packages
|
||||
run: make packages VERSION=${{ steps.version.outputs.VERSION }}
|
||||
|
||||
- name: Publish release distributions to PyPI
|
||||
uses: pypa/gh-action-pypi-publish@release/v1
|
||||
|
||||
- name: Create deploy bundle
|
||||
run: templates/generate-all deploy.zip ${{ steps.version.outputs.VERSION }}
|
||||
|
||||
- uses: ncipollo/release-action@v1
|
||||
with:
|
||||
artifacts: deploy.zip
|
||||
generateReleaseNotes: true
|
||||
makeLatest: false
|
||||
prerelease: true
|
||||
skipIfReleaseExists: true
|
||||
|
||||
- name: Build container
|
||||
run: make container VERSION=${{ steps.version.outputs.VERSION }}
|
||||
|
||||
- name: Extract metadata for container
|
||||
id: meta
|
||||
uses: docker/metadata-action@v4
|
||||
with:
|
||||
images: trustgraph/trustgraph-flow
|
||||
tags: |
|
||||
type=ref,event=branch
|
||||
type=ref,event=pr
|
||||
type=semver,pattern={{version}}
|
||||
type=semver,pattern={{major}}.{{minor}}
|
||||
type=sha
|
||||
|
||||
- name: Build and push Docker image
|
||||
id: push
|
||||
uses: docker/build-push-action@3b5e8027fcad23fda98b2e3ac259d8d67585f671
|
||||
with:
|
||||
context: .
|
||||
file: ./Containerfile
|
||||
push: true
|
||||
tags: ${{ steps.meta.outputs.tags }}
|
||||
labels: ${{ steps.meta.outputs.labels }}
|
||||
|
||||
|
|
@ -13,7 +13,7 @@ RUN dnf install -y python3 python3-pip python3-wheel python3-aiohttp \
|
|||
|
||||
RUN pip3 install torch --index-url https://download.pytorch.org/whl/cpu
|
||||
|
||||
RUN pip3 install anthropic boto3 cohere openai google-cloud-aiplatform ollama \
|
||||
RUN pip3 install anthropic boto3 cohere openai google-cloud-aiplatform ollama google-generativeai \
|
||||
langchain langchain-core langchain-huggingface langchain-text-splitters \
|
||||
langchain-community pymilvus sentence-transformers transformers \
|
||||
huggingface-hub pulsar-client cassandra-driver pyarrow pyyaml \
|
||||
|
|
|
|||
4
Makefile
4
Makefile
|
|
@ -1,6 +1,7 @@
|
|||
|
||||
# VERSION=$(shell git describe | sed 's/^v//')
|
||||
VERSION=0.11.20
|
||||
|
||||
VERSION=0.0.0
|
||||
|
||||
DOCKER=podman
|
||||
|
||||
|
|
@ -35,6 +36,7 @@ CONTAINER=docker.io/trustgraph/trustgraph-flow
|
|||
|
||||
update-package-versions:
|
||||
mkdir -p trustgraph-cli/trustgraph
|
||||
mkdir -p trustgraph/trustgraph
|
||||
echo __version__ = \"${VERSION}\" > trustgraph-base/trustgraph/base_version.py
|
||||
echo __version__ = \"${VERSION}\" > trustgraph-flow/trustgraph/flow_version.py
|
||||
echo __version__ = \"${VERSION}\" > trustgraph-vertexai/trustgraph/vertexai_version.py
|
||||
|
|
|
|||
18
docs/README.agent-demo
Normal file
18
docs/README.agent-demo
Normal file
|
|
@ -0,0 +1,18 @@
|
|||
podman-compose -f docker-compose.yaml up -d
|
||||
|
||||
|
||||
tg-processor-state
|
||||
|
||||
tg-load-text --keyword cats animals home-life --name "Mark's cats" --description "This document describes Mark's cats" --copyright-notice 'Public domain' --publication-organization 'trustgraph.ai' --publication-date 2024-10-23 --copyright-holder 'trustgraph.ai' --copyright-year 2024 --publication-description 'Uploading to Github' --url https://example.com --id TG-000001 ../trustgraph/README.cats
|
||||
|
||||
tg-load-text --keyword nasa challenger space-shuttle shuttle orbiter --name 'Challenger Report Volume 1' --description 'The findings of the Presidential Commission regarding the circumstances surrounding the Challenger accident are reported and recommendations for corrective action are outlined' --copyright-notice 'Work of the US Gov. Public Use Permitted' --publication-organization 'NASA' --publication-date 1986-06-06 --copyright-holder 'US Government' --copyright-year 1986 --publication-description 'The findings of the Commission regarding the circumstances surrounding the Challenger accident are reported' --url https://ntrs.nasa.gov/citations/19860015255 --id AD-A171402 ../trustgraph/README.challenger
|
||||
|
||||
|
||||
tg-graph-show
|
||||
|
||||
tg-query-graph-rag -q 'Tell me cat facts'
|
||||
|
||||
|
||||
tg-invoke-agent -v -q "How many cats does Mark have? Calculate that number raised to 0.4 power. Is that number lower than the numeric part of the mission identifier of the Space Shuttle Challenger on its last mission? If so, give me an apple pie recipe, otherwise return a poem about cheese."
|
||||
|
||||
|
||||
35
docs/README.cats
Normal file
35
docs/README.cats
Normal file
|
|
@ -0,0 +1,35 @@
|
|||
|
||||
My name is Mark.
|
||||
|
||||
I have 2 cats:
|
||||
- Fred is a big, fat, orange, stripy cat. He is 12 years old and has 4 legs.
|
||||
- Hope is a small, black cat. She is 7 years old and also has 4 legs.
|
||||
|
||||
Fred has 4 legs.
|
||||
|
||||
Hope has 4 legs.
|
||||
|
||||
Fred and Hope are nice animals, but occasionally they fight.
|
||||
|
||||
Fred is lazy and sleeps a lot. Hope is energetic, runs around a lot and
|
||||
climbs trees.
|
||||
|
||||
Both cats have tails and whiskers like all cats do.
|
||||
|
||||
Cats have the species name Felis catus.
|
||||
|
||||
The cat (Felis catus), also referred to as domestic cat or house cat, is a
|
||||
small domesticated carnivorous mammal. It is the only domesticated species of
|
||||
the family Felidae. Advances in archaeology and genetics have shown that the
|
||||
domestication of the cat occurred in the Near East around 7500 BC. It is
|
||||
commonly kept as a pet and farm cat, but also ranges freely as a feral cat
|
||||
avoiding human contact. Valued by humans for companionship and its ability to
|
||||
kill vermin, the cat's retractable claws are adapted to killing small prey
|
||||
like mice and rats. It has a strong, flexible body, quick reflexes, and sharp
|
||||
teeth, and its night vision and sense of smell are well developed. It is a
|
||||
social species, but a solitary hunter and a crepuscular predator. Cat
|
||||
communication includes vocalizations—including meowing, purring, trilling,
|
||||
hissing, growling, and grunting–as well as body language. It can hear sounds
|
||||
too faint or too high in frequency for human ears, such as those made by small
|
||||
mammals. It secretes and perceives pheromones.
|
||||
|
||||
54
docs/README.challenger
Normal file
54
docs/README.challenger
Normal file
|
|
@ -0,0 +1,54 @@
|
|||
|
||||
On January 28, 1986, the Space Shuttle Challenger broke apart 73 seconds into
|
||||
its flight, killing all seven crew members aboard. The spacecraft
|
||||
disintegrated 46,000 feet (14 km) above the Atlantic Ocean, off the coast of
|
||||
Cape Canaveral, Florida, at 11:39 a.m. EST (16:39 UTC). It was the first fatal
|
||||
accident involving an American spacecraft while in flight.
|
||||
|
||||
The mission, designated STS-51-L, was the 10th flight for the orbiter and the
|
||||
25th flight of the Space Shuttle fleet. The crew was scheduled to deploy a
|
||||
communications satellite and study Halley's Comet while they were in orbit, in
|
||||
addition to taking schoolteacher Christa McAuliffe into space under the
|
||||
Teacher In Space program. The latter task resulted in a higher-than-usual
|
||||
media interest in and coverage of the mission; the launch and subsequent
|
||||
disaster were seen live in many schools across the United States.
|
||||
|
||||
The cause of the disaster was the failure of the primary and secondary O-ring
|
||||
seals in a joint in the shuttle's right solid rocket booster (SRB). The
|
||||
record-low temperatures on the morning of the launch had stiffened the rubber
|
||||
O-rings, reducing their ability to seal the joints. Shortly after liftoff, the
|
||||
seals were breached, and hot pressurized gas from within the SRB leaked
|
||||
through the joint and burned through the aft attachment strut connecting it to
|
||||
the external propellant tank (ET), then into the tank itself. The collapse of
|
||||
the ET's internal structures and the rotation of the SRB that followed threw
|
||||
the shuttle stack, traveling at a speed of Mach 1.92, into a direction that
|
||||
allowed aerodynamic forces to tear the orbiter apart. Both SRBs detached from
|
||||
the now-destroyed ET and continued to fly uncontrollably until the range
|
||||
safety officer destroyed them.
|
||||
|
||||
The crew compartment, human remains, and many other fragments from the shuttle
|
||||
were recovered from the ocean floor after a three-month search-and-recovery
|
||||
operation. The exact timing of the deaths of the crew is unknown, but several
|
||||
crew members are thought to have survived the initial breakup of the
|
||||
spacecraft. The orbiter had no escape system, and the impact of the crew
|
||||
compartment at terminal velocity with the ocean surface was too violent to be
|
||||
survivable.
|
||||
|
||||
The disaster resulted in a 32-month hiatus in the Space Shuttle
|
||||
program. President Ronald Reagan created the Rogers Commission to investigate
|
||||
the accident. The commission criticized NASA's organizational culture and
|
||||
decision-making processes that had contributed to the accident. Test data
|
||||
since 1977 demonstrated a potentially catastrophic flaw in the SRBs' O-rings,
|
||||
but neither NASA nor SRB manufacturer Morton Thiokol had addressed this known
|
||||
defect. NASA managers also disregarded engineers' warnings about the dangers
|
||||
of launching in cold temperatures and did not report these technical concerns
|
||||
to their superiors.
|
||||
|
||||
As a result of this disaster, NASA established the Office of Safety,
|
||||
Reliability, and Quality Assurance, and arranged for deployment of commercial
|
||||
satellites from expendable launch vehicles rather than from a crewed
|
||||
orbiter. To replace Challenger, the construction of a new Space Shuttle
|
||||
orbiter, Endeavour, was approved in 1987, and the new orbiter first flew in
|
||||
1992. Subsequent missions were launched with redesigned SRBs and their crews
|
||||
wore pressurized suits during ascent and reentry.
|
||||
|
||||
|
|
@ -186,7 +186,7 @@ To change the `Ollama` model, first make sure the desired model has been pulled
|
|||
### OpenAI API
|
||||
|
||||
```
|
||||
export OPENAI_KEY=<TOKEN-GOES-HERE>
|
||||
export OPENAI_TOKEN=<TOKEN-GOES-HERE>
|
||||
docker compose -f tg-launch-openai-cassandra.yaml up -d # Using Cassandra as the graph store
|
||||
docker compose -f tg-launch-openai-neo4j.yaml up -d # Using Neo4j as the graph store
|
||||
```
|
||||
|
|
@ -458,4 +458,4 @@ docker compose -f tg-launch-<model-deployment>-<graph-store>.yaml down -v
|
|||
> To confirm all Docker volumes have been removed, check that the following list is empty:
|
||||
> ```
|
||||
> docker volume ls
|
||||
> ```
|
||||
> ```
|
||||
|
|
|
|||
32
schema.ttl
Normal file
32
schema.ttl
Normal file
|
|
@ -0,0 +1,32 @@
|
|||
@prefix ns1: <http://trustgraph.ai/e/> .
|
||||
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
|
||||
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
|
||||
@prefix schema: <https://schema.org/> .
|
||||
@prefix skos: <http://www.w3.org/2004/02/skos/core#> .
|
||||
|
||||
schema:subjectOf rdfs:label "subject of" .
|
||||
skos:definition rdfs:label "definition" .
|
||||
|
||||
rdf:type rdfs:label "type" .
|
||||
|
||||
schema:DigitalDocument rdfs:label "digital document" .
|
||||
schema:Organization rdfs:label "organization" .
|
||||
schema:PublicationEvent rdfs:label "publication event" .
|
||||
|
||||
schema:copyrightNotice rdfs:label "copyright notice" .
|
||||
schema:copyrightHolder rdfs:label "copyright holder" .
|
||||
schema:copyrightYear rdfs:label "copyright year" .
|
||||
schema:license rdfs:label "license" .
|
||||
schema:publication rdfs:label "publication" .
|
||||
schema:startDate rdfs:label "start date" .
|
||||
schema:endDate rdfs:label "end date" .
|
||||
schema:publishedBy rdfs:label "published by" .
|
||||
schema:datePublished rdfs:label "date published" .
|
||||
schema:publication rdfs:label "publication" .
|
||||
schema:datePublished rdfs:label "date published" .
|
||||
schema:url rdfs:label "url" .
|
||||
schema:identifier rdfs:label "identifier" .
|
||||
schema:keywords rdfs:label "keyword" .
|
||||
|
||||
skos:definition rdfs:label "definition" .
|
||||
|
||||
|
|
@ -7,6 +7,7 @@
|
|||
import "patterns/triple-store-neo4j.jsonnet",
|
||||
import "patterns/graph-rag.jsonnet",
|
||||
import "patterns/llm-azure.jsonnet",
|
||||
import "patterns/llm-azure-openai.jsonnet",
|
||||
import "patterns/llm-bedrock.jsonnet",
|
||||
import "patterns/llm-claude.jsonnet",
|
||||
import "patterns/llm-cohere.jsonnet",
|
||||
|
|
|
|||
|
|
@ -1,11 +1,13 @@
|
|||
{
|
||||
"azure": import "components/azure.jsonnet",
|
||||
"azure-openai": import "components/azure-openai.jsonnet",
|
||||
"bedrock": import "components/bedrock.jsonnet",
|
||||
"claude": import "components/claude.jsonnet",
|
||||
"cohere": import "components/cohere.jsonnet",
|
||||
"document-rag": import "components/document-rag.jsonnet",
|
||||
"embeddings-hf": import "components/embeddings-hf.jsonnet",
|
||||
"embeddings-ollama": import "components/embeddings-ollama.jsonnet",
|
||||
"googleaistudio": import "components/googleaistudio.jsonnet",
|
||||
"grafana": import "components/grafana.jsonnet",
|
||||
"graph-rag": import "components/graph-rag.jsonnet",
|
||||
"triple-store-cassandra": import "components/cassandra.jsonnet",
|
||||
|
|
@ -14,11 +16,10 @@
|
|||
"ollama": import "components/ollama.jsonnet",
|
||||
"openai": import "components/openai.jsonnet",
|
||||
"override-recursive-chunker": import "components/chunker-recursive.jsonnet",
|
||||
"prompt-template-definitions": import "components/null.jsonnet",
|
||||
"prompt-template-document-query": import "components/null.jsonnet",
|
||||
"prompt-template-kq-query": import "components/null.jsonnet",
|
||||
"prompt-template-relationships": import "components/null.jsonnet",
|
||||
"prompt-template-rows-template": import "components/null.jsonnet",
|
||||
|
||||
"prompt-template": import "components/prompt-template.jsonnet",
|
||||
"prompt-overrides": import "components/prompt-overrides.jsonnet",
|
||||
|
||||
"pulsar": import "components/pulsar.jsonnet",
|
||||
"pulsar-manager": import "components/pulsar-manager.jsonnet",
|
||||
"trustgraph-base": import "components/trustgraph.jsonnet",
|
||||
|
|
@ -27,6 +28,8 @@
|
|||
"vertexai": import "components/vertexai.jsonnet",
|
||||
"null": {},
|
||||
|
||||
"agent-manager-react": import "components/agent-manager-react.jsonnet",
|
||||
|
||||
// FIXME: Dupes
|
||||
"cassandra": import "components/cassandra.jsonnet",
|
||||
"neo4j": import "components/neo4j.jsonnet",
|
||||
|
|
|
|||
60
templates/components/agent-manager-react.jsonnet
Normal file
60
templates/components/agent-manager-react.jsonnet
Normal file
|
|
@ -0,0 +1,60 @@
|
|||
local base = import "base/base.jsonnet";
|
||||
local images = import "values/images.jsonnet";
|
||||
local url = import "values/url.jsonnet";
|
||||
local prompts = import "prompts/mixtral.jsonnet";
|
||||
local default_prompts = import "prompts/default-prompts.jsonnet";
|
||||
|
||||
{
|
||||
|
||||
tools:: [],
|
||||
|
||||
"agent-manager" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local container =
|
||||
engine.container("agent-manager")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"agent-manager-react",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"--tool-type",
|
||||
] + [
|
||||
tool.id + "=" + tool.type
|
||||
for tool in $.tools
|
||||
] + [
|
||||
"--tool-description"
|
||||
] + [
|
||||
tool.id + "=" + tool.description
|
||||
for tool in $.tools
|
||||
] + [
|
||||
"--tool-argument"
|
||||
] + [
|
||||
"%s=%s:%s:%s" % [
|
||||
tool.id, arg.name, arg.type, arg.description
|
||||
]
|
||||
for tool in $.tools
|
||||
for arg in tool.arguments
|
||||
]
|
||||
)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"agent-manager", [ container ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
service,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
} + default_prompts
|
||||
|
||||
84
templates/components/azure-openai.jsonnet
Normal file
84
templates/components/azure-openai.jsonnet
Normal file
|
|
@ -0,0 +1,84 @@
|
|||
local base = import "base/base.jsonnet";
|
||||
local images = import "values/images.jsonnet";
|
||||
local url = import "values/url.jsonnet";
|
||||
local prompts = import "prompts/mixtral.jsonnet";
|
||||
|
||||
{
|
||||
|
||||
"azure-openai-model":: "GPT-3.5-Turbo",
|
||||
"azure-openai-max-output-tokens":: 4192,
|
||||
"azure-openai-temperature":: 0.0,
|
||||
|
||||
"text-completion" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local envSecrets = engine.envSecrets("azure-openai-credentials")
|
||||
.with_env_var("AZURE_TOKEN", "azure-token");
|
||||
|
||||
local container =
|
||||
engine.container("text-completion")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-azure-openai",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-m",
|
||||
$["azure-openai-model"],
|
||||
"-x",
|
||||
std.toString($["azure-openai-max-output-tokens"]),
|
||||
"-t",
|
||||
"%0.3f" % $["azure-openai-temperature"],
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerRag =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-azure",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-x",
|
||||
std.toString($["azure-openai-max-output-tokens"]),
|
||||
"-t",
|
||||
"%0.3f" % $["azure-openai-temperature"],
|
||||
"-i",
|
||||
"non-persistent://tg/request/text-completion-rag",
|
||||
"-o",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local containerSetRag = engine.containers(
|
||||
"text-completion-rag", [ containerRag ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
local serviceRag =
|
||||
engine.internalService(containerSetRag)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
envSecrets,
|
||||
containerSet,
|
||||
containerSetRag,
|
||||
service,
|
||||
serviceRag,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
} + prompts
|
||||
|
||||
|
|
@ -5,8 +5,6 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
|
||||
{
|
||||
|
||||
"azure-token":: "${AZURE_TOKEN}",
|
||||
"azure-endpoint":: "${AZURE_ENDPOINT}",
|
||||
"azure-max-output-tokens":: 4096,
|
||||
"azure-temperature":: 0.0,
|
||||
|
||||
|
|
@ -14,6 +12,10 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
|
||||
create:: function(engine)
|
||||
|
||||
local envSecrets = engine.envSecrets("azure-credentials")
|
||||
.with_env_var("AZURE_TOKEN", "azure-token")
|
||||
.with_env_var("AZURE_ENDPOINT", "azure-endpoint");
|
||||
|
||||
local container =
|
||||
engine.container("text-completion")
|
||||
.with_image(images.trustgraph)
|
||||
|
|
@ -21,15 +23,32 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
"text-completion-azure",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-k",
|
||||
$["azure-token"],
|
||||
"-e",
|
||||
$["azure-endpoint"],
|
||||
"-x",
|
||||
std.toString($["azure-max-output-tokens"]),
|
||||
"-t",
|
||||
std.toString($["azure-temperature"]),
|
||||
"%0.3f" % $["azure-temperature"],
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerRag =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-azure",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-x",
|
||||
std.toString($["azure-max-output-tokens"]),
|
||||
"-t",
|
||||
"%0.3f" % $["azure-temperature"],
|
||||
"-i",
|
||||
"non-persistent://tg/request/text-completion-rag",
|
||||
"-o",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
|
|
@ -37,57 +56,25 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
service,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
"text-completion-rag" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local container =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-azure",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-k",
|
||||
$["azure-token"],
|
||||
"-e",
|
||||
$["azure-endpoint"],
|
||||
"-x",
|
||||
std.toString($["azure-max-output-tokens"]),
|
||||
"-t",
|
||||
std.toString($["azure-temperature"]),
|
||||
"-i",
|
||||
"non-persistent://tg/request/text-completion-rag",
|
||||
"-o",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
])
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion-rag", [ container ]
|
||||
local containerSetRag = engine.containers(
|
||||
"text-completion-rag", [ containerRag ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
service,
|
||||
])
|
||||
local serviceRag =
|
||||
engine.internalService(containerSetRag)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
envSecrets,
|
||||
containerSet,
|
||||
containerSetRag,
|
||||
service,
|
||||
serviceRag,
|
||||
])
|
||||
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -6,9 +6,6 @@ local chunker = import "chunker-recursive.jsonnet";
|
|||
|
||||
{
|
||||
|
||||
"aws-id-key":: "${AWS_ID_KEY}",
|
||||
"aws-secret-key":: "${AWS_SECRET_KEY}",
|
||||
"aws-region":: "us-west-2",
|
||||
"bedrock-max-output-tokens":: 4096,
|
||||
"bedrock-temperature":: 0.0,
|
||||
"bedrock-model":: "mistral.mixtral-8x7b-instruct-v0:1",
|
||||
|
|
@ -17,6 +14,11 @@ local chunker = import "chunker-recursive.jsonnet";
|
|||
|
||||
create:: function(engine)
|
||||
|
||||
local envSecrets = engine.envSecrets("bedrock-credentials")
|
||||
.with_env_var("AWS_ID_KEY", "aws-id-key")
|
||||
.with_env_var("AWS_SECRET", "aws-secret")
|
||||
.with_env_var("AWS_REGION", "aws-region");
|
||||
|
||||
local container =
|
||||
engine.container("text-completion")
|
||||
.with_image(images.trustgraph)
|
||||
|
|
@ -24,58 +26,28 @@ local chunker = import "chunker-recursive.jsonnet";
|
|||
"text-completion-bedrock",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-z",
|
||||
$["aws-id-key"],
|
||||
"-k",
|
||||
$["aws-secret-key"],
|
||||
"-r",
|
||||
$["aws-region"],
|
||||
"-x",
|
||||
std.toString($["bedrock-max-output-tokens"]),
|
||||
"-t",
|
||||
std.toString($["bedrock-temperature"]),
|
||||
"%0.3f" % $["bedrock-temperature"],
|
||||
"-m",
|
||||
$["bedrock-model"],
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
service,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
"text-completion-rag" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local container =
|
||||
local containerRag =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-bedrock",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-z",
|
||||
$["aws-id-key"],
|
||||
"-k",
|
||||
$["aws-secret-key"],
|
||||
"-r",
|
||||
$["aws-region"],
|
||||
"-x",
|
||||
std.toString($["bedrock-max-output-tokens"]),
|
||||
"-t",
|
||||
std.toString($["bedrock-temperature"]),
|
||||
"%0.3f" % $["bedrock-temperature"],
|
||||
"-m",
|
||||
$["bedrock-model"],
|
||||
"-i",
|
||||
|
|
@ -83,24 +55,35 @@ local chunker = import "chunker-recursive.jsonnet";
|
|||
"-o",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion-rag", [ container ]
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local containerSetRag = engine.containers(
|
||||
"text-completion-rag", [ containerRag ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
local serviceRag =
|
||||
engine.internalService(containerSetRag)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
envSecrets,
|
||||
containerSet,
|
||||
containerSetRag,
|
||||
service,
|
||||
serviceRag,
|
||||
])
|
||||
|
||||
|
||||
}
|
||||
},
|
||||
|
||||
} + prompts + chunker
|
||||
|
||||
|
|
|
|||
|
|
@ -24,7 +24,7 @@ cassandra + {
|
|||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"stop-triples", [ container ]
|
||||
"store-triples", [ container ]
|
||||
);
|
||||
|
||||
local service =
|
||||
|
|
|
|||
|
|
@ -5,7 +5,6 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
|
||||
{
|
||||
|
||||
"claude-key":: "${CLAUDE_KEY}",
|
||||
"claude-max-output-tokens":: 4096,
|
||||
"claude-temperature":: 0.0,
|
||||
|
||||
|
|
@ -13,6 +12,9 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
|
||||
create:: function(engine)
|
||||
|
||||
local envSecrets = engine.envSecrets("claude-credentials")
|
||||
.with_env_var("CLAUDE_KEY_TOKEN", "claude-key");
|
||||
|
||||
local container =
|
||||
engine.container("text-completion")
|
||||
.with_image(images.trustgraph)
|
||||
|
|
@ -20,13 +22,32 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
"text-completion-claude",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-k",
|
||||
$["claude-key"],
|
||||
"-x",
|
||||
std.toString($["claude-max-output-tokens"]),
|
||||
"-t",
|
||||
std.toString($["claude-temperature"]),
|
||||
"%0.3f" % $["claude-temperature"],
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerRag =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-claude",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-x",
|
||||
std.toString($["claude-max-output-tokens"]),
|
||||
"-t",
|
||||
"%0.3f" % $["claude-temperature"],
|
||||
"-i",
|
||||
"non-persistent://tg/request/text-completion-rag",
|
||||
"-o",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
|
|
@ -34,57 +55,27 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
service,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
"text-completion-rag" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local container =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-claude",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-k",
|
||||
$["claude-key"],
|
||||
"-x",
|
||||
std.toString($["claude-max-output-tokens"]),
|
||||
"-t",
|
||||
std.toString($["claude-temperature"]),
|
||||
"-i",
|
||||
"non-persistent://tg/request/text-completion-rag",
|
||||
"-o",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
])
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion-rag", [ container ]
|
||||
local containerSetRag = engine.containers(
|
||||
"text-completion-rag", [ containerRag ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
local serviceRag =
|
||||
engine.internalService(containerSetRag)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
envSecrets,
|
||||
containerSet,
|
||||
containerSetRag,
|
||||
service,
|
||||
serviceRag,
|
||||
])
|
||||
|
||||
|
||||
}
|
||||
},
|
||||
|
||||
} + prompts
|
||||
|
||||
|
|
|
|||
|
|
@ -9,13 +9,15 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
"chunk-size":: 150,
|
||||
"chunk-overlap":: 10,
|
||||
|
||||
"cohere-key":: "${COHERE_KEY}",
|
||||
"cohere-temperature":: 0.0,
|
||||
|
||||
"text-completion" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local envSecrets = engine.envSecrets("cohere-credentials")
|
||||
.with_env_var("COHERE_KEY", "cohere-key");
|
||||
|
||||
local container =
|
||||
engine.container("text-completion")
|
||||
.with_image(images.trustgraph)
|
||||
|
|
@ -23,44 +25,21 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
"text-completion-cohere",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-k",
|
||||
$["cohere-key"],
|
||||
"-t",
|
||||
std.toString($["cohere-temperature"]),
|
||||
"%0.3f" % $["cohere-temperature"],
|
||||
])
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
service,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
"text-completion-rag" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local container =
|
||||
local containerRag =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-cohere",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-k",
|
||||
$["cohere-key"],
|
||||
"-t",
|
||||
std.toString($["cohere-temperature"]),
|
||||
"%0.3f" % $["cohere-temperature"],
|
||||
"-i",
|
||||
"non-persistent://tg/request/text-completion-rag",
|
||||
"-o",
|
||||
|
|
@ -70,20 +49,30 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion-rag", [ container ]
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local containerSetRag = engine.containers(
|
||||
"text-completion-rag", [ containerRag ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
local serviceRag =
|
||||
engine.internalService(containerSetRag)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
envSecrets,
|
||||
containerSet,
|
||||
containerSetRag,
|
||||
service,
|
||||
serviceRag,
|
||||
])
|
||||
|
||||
|
||||
}
|
||||
},
|
||||
|
||||
} + prompts
|
||||
|
||||
|
|
|
|||
86
templates/components/googleaistudio.jsonnet
Normal file
86
templates/components/googleaistudio.jsonnet
Normal file
|
|
@ -0,0 +1,86 @@
|
|||
local base = import "base/base.jsonnet";
|
||||
local images = import "values/images.jsonnet";
|
||||
local url = import "values/url.jsonnet";
|
||||
local prompts = import "prompts/mixtral.jsonnet";
|
||||
|
||||
{
|
||||
|
||||
"googleaistudio-max-output-tokens":: 4096,
|
||||
"googleaistudio-temperature":: 0.0,
|
||||
"googleaistudio-model":: "gemini-1.5-flash-002",
|
||||
|
||||
"text-completion" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local envSecrets = engine.envSecrets("bedrock-credentials")
|
||||
.with_env_var("GOOGLE_AI_STUDIO_KEY", "googleaistudio-key");
|
||||
|
||||
local container =
|
||||
engine.container("text-completion")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-googleaistudio",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-x",
|
||||
std.toString($["googleaistudio-max-output-tokens"]),
|
||||
"-t",
|
||||
"%0.3f" % $["googleaistudio-temperature"],
|
||||
"-m",
|
||||
$["googleaistudio-model"],
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerRag =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-googleaistudio",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-x",
|
||||
std.toString($["googleaistudio-max-output-tokens"]),
|
||||
"-t",
|
||||
"%0.3f" % $["googleaistudio-temperature"],
|
||||
"-m",
|
||||
$["googleaistudio-model"],
|
||||
"-i",
|
||||
"non-persistent://tg/request/text-completion-rag",
|
||||
"-o",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local containerSetRag = engine.containers(
|
||||
"text-completion-rag", [ containerRag ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
local serviceRag =
|
||||
engine.internalService(containerSetRag)
|
||||
.with_port(8000, 8000, "metrics");
|
||||
|
||||
engine.resources([
|
||||
envSecrets,
|
||||
containerSet,
|
||||
containerSetRag,
|
||||
service,
|
||||
serviceRag,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
} + prompts
|
||||
|
||||
|
|
@ -6,12 +6,14 @@ local prompts = import "prompts/slm.jsonnet";
|
|||
{
|
||||
|
||||
"llamafile-model":: "LLaMA_CPP",
|
||||
"llamafile-url":: "${LLAMAFILE_URL}",
|
||||
|
||||
"text-completion" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local envSecrets = engine.envSecrets("llamafile-credentials")
|
||||
.with_env_var("LLAMAFILE_URL", "llamafile-url");
|
||||
|
||||
local container =
|
||||
engine.container("text-completion")
|
||||
.with_image(images.trustgraph)
|
||||
|
|
@ -21,27 +23,12 @@ local prompts = import "prompts/slm.jsonnet";
|
|||
url.pulsar,
|
||||
"-m",
|
||||
$["llamafile-model"],
|
||||
"-r",
|
||||
$["llamafile-url"],
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
"text-completion-rag" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local container =
|
||||
local containerRag =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
|
|
@ -50,26 +37,40 @@ local prompts = import "prompts/slm.jsonnet";
|
|||
url.pulsar,
|
||||
"-m",
|
||||
$["llamafile-model"],
|
||||
"-r",
|
||||
$["llamafile-url"],
|
||||
"-i",
|
||||
"non-persistent://tg/request/text-completion-rag",
|
||||
"-o",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion-rag", [ container ]
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local containerSetRag = engine.containers(
|
||||
"text-completion-rag", [ containerRag ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8080, 8080, "metrics");
|
||||
|
||||
local serviceRag =
|
||||
engine.internalService(containerSetRag)
|
||||
.with_port(8080, 8080, "metrics");
|
||||
|
||||
engine.resources([
|
||||
envSecrets,
|
||||
containerSet,
|
||||
containerSetRag,
|
||||
service,
|
||||
serviceRag,
|
||||
])
|
||||
|
||||
|
||||
}
|
||||
},
|
||||
|
||||
} + prompts
|
||||
|
||||
|
|
|
|||
|
|
@ -1,17 +1,19 @@
|
|||
local base = import "base/base.jsonnet";
|
||||
local images = import "values/images.jsonnet";
|
||||
local url = import "values/url.jsonnet";
|
||||
local prompts = import "prompts/slm.jsonnet";
|
||||
local prompts = import "prompts/mixtral.jsonnet";
|
||||
|
||||
{
|
||||
|
||||
"ollama-model":: "gemma2:9b",
|
||||
"ollama-url":: "${OLLAMA_HOST}",
|
||||
|
||||
"text-completion" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local envSecrets = engine.envSecrets("ollama-credentials")
|
||||
.with_env_var("OLLAMA_HOST", "ollama-host");
|
||||
|
||||
local container =
|
||||
engine.container("text-completion")
|
||||
.with_image(images.trustgraph)
|
||||
|
|
@ -21,32 +23,12 @@ local prompts = import "prompts/slm.jsonnet";
|
|||
url.pulsar,
|
||||
"-m",
|
||||
$["ollama-model"],
|
||||
"-r",
|
||||
$["ollama-url"],
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8080, 8080, "metrics");
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
service,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
"text-completion-rag" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local container =
|
||||
local containerRag =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
|
|
@ -55,31 +37,40 @@ local prompts = import "prompts/slm.jsonnet";
|
|||
url.pulsar,
|
||||
"-m",
|
||||
$["ollama-model"],
|
||||
"-r",
|
||||
$["ollama-url"],
|
||||
"-i",
|
||||
"non-persistent://tg/request/text-completion-rag",
|
||||
"-o",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion-rag", [ container ]
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local containerSetRag = engine.containers(
|
||||
"text-completion-rag", [ containerRag ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8080, 8080, "metrics");
|
||||
|
||||
local serviceRag =
|
||||
engine.internalService(containerSetRag)
|
||||
.with_port(8080, 8080, "metrics");
|
||||
|
||||
engine.resources([
|
||||
envSecrets,
|
||||
containerSet,
|
||||
containerSetRag,
|
||||
service,
|
||||
serviceRag,
|
||||
])
|
||||
|
||||
|
||||
}
|
||||
},
|
||||
|
||||
} + prompts
|
||||
|
||||
|
|
|
|||
|
|
@ -5,7 +5,6 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
|
||||
{
|
||||
|
||||
"openai-key":: "${OPENAI_KEY}",
|
||||
"openai-max-output-tokens":: 4096,
|
||||
"openai-temperature":: 0.0,
|
||||
"openai-model":: "GPT-3.5-Turbo",
|
||||
|
|
@ -14,6 +13,9 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
|
||||
create:: function(engine)
|
||||
|
||||
local envSecrets = engine.envSecrets("openai-credentials")
|
||||
.with_env_var("OPENAI_TOKEN", "openai-token");
|
||||
|
||||
local container =
|
||||
engine.container("text-completion")
|
||||
.with_image(images.trustgraph)
|
||||
|
|
@ -21,50 +23,28 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
"text-completion-openai",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-k",
|
||||
$["openai-key"],
|
||||
"-x",
|
||||
std.toString($["openai-max-output-tokens"]),
|
||||
"-t",
|
||||
std.toString($["openai-temperature"]),
|
||||
"%0.3f" % $["openai-temperature"],
|
||||
"-m",
|
||||
$["openai-model"],
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8080, 8080, "metrics");
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
service,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
"text-completion-rag" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local container =
|
||||
local containerRag =
|
||||
engine.container("text-completion-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"text-completion-openai",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-k",
|
||||
$["openai-key"],
|
||||
"-x",
|
||||
std.toString($["openai-max-output-tokens"]),
|
||||
"-t",
|
||||
std.toString($["openai-temperature"]),
|
||||
"%0.3f" % $["openai-temperature"],
|
||||
"-m",
|
||||
$["openai-model"],
|
||||
"-i",
|
||||
|
|
@ -72,24 +52,35 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
"-o",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
])
|
||||
.with_env_var_secrets(envSecrets)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"text-completion-rag", [ container ]
|
||||
"text-completion", [ container ]
|
||||
);
|
||||
|
||||
local containerSetRag = engine.containers(
|
||||
"text-completion-rag", [ containerRag ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8080, 8080, "metrics");
|
||||
|
||||
local serviceRag =
|
||||
engine.internalService(containerSetRag)
|
||||
.with_port(8080, 8080, "metrics");
|
||||
|
||||
engine.resources([
|
||||
envSecrets,
|
||||
containerSet,
|
||||
containerSetRag,
|
||||
service,
|
||||
serviceRag,
|
||||
])
|
||||
|
||||
|
||||
}
|
||||
},
|
||||
|
||||
} + prompts
|
||||
|
||||
|
|
|
|||
|
|
@ -1,81 +0,0 @@
|
|||
local base = import "base/base.jsonnet";
|
||||
local images = import "values/images.jsonnet";
|
||||
local url = import "values/url.jsonnet";
|
||||
local prompts = import "prompts/mixtral.jsonnet";
|
||||
|
||||
{
|
||||
|
||||
"prompt" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local container =
|
||||
engine.container("prompt")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"prompt-generic",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"--text-completion-request-queue",
|
||||
"non-persistent://tg/request/text-completion",
|
||||
"--text-completion-response-queue",
|
||||
"non-persistent://tg/response/text-completion-response",
|
||||
])
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"prompt", [ container ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8080, 8080, "metrics");
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
service,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
"prompt-rag" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
||||
local container =
|
||||
engine.container("prompt-rag")
|
||||
.with_image(images.trustgraph)
|
||||
.with_command([
|
||||
"prompt-generic",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
"-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",
|
||||
])
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
local containerSet = engine.containers(
|
||||
"prompt-rag", [ container ]
|
||||
);
|
||||
|
||||
local service =
|
||||
engine.internalService(containerSet)
|
||||
.with_port(8080, 8080, "metrics");
|
||||
|
||||
engine.resources([
|
||||
containerSet,
|
||||
service,
|
||||
])
|
||||
|
||||
},
|
||||
|
||||
}
|
||||
|
||||
28
templates/components/prompt-overrides.jsonnet
Normal file
28
templates/components/prompt-overrides.jsonnet
Normal file
|
|
@ -0,0 +1,28 @@
|
|||
local base = import "base/base.jsonnet";
|
||||
local images = import "values/images.jsonnet";
|
||||
local url = import "values/url.jsonnet";
|
||||
local prompts = import "prompts/mixtral.jsonnet";
|
||||
local default_prompts = import "prompts/default-prompts.jsonnet";
|
||||
|
||||
{
|
||||
|
||||
with:: function(key, value)
|
||||
if (key == "system-template") then
|
||||
self + {
|
||||
prompts +:: {
|
||||
"system-template": value,
|
||||
}
|
||||
}
|
||||
else
|
||||
self + {
|
||||
prompts +:: {
|
||||
templates +:: {
|
||||
[key] +:: {
|
||||
prompt: value
|
||||
}
|
||||
}
|
||||
}
|
||||
},
|
||||
|
||||
} + default_prompts
|
||||
|
||||
|
|
@ -6,6 +6,38 @@ local default_prompts = import "prompts/default-prompts.jsonnet";
|
|||
|
||||
{
|
||||
|
||||
prompts:: default_prompts,
|
||||
|
||||
local prompt_template_args = [ "--prompt" ] + [
|
||||
p.key + "=" + p.value.prompt,
|
||||
for p in std.objectKeysValuesAll($.prompts.templates)
|
||||
],
|
||||
|
||||
local prompt_response_type_args = [ "--prompt-response-type" ] + [
|
||||
p.key + "=" + p.value["response-type"],
|
||||
for p in std.objectKeysValuesAll($.prompts.templates)
|
||||
if std.objectHas(p.value, "response-type")
|
||||
],
|
||||
|
||||
local prompt_schema_args = [ "--prompt-schema" ] + [
|
||||
(
|
||||
p.key + "=" +
|
||||
std.manifestJsonMinified(p.value["schema"])
|
||||
)
|
||||
for p in std.objectKeysValuesAll($.prompts.templates)
|
||||
if std.objectHas(p.value, "schema")
|
||||
],
|
||||
|
||||
local prompt_term_args = [ "--prompt-term" ] + [
|
||||
p.key + "=" + t.key + ":" + t.value
|
||||
for p in std.objectKeysValuesAll($.prompts.templates)
|
||||
if std.objectHas(p.value, "terms")
|
||||
for t in std.objectKeysValuesAll(p.value.terms)
|
||||
],
|
||||
|
||||
local prompt_args = prompt_template_args + prompt_response_type_args +
|
||||
prompt_schema_args + prompt_term_args,
|
||||
|
||||
"prompt" +: {
|
||||
|
||||
create:: function(engine)
|
||||
|
|
@ -17,23 +49,17 @@ local default_prompts = import "prompts/default-prompts.jsonnet";
|
|||
"prompt-template",
|
||||
"-p",
|
||||
url.pulsar,
|
||||
|
||||
"--text-completion-request-queue",
|
||||
"non-persistent://tg/request/text-completion",
|
||||
"--text-completion-response-queue",
|
||||
"non-persistent://tg/response/text-completion-response",
|
||||
"--definition-template",
|
||||
$["prompt-definition-template"],
|
||||
"--relationship-template",
|
||||
$["prompt-relationship-template"],
|
||||
"--topic-template",
|
||||
$["prompt-topic-template"],
|
||||
"--knowledge-query-template",
|
||||
$["prompt-knowledge-query-template"],
|
||||
"--document-query-template",
|
||||
$["prompt-document-query-template"],
|
||||
"--rows-template",
|
||||
$["prompt-rows-template"],
|
||||
])
|
||||
|
||||
"--system-prompt",
|
||||
$["prompts"]["system-template"],
|
||||
|
||||
] + prompt_args
|
||||
)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
|
|
@ -71,19 +97,12 @@ local default_prompts = import "prompts/default-prompts.jsonnet";
|
|||
"non-persistent://tg/request/text-completion-rag",
|
||||
"--text-completion-response-queue",
|
||||
"non-persistent://tg/response/text-completion-rag-response",
|
||||
"--definition-template",
|
||||
$["prompt-definition-template"],
|
||||
"--relationship-template",
|
||||
$["prompt-relationship-template"],
|
||||
"--topic-template",
|
||||
$["prompt-topic-template"],
|
||||
"--knowledge-query-template",
|
||||
$["prompt-knowledge-query-template"],
|
||||
"--document-query-template",
|
||||
$["prompt-document-query-template"],
|
||||
"--rows-template",
|
||||
$["prompt-rows-template"],
|
||||
])
|
||||
|
||||
"--system-prompt",
|
||||
$["prompts"]["system-template"],
|
||||
|
||||
] + prompt_args
|
||||
)
|
||||
.with_limits("0.5", "128M")
|
||||
.with_reservations("0.1", "128M");
|
||||
|
||||
|
|
|
|||
|
|
@ -37,7 +37,7 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
"-x",
|
||||
std.toString($["vertexai-max-output-tokens"]),
|
||||
"-t",
|
||||
std.toString($["vertexai-temperature"]),
|
||||
"%0.3f" % $["vertexai-temperature"],
|
||||
"-m",
|
||||
$["vertexai-model"],
|
||||
])
|
||||
|
|
@ -87,7 +87,7 @@ local prompts = import "prompts/mixtral.jsonnet";
|
|||
"-x",
|
||||
std.toString($["vertexai-max-output-tokens"]),
|
||||
"-t",
|
||||
std.toString($["vertexai-temperature"]),
|
||||
"%0.3f" % $["vertexai-temperature"],
|
||||
"-m",
|
||||
$["vertexai-model"],
|
||||
"-i",
|
||||
|
|
|
|||
|
|
@ -18,12 +18,15 @@
|
|||
reservations: {},
|
||||
ports: [],
|
||||
volumes: [],
|
||||
environment: {},
|
||||
|
||||
with_image:: function(x) self + { image: x },
|
||||
|
||||
with_command:: function(x) self + { command: x },
|
||||
|
||||
with_environment:: function(x) self + { environment: x },
|
||||
with_environment:: function(x) self + {
|
||||
environment: super.environment + x,
|
||||
},
|
||||
|
||||
with_limits:: function(c, m) self + { limits: { cpus: c, memory: m } },
|
||||
|
||||
|
|
@ -45,6 +48,16 @@
|
|||
]
|
||||
},
|
||||
|
||||
with_env_var_secrets::
|
||||
function(vars)
|
||||
std.foldl(
|
||||
function(obj, x) obj.with_environment(
|
||||
{ [x]: "${" + x + "}" }
|
||||
),
|
||||
vars.variables,
|
||||
self
|
||||
),
|
||||
|
||||
add:: function() {
|
||||
services +: {
|
||||
[container.name]: {
|
||||
|
|
@ -62,7 +75,7 @@
|
|||
{ command: container.command }
|
||||
else {}) +
|
||||
|
||||
(if std.objectHas(container, "environment") then
|
||||
(if ! std.isEmpty(container.environment) then
|
||||
{ environment: container.environment }
|
||||
else {}) +
|
||||
|
||||
|
|
@ -170,6 +183,27 @@
|
|||
|
||||
},
|
||||
|
||||
envSecrets:: function(name)
|
||||
{
|
||||
|
||||
local volume = self,
|
||||
|
||||
name: name,
|
||||
|
||||
volid:: name,
|
||||
|
||||
variables:: [],
|
||||
|
||||
with_env_var::
|
||||
function(name, key) self + {
|
||||
variables: super.variables + [name],
|
||||
},
|
||||
|
||||
add:: function() {
|
||||
}
|
||||
|
||||
},
|
||||
|
||||
containers:: function(name, containers)
|
||||
{
|
||||
|
||||
|
|
|
|||
|
|
@ -10,12 +10,20 @@
|
|||
reservations: {},
|
||||
ports: [],
|
||||
volumes: [],
|
||||
environment: [],
|
||||
|
||||
with_image:: function(x) self + { image: x },
|
||||
|
||||
with_command:: function(x) self + { command: x },
|
||||
|
||||
with_environment:: function(x) self + { environment: x },
|
||||
with_environment:: function(x) self + {
|
||||
environment: super.environment + [
|
||||
{
|
||||
name: v.key, value: v.value
|
||||
}
|
||||
for v in std.objectKeysValues(x)
|
||||
],
|
||||
},
|
||||
|
||||
with_limits:: function(c, m) self + { limits: { cpu: c, memory: m } },
|
||||
|
||||
|
|
@ -37,6 +45,24 @@
|
|||
]
|
||||
},
|
||||
|
||||
with_env_var_secrets::
|
||||
function(vars)
|
||||
std.foldl(
|
||||
function(obj, x) obj + {
|
||||
environment: super.environment + [{
|
||||
name: x,
|
||||
valueFrom: {
|
||||
secretKeyRef: {
|
||||
name: vars.name,
|
||||
key: vars.keyMap[x],
|
||||
}
|
||||
}
|
||||
}]
|
||||
},
|
||||
vars.variables,
|
||||
self
|
||||
),
|
||||
|
||||
add:: function() [
|
||||
|
||||
{
|
||||
|
|
@ -97,16 +123,11 @@
|
|||
(if std.objectHas(container, "command") then
|
||||
{ command: container.command }
|
||||
else {}) +
|
||||
(if std.objectHas(container, "environment") then
|
||||
{ env: [ {
|
||||
name: e.key, value: e.value
|
||||
}
|
||||
for e in
|
||||
std.objectKeysValues(
|
||||
container.environment
|
||||
)
|
||||
]
|
||||
}
|
||||
|
||||
(if ! std.isEmpty(container.environment) then
|
||||
{
|
||||
env: container.environment,
|
||||
}
|
||||
else {}) +
|
||||
|
||||
(if std.length(container.volumes) > 0 then
|
||||
|
|
@ -283,6 +304,34 @@
|
|||
|
||||
},
|
||||
|
||||
envSecrets:: function(name)
|
||||
{
|
||||
|
||||
local volume = self,
|
||||
|
||||
name: name,
|
||||
|
||||
variables: [],
|
||||
keyMap: {},
|
||||
|
||||
with_size:: function(size) self + { size: size },
|
||||
|
||||
add:: function() [
|
||||
],
|
||||
|
||||
volRef:: function() {
|
||||
name: volume.name,
|
||||
secret: { secretName: volume.name },
|
||||
},
|
||||
|
||||
with_env_var::
|
||||
function(name, key) self + {
|
||||
variables: super.variables + [name],
|
||||
keyMap: super.keyMap + { [name]: key },
|
||||
},
|
||||
|
||||
},
|
||||
|
||||
containers:: function(name, containers)
|
||||
{
|
||||
|
||||
|
|
|
|||
140
templates/generate
Executable file
140
templates/generate
Executable file
|
|
@ -0,0 +1,140 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
import _jsonnet as j
|
||||
import json
|
||||
import yaml
|
||||
import logging
|
||||
import os
|
||||
import sys
|
||||
import zipfile
|
||||
|
||||
logger = logging.getLogger("generate")
|
||||
logging.basicConfig(level=logging.INFO, format='%(message)s')
|
||||
|
||||
private_json = "Put your GCP private.json here"
|
||||
|
||||
class Generator:
|
||||
|
||||
def __init__(self, config, base="./templates/", version="0.0.0"):
|
||||
|
||||
self.jsonnet_base = base
|
||||
self.config = config
|
||||
self.version = f"\"{version}\"".encode("utf-8")
|
||||
|
||||
def process(self, config):
|
||||
|
||||
res = j.evaluate_snippet("config", config, import_callback=self.load)
|
||||
return json.loads(res)
|
||||
|
||||
def load(self, dir, filename):
|
||||
|
||||
logger.debug("Request jsonnet: %s %s", dir, filename)
|
||||
|
||||
if filename == "config.json" and dir == "":
|
||||
path = os.path.join(".", dir, filename)
|
||||
return str(path), self.config
|
||||
|
||||
if filename == "version.jsonnet" and dir == "./templates/values/":
|
||||
path = os.path.join(".", dir, filename)
|
||||
return str(path), self.version
|
||||
|
||||
if dir:
|
||||
candidates = [
|
||||
os.path.join(".", dir, filename),
|
||||
os.path.join(".", filename)
|
||||
]
|
||||
else:
|
||||
candidates = [
|
||||
os.path.join(".", filename)
|
||||
]
|
||||
|
||||
try:
|
||||
|
||||
if filename == "vertexai/private.json":
|
||||
|
||||
return candidates[0], private_json.encode("utf-8")
|
||||
|
||||
for c in candidates:
|
||||
logger.debug("Try: %s", c)
|
||||
|
||||
if os.path.isfile(c):
|
||||
with open(c, "rb") as f:
|
||||
logger.debug("Loading: %s", c)
|
||||
return str(c), f.read()
|
||||
|
||||
raise RuntimeError(
|
||||
f"Could not load file={filename} dir={dir}"
|
||||
)
|
||||
|
||||
except:
|
||||
|
||||
path = os.path.join(self.jsonnet_base, filename)
|
||||
logger.debug("Try: %s", path)
|
||||
with open(path, "rb") as f:
|
||||
logger.debug("Loaded: %s", path)
|
||||
return str(path), f.read()
|
||||
|
||||
def main():
|
||||
|
||||
if len(sys.argv) != 3:
|
||||
print()
|
||||
print("Usage:")
|
||||
print(" generate <outfile> <version> < input.json")
|
||||
print()
|
||||
raise RuntimeError("Arg error")
|
||||
|
||||
outfile = sys.argv[1]
|
||||
version = sys.argv[2]
|
||||
|
||||
cfg = sys.stdin.read()
|
||||
cfg = json.loads(cfg)
|
||||
|
||||
logger.info(f"Outputting to {outfile}...")
|
||||
|
||||
with zipfile.ZipFile(outfile, mode='w') as out:
|
||||
|
||||
def output(name, content):
|
||||
logger.info(f"Adding {name}...")
|
||||
out.writestr(name, content)
|
||||
|
||||
fname = "tg-launch.yaml"
|
||||
|
||||
platform = "docker-compose"
|
||||
|
||||
with open(f"./templates/config-to-{platform}.jsonnet", "r") as f:
|
||||
wrapper = f.read()
|
||||
|
||||
gen = Generator(json.dumps(cfg).encode("utf-8"), version=version)
|
||||
|
||||
processed = gen.process(wrapper)
|
||||
|
||||
y = yaml.dump(processed)
|
||||
|
||||
output(fname, y)
|
||||
|
||||
# Placeholder for the private.json file. Won't put actual credentials
|
||||
# here.
|
||||
output("docker-compose/vertexai/private.json", private_json)
|
||||
|
||||
# Grafana config
|
||||
with open("grafana/dashboards/dashboard.json") as f:
|
||||
output(
|
||||
"docker-compose/grafana/dashboards/dashboard.json", f.read()
|
||||
)
|
||||
|
||||
with open("grafana/provisioning/dashboard.yml") as f:
|
||||
output(
|
||||
"docker-compose/grafana/provisioning/dashboard.yml", f.read()
|
||||
)
|
||||
|
||||
with open("grafana/provisioning/datasource.yml") as f:
|
||||
output(
|
||||
"docker-compose/grafana/provisioning/datasource.yml", f.read()
|
||||
)
|
||||
|
||||
# Prometheus config
|
||||
with open("prometheus/prometheus.yml") as f:
|
||||
output("docker-compose/prometheus/prometheus.yml", f.read())
|
||||
|
||||
main()
|
||||
|
||||
|
|
@ -122,8 +122,8 @@ def generate_all(output, version):
|
|||
"docker-compose", "minikube-k8s", "gcp-k8s"
|
||||
]:
|
||||
for model in [
|
||||
"azure", "bedrock", "claude", "cohere", "llamafile", "ollama",
|
||||
"openai", "vertexai"
|
||||
"azure", "azure-openai", "bedrock", "claude", "cohere",
|
||||
"googleaistudio", "llamafile", "ollama", "openai", "vertexai",
|
||||
]:
|
||||
for graph in [ "cassandra", "neo4j" ]:
|
||||
|
||||
|
|
|
|||
32
templates/patterns/llm-azure-openai.jsonnet
Normal file
32
templates/patterns/llm-azure-openai.jsonnet
Normal file
|
|
@ -0,0 +1,32 @@
|
|||
{
|
||||
pattern: {
|
||||
name: "azure-openai",
|
||||
icon: "🤖💬",
|
||||
title: "Add Azure OpenAI LLM endpoint for text completion",
|
||||
description: "This pattern integrates an Azure OpenAI LLM endpoint hosted in the Azure cloud for text completion operations. You need an Azure subscription to be able to use this service.",
|
||||
requires: ["pulsar", "trustgraph"],
|
||||
features: ["llm"],
|
||||
args: [
|
||||
{
|
||||
name: "azure-openai-max-output-tokens",
|
||||
label: "Maximum output tokens",
|
||||
type: "integer",
|
||||
description: "Limit on number tokens to generate",
|
||||
default: 4096,
|
||||
required: true,
|
||||
},
|
||||
{
|
||||
name: "azure-openai-temperature",
|
||||
label: "Temperature",
|
||||
type: "slider",
|
||||
description: "Controlling predictability / creativity balance",
|
||||
min: 0,
|
||||
max: 1,
|
||||
step: 0.05,
|
||||
default: 0.5,
|
||||
},
|
||||
],
|
||||
category: [ "llm" ],
|
||||
},
|
||||
module: "components/azure.jsonnet",
|
||||
}
|
||||
32
templates/patterns/llm-googleaistudio.jsonnet
Normal file
32
templates/patterns/llm-googleaistudio.jsonnet
Normal file
|
|
@ -0,0 +1,32 @@
|
|||
{
|
||||
pattern: {
|
||||
name: "googleaistudio",
|
||||
icon: "🤖💬",
|
||||
title: "Add GoogleAIStudio for text completion",
|
||||
description: "This pattern integrates a GoogleAIStudio LLM service for text completion operations. You need a GoogleAISTudio API key to be able to use this service.",
|
||||
requires: ["pulsar", "trustgraph"],
|
||||
features: ["llm"],
|
||||
args: [
|
||||
{
|
||||
name: "googleaistudio-max-output-tokens",
|
||||
label: "Maximum output tokens",
|
||||
type: "integer",
|
||||
description: "Limit on number tokens to generate",
|
||||
default: 4096,
|
||||
required: true,
|
||||
},
|
||||
{
|
||||
name: "googleaistudio-temperature",
|
||||
label: "Temperature",
|
||||
type: "slider",
|
||||
description: "Controlling predictability / creativity balance",
|
||||
min: 0,
|
||||
max: 1,
|
||||
step: 0.05,
|
||||
default: 0.5,
|
||||
},
|
||||
],
|
||||
category: [ "llm" ],
|
||||
},
|
||||
module: "components/googleaistudio.jsonnet",
|
||||
}
|
||||
|
|
@ -1,18 +1,42 @@
|
|||
|
||||
// For Cohere. Not currently overriding prompts
|
||||
|
||||
{
|
||||
local prompts = import "default-prompts.jsonnet";
|
||||
|
||||
// "prompt-definition-template": "PROMPT GOES HERE",
|
||||
prompts + {
|
||||
|
||||
// "prompt-relationship-template":: "PROMPT GOES HERE",
|
||||
// "system-template":: "PROMPT GOES HERE.",
|
||||
|
||||
// "prompt-topic-template":: "PROMPT GOES HERE",
|
||||
"templates" +:: {
|
||||
|
||||
// "prompt-knowledge-query-template":: "PROMPT GOES HERE",
|
||||
"question" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
// "prompt-document-query-template":: "PROMPT GOES HERE",
|
||||
"extract-definitions" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
// "prompt-rows-template":: "PROMPT GOES HERE",
|
||||
"extract-relationships" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-topics" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-rows" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"kg-prompt" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"document-prompt" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -4,16 +4,111 @@
|
|||
|
||||
{
|
||||
|
||||
"prompt-definition-template":: "<instructions>\nStudy the following text and derive definitions for any discovered entities.\nDo not provide definitions for entities whose definitions are incomplete\nor unknown.\nOutput relationships in JSON format as an arary of objects with fields:\n- entity: the name of the entity\n- definition: English text which defines the entity\n</instructions>\n\n<text>\n{text}\n</text>\n\n<requirements>\nYou will respond only with raw JSON format data. Do not provide\nexplanations. Do not use special characters in the abstract text. The\nabstract will be written as plain text. Do not add markdown formatting\nor headers or prefixes. Do not include null or unknown definitions.\n</requirements>",
|
||||
"system-template":: "You are a helpful assistant.",
|
||||
|
||||
"prompt-relationship-template":: "<instructions>\nStudy the following text and derive entity relationships. For each\nrelationship, derive the subject, predicate and object of the relationship.\nOutput relationships in JSON format as an arary of objects with fields:\n- subject: the subject of the relationship\n- predicate: the predicate\n- object: the object of the relationship\n- object-entity: false if the object is a simple data type: name, value or date. true if it is an entity.\n</instructions>\n\n<text>\n{text}\n</text>\n\n<requirements>\nYou will respond only with raw JSON format data. Do not provide\nexplanations. Do not use special characters in the abstract text. The\nabstract must be written as plain text. Do not add markdown formatting\nor headers or prefixes.\n</requirements>",
|
||||
"templates":: {
|
||||
|
||||
"prompt-topic-template":: "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\nHere 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.",
|
||||
"question":: {
|
||||
"prompt": "{{question}}",
|
||||
},
|
||||
|
||||
"prompt-knowledge-query-template":: "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.\n\nHere's the knowledge statements:\n{graph}\n\nUse only the provided knowledge statements to respond to the following:\n{query}\n",
|
||||
"extract-definitions":: {
|
||||
"prompt": "<instructions>\nStudy the following text and derive definitions for any discovered entities.\nDo not provide definitions for entities whose definitions are incomplete\nor unknown.\nOutput relationships in JSON format as an arary of objects with fields:\n- entity: the name of the entity\n- definition: English text which defines the entity\n</instructions>\n\n<text>\n{{text}}\n</text>\n\n<requirements>\nYou will respond only with raw JSON format data. Do not provide\nexplanations. Do not use special characters in the abstract text. The\nabstract will be written as plain text. Do not add markdown formatting\nor headers or prefixes. Do not include null or unknown definitions.\n</requirements>",
|
||||
"response-type": "json",
|
||||
"schema": {
|
||||
"type": "array",
|
||||
"items": {
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"entity": {
|
||||
"type": "string"
|
||||
},
|
||||
"definition": {
|
||||
"type": "string"
|
||||
}
|
||||
},
|
||||
"required": [
|
||||
"entity",
|
||||
"definition"
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
|
||||
"prompt-document-query-template":: "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.\n\nHere is the context:\n{documents}\n\nUse only the provided knowledge statements to respond to the following:\n{query}\n",
|
||||
"extract-relationships":: {
|
||||
"prompt": "<instructions>\nStudy the following text and derive entity relationships. For each\nrelationship, derive the subject, predicate and object of the relationship.\nOutput relationships in JSON format as an arary of objects with fields:\n- subject: the subject of the relationship\n- predicate: the predicate\n- object: the object of the relationship\n- object-entity: false if the object is a simple data type: name, value or date. true if it is an entity.\n</instructions>\n\n<text>\n{{text}}\n</text>\n\n<requirements>\nYou will respond only with raw JSON format data. Do not provide\nexplanations. Do not use special characters in the abstract text. The\nabstract must be written as plain text. Do not add markdown formatting\nor headers or prefixes.\n</requirements>",
|
||||
"response-type": "json",
|
||||
"schema": {
|
||||
"type": "array",
|
||||
"items": {
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"subject": {
|
||||
"type": "string"
|
||||
},
|
||||
"predicate": {
|
||||
"type": "string"
|
||||
},
|
||||
"object": {
|
||||
"type": "string"
|
||||
},
|
||||
"object-entity": {
|
||||
"type": "boolean"
|
||||
},
|
||||
},
|
||||
"required": [
|
||||
"subject",
|
||||
"predicate",
|
||||
"object",
|
||||
"object-entity"
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
|
||||
"prompt-rows-template":: "<instructions>\nStudy the following text and derive objects which match the schema provided.\n\nYou must output an array of JSON objects for each object you discover\nwhich matches the schema. For each object, output a JSON object whose fields\ncarry the name field specified in the schema.\n</instructions>\n\n<schema>\n{schema}\n</schema>\n\n<text>\n{text}\n</text>\n\n<requirements>\nYou will respond only with raw JSON format data. Do not provide\nexplanations. Do not add markdown formatting or headers or prefixes.\n</requirements>",
|
||||
"extract-topics":: {
|
||||
"prompt": "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\nHere 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.",
|
||||
"response-type": "json",
|
||||
"schema": {
|
||||
"type": "array",
|
||||
"items": {
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"topic": {
|
||||
"type": "string"
|
||||
},
|
||||
"definition": {
|
||||
"type": "string"
|
||||
}
|
||||
},
|
||||
"required": [
|
||||
"topic",
|
||||
"definition"
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
|
||||
"extract-rows":: {
|
||||
"prompt": "<instructions>\nStudy the following text and derive objects which match the schema provided.\n\nYou must output an array of JSON objects for each object you discover\nwhich matches the schema. For each object, output a JSON object whose fields\ncarry the name field specified in the schema.\n</instructions>\n\n<schema>\n{{schema}}\n</schema>\n\n<text>\n{{text}}\n</text>\n\n<requirements>\nYou will respond only with raw JSON format data. Do not provide\nexplanations. Do not add markdown formatting or headers or prefixes.\n</requirements>",
|
||||
"response-type": "json",
|
||||
},
|
||||
|
||||
"kg-prompt":: {
|
||||
"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.\n\nHere's the knowledge statements:\n{% for edge in knowledge %}({{edge.s}})-[{{edge.p}}]->({{edge.o}})\n{%endfor%}\n\nUse only the provided knowledge statements to respond to the following:\n{{query}}\n",
|
||||
"response-type": "text",
|
||||
},
|
||||
|
||||
"document-prompt":: {
|
||||
"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.\n\nHere is the context:\n{{documents}}\n\nUse only the provided knowledge statements to respond to the following:\n{{query}}\n",
|
||||
"response-type": "text",
|
||||
},
|
||||
|
||||
"agent-react":: {
|
||||
"prompt": "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 %}",
|
||||
"response-type": "json"
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
|
|
@ -1,17 +1,42 @@
|
|||
|
||||
// For VertexAI Gemini. Not currently overriding prompts
|
||||
{
|
||||
|
||||
// "prompt-definition-template": "PROMPT GOES HERE",
|
||||
local prompts = import "default-prompts.jsonnet";
|
||||
|
||||
// "prompt-relationship-template":: "PROMPT GOES HERE",
|
||||
prompts + {
|
||||
|
||||
// "prompt-topic-template":: "PROMPT GOES HERE",
|
||||
// "system-template":: "PROMPT GOES HERE.",
|
||||
|
||||
// "prompt-knowledge-query-template":: "PROMPT GOES HERE",
|
||||
"templates" +:: {
|
||||
|
||||
// "prompt-document-query-template":: "PROMPT GOES HERE",
|
||||
"question" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
// "prompt-rows-template":: "PROMPT GOES HERE",
|
||||
"extract-definitions" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-relationships" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-topics" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-rows" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"kg-prompt" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"document-prompt" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -1,18 +1,42 @@
|
|||
|
||||
// For Mixtral. Not currently overriding prompts
|
||||
|
||||
{
|
||||
local prompts = import "default-prompts.jsonnet";
|
||||
|
||||
// "prompt-definition-template": "PROMPT GOES HERE",
|
||||
prompts + {
|
||||
|
||||
// "prompt-relationship-template":: "PROMPT GOES HERE",
|
||||
// "system-template":: "PROMPT GOES HERE.",
|
||||
|
||||
// "prompt-topic-template":: "PROMPT GOES HERE",
|
||||
"templates" +:: {
|
||||
|
||||
// "prompt-knowledge-query-template":: "PROMPT GOES HERE",
|
||||
"question" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
// "prompt-document-query-template":: "PROMPT GOES HERE",
|
||||
"extract-definitions" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
// "prompt-rows-template":: "PROMPT GOES HERE",
|
||||
"extract-relationships" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-topics" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-rows" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"kg-prompt" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"document-prompt" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -1,23 +1,42 @@
|
|||
// For OpenAI LLMs. Not currently overriding prompts
|
||||
|
||||
// For OpenAI LLMs
|
||||
local prompts = import "default-prompts.jsonnet";
|
||||
|
||||
local base = import "base/base.jsonnet";
|
||||
local images = import "values/images.jsonnet";
|
||||
local url = import "values/url.jsonnet";
|
||||
prompts + {
|
||||
|
||||
{
|
||||
// "system-template":: "PROMPT GOES HERE.",
|
||||
|
||||
// "prompt-definition-template": "PROMPT GOES HERE",
|
||||
"templates" +:: {
|
||||
|
||||
// "prompt-relationship-template":: "PROMPT GOES HERE",
|
||||
"question" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
// "prompt-topic-template":: "PROMPT GOES HERE",
|
||||
"extract-definitions" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
// "prompt-knowledge-query-template":: "PROMPT GOES HERE",
|
||||
"extract-relationships" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
// "prompt-document-query-template":: "PROMPT GOES HERE",
|
||||
"extract-topics" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
// "prompt-rows-template":: "PROMPT GOES HERE",
|
||||
"extract-rows" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"kg-prompt" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"document-prompt" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -1,7 +1,44 @@
|
|||
// For SLM. Not currently overriding prompts
|
||||
|
||||
// For basic SLMs, use prompt-generic
|
||||
local prompts = import "default-prompts.jsonnet";
|
||||
|
||||
prompts + {
|
||||
|
||||
// "system-template":: "PROMPT GOES HERE.",
|
||||
|
||||
"templates" +:: {
|
||||
|
||||
"question" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-definitions" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-relationships" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-topics" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"extract-rows" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"kg-prompt" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
"document-prompt" +:: {
|
||||
// "prompt": "PROMPT GOES HERE",
|
||||
},
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
local prompts = import "components/prompt-generic.jsonnet";
|
||||
|
||||
prompts
|
||||
|
||||
|
|
|
|||
|
|
@ -13,10 +13,10 @@ local images = import "values/images.jsonnet";
|
|||
engine.container("cassandra")
|
||||
.with_image(images.cassandra)
|
||||
.with_environment({
|
||||
JVM_OPTS: "-Xms256M -Xmx256M",
|
||||
JVM_OPTS: "-Xms300M -Xmx300M",
|
||||
})
|
||||
.with_limits("1.0", "800M")
|
||||
.with_reservations("0.5", "800M")
|
||||
.with_limits("1.0", "1000M")
|
||||
.with_reservations("0.5", "1000M")
|
||||
.with_port(9042, 9042, "cassandra")
|
||||
.with_volume_mount(vol, "/var/lib/cassandra");
|
||||
|
||||
|
|
|
|||
|
|
@ -37,7 +37,7 @@ local images = import "values/images.jsonnet";
|
|||
|
||||
local service =
|
||||
engine.service(containerSet)
|
||||
.with_port(2379, 2379, 30379, "api");
|
||||
.with_port(2379, 2379, "api");
|
||||
|
||||
engine.resources([
|
||||
vol,
|
||||
|
|
@ -117,7 +117,7 @@ local images = import "values/images.jsonnet";
|
|||
local service =
|
||||
engine.service(containerSet)
|
||||
.with_port(9091, 9091, "api")
|
||||
.with_port(19530, 19530, "api2);
|
||||
.with_port(19530, 19530, "api2");
|
||||
|
||||
engine.resources([
|
||||
vol,
|
||||
|
|
|
|||
|
|
@ -12,8 +12,8 @@ local images = import "values/images.jsonnet";
|
|||
local container =
|
||||
engine.container("qdrant")
|
||||
.with_image(images.qdrant)
|
||||
.with_limits("1.0", "256M")
|
||||
.with_reservations("0.5", "256M")
|
||||
.with_limits("1.0", "1024M")
|
||||
.with_reservations("0.5", "1024M")
|
||||
.with_port(6333, 6333, "api")
|
||||
.with_port(6334, 6334, "api2")
|
||||
.with_volume_mount(vol, "/qdrant/storage");
|
||||
|
|
|
|||
|
|
@ -3,9 +3,7 @@ local components = import "components.jsonnet";
|
|||
|
||||
local apply = function(p, components)
|
||||
|
||||
local component = components[p.name];
|
||||
|
||||
(component + {
|
||||
local base = {
|
||||
|
||||
with:: function(k, v) self + {
|
||||
[k]:: v
|
||||
|
|
@ -18,7 +16,11 @@ local apply = function(p, components)
|
|||
self
|
||||
),
|
||||
|
||||
}).with_params(p.parameters);
|
||||
};
|
||||
|
||||
local component = base + components[p.name];
|
||||
|
||||
component.with_params(p.parameters);
|
||||
|
||||
local decode = function(config)
|
||||
local add = function(state, c) state + apply(c, components);
|
||||
|
|
|
|||
27
tests/README.prompts
Normal file
27
tests/README.prompts
Normal file
|
|
@ -0,0 +1,27 @@
|
|||
|
||||
test-prompt-... is tested with this prompt set...
|
||||
|
||||
prompt-template \
|
||||
-p pulsar://localhost:6650 \
|
||||
--system-prompt 'You are a {{attitude}}, you are called {{name}}' \
|
||||
--global-term \
|
||||
'name=Craig' \
|
||||
'attitude=LOUD, SHOUTY ANNOYING BOT' \
|
||||
--prompt \
|
||||
'question={{question}}' \
|
||||
'french-question={{question}}' \
|
||||
"analyze=Find the name and age in this text, and output a JSON structure containing just the name and age fields: {{description}}. Don't add markup, just output the raw JSON object." \
|
||||
"graph-query=Study the following knowledge graph, and then answer the question.\\n\nGraph:\\n{% for edge in knowledge %}({{edge.0}})-[{{edge.1}}]->({{edge.2}})\\n{%endfor%}\\nQuestion:\\n{{question}}" \
|
||||
"extract-definition=Analyse the text provided, and then return a list of terms and definitions. The output should be a JSON array, each item in the array is an object with fields 'term' and 'definition'.Don't add markup, just output the raw JSON object. Here is the text:\\n{{text}}" \
|
||||
--prompt-response-type \
|
||||
'question=text' \
|
||||
'analyze=json' \
|
||||
'graph-query=text' \
|
||||
'extract-definition=json' \
|
||||
--prompt-term \
|
||||
'question=name:Bonny' \
|
||||
'french-question=attitude:French-speaking bot' \
|
||||
--prompt-schema \
|
||||
'analyze={ "type" : "object", "properties" : { "age": { "type" : "number" }, "name": { "type" : "string" } } }' \
|
||||
'extract-definition={ "type": "array", "items": { "type": "object", "properties": { "term": { "type": "string" }, "definition": { "type": "string" } }, "required": [ "term", "definition" ] } }'
|
||||
|
||||
44
tests/test-agent
Executable file
44
tests/test-agent
Executable file
|
|
@ -0,0 +1,44 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
import json
|
||||
import textwrap
|
||||
from trustgraph.clients.agent_client import AgentClient
|
||||
|
||||
def wrap(text, width=75):
|
||||
|
||||
if text is None: text = "n/a"
|
||||
|
||||
out = textwrap.wrap(
|
||||
text, width=width
|
||||
)
|
||||
return "\n".join(out)
|
||||
|
||||
def output(text, prefix="> ", width=78):
|
||||
|
||||
out = textwrap.indent(
|
||||
text, prefix=prefix
|
||||
)
|
||||
print(out)
|
||||
|
||||
p = AgentClient(pulsar_host="pulsar://localhost:6650")
|
||||
|
||||
q = "How many cats does Mark have? Calculate that number raised to 0.4 power. Is that number lower than the numeric part of the mission identifier of the Space Shuttle Challenger on its last mission? If so, give me an apple pie recipe, otherwise return a poem about cheese."
|
||||
|
||||
output(wrap(q), "\U00002753 ")
|
||||
print()
|
||||
|
||||
def think(x):
|
||||
output(wrap(x), "\U0001f914 ")
|
||||
print()
|
||||
|
||||
def observe(x):
|
||||
output(wrap(x), "\U0001f4a1 ")
|
||||
print()
|
||||
|
||||
resp = p.request(
|
||||
question=q, think=think, observe=observe,
|
||||
)
|
||||
|
||||
output(resp, "\U0001f4ac ")
|
||||
print()
|
||||
|
||||
|
|
@ -7,7 +7,13 @@ p = PromptClient(pulsar_host="pulsar://localhost:6650")
|
|||
|
||||
chunk = """I noticed a cat in my garden. It is a four-legged animal
|
||||
which is a mammal and can be tame or wild. I wonder if it will be friends
|
||||
with me. I think the cat's name is Fred and it has 4 legs"""
|
||||
with me. I think the cat's name is Fred and it has 4 legs.
|
||||
|
||||
A cat is a small mammal.
|
||||
|
||||
A grapefruit is a citrus fruit.
|
||||
|
||||
"""
|
||||
|
||||
resp = p.request_definitions(
|
||||
chunk=chunk,
|
||||
|
|
|
|||
19
tests/test-lang-topics
Executable file
19
tests/test-lang-topics
Executable file
|
|
@ -0,0 +1,19 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
import pulsar
|
||||
from trustgraph.clients.prompt_client import PromptClient
|
||||
|
||||
p = PromptClient(pulsar_host="pulsar://localhost:6650")
|
||||
|
||||
chunk = """I noticed a cat in my garden. It is a four-legged animal
|
||||
which is a mammal and can be tame or wild. I wonder if it will be friends
|
||||
with me. I think the cat's name is Fred and it has 4 legs"""
|
||||
|
||||
resp = p.request_topics(
|
||||
chunk=chunk,
|
||||
)
|
||||
|
||||
for d in resp:
|
||||
print(d.topic)
|
||||
print(" ", d.definition)
|
||||
|
||||
|
|
@ -5,9 +5,10 @@ from trustgraph.clients.llm_client import LlmClient
|
|||
|
||||
llm = LlmClient(pulsar_host="pulsar://localhost:6650")
|
||||
|
||||
system = "You are a lovely assistant."
|
||||
prompt="Write a funny limerick about a llama"
|
||||
|
||||
resp = llm.request(prompt)
|
||||
resp = llm.request(system, prompt)
|
||||
|
||||
print(resp)
|
||||
|
||||
|
|
|
|||
18
tests/test-prompt-analyze
Executable file
18
tests/test-prompt-analyze
Executable file
|
|
@ -0,0 +1,18 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
import json
|
||||
from trustgraph.clients.prompt_client import PromptClient
|
||||
|
||||
p = PromptClient(pulsar_host="pulsar://localhost:6650")
|
||||
|
||||
description = """Fred is a 4-legged cat who is 12 years old"""
|
||||
|
||||
resp = p.request(
|
||||
id="analyze",
|
||||
terms = {
|
||||
"description": description,
|
||||
}
|
||||
)
|
||||
|
||||
print(json.dumps(resp, indent=4))
|
||||
|
||||
46
tests/test-prompt-extraction
Executable file
46
tests/test-prompt-extraction
Executable file
|
|
@ -0,0 +1,46 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
import json
|
||||
from trustgraph.clients.prompt_client import PromptClient
|
||||
|
||||
p = PromptClient(pulsar_host="pulsar://localhost:6650")
|
||||
|
||||
chunk="""
|
||||
The Space Shuttle was a reusable spacecraft that transported astronauts and cargo to and from Earth's orbit. It was designed to launch like a rocket, maneuver in orbit like a spacecraft, and land like an airplane. The Space Shuttle was NASA's space transportation system and was used for many purposes, including:
|
||||
|
||||
Carrying astronauts
|
||||
The Space Shuttle could carry up to seven astronauts at a time.
|
||||
|
||||
Launching, recovering, and repairing satellites
|
||||
The Space Shuttle could launch satellites into orbit, recover them, and repair them.
|
||||
Building the International Space Station
|
||||
The Space Shuttle carried large parts into space to build the International Space Station.
|
||||
Conducting research
|
||||
Astronauts conducted experiments in the Space Shuttle, which was like a science lab in space.
|
||||
|
||||
The Space Shuttle was retired in 2011 after the Columbia accident in 2003. The Columbia Accident Investigation Board report found that the Space Shuttle was unsafe and expensive to make safe.
|
||||
Here are some other facts about the Space Shuttle:
|
||||
|
||||
The Space Shuttle was 184 ft tall and had a diameter of 29 ft.
|
||||
|
||||
The Space Shuttle had a mass of 4,480,000 lb.
|
||||
The Space Shuttle's first flight was on April 12, 1981.
|
||||
The Space Shuttle's last mission was in 2011.
|
||||
"""
|
||||
|
||||
q = "Tell me some facts in the knowledge graph"
|
||||
|
||||
resp = p.request(
|
||||
id="extract-definition",
|
||||
terms = {
|
||||
"text": chunk,
|
||||
}
|
||||
)
|
||||
|
||||
print(resp)
|
||||
|
||||
for fact in resp:
|
||||
print(fact["term"], "::")
|
||||
print(fact["definition"])
|
||||
print()
|
||||
|
||||
18
tests/test-prompt-french-question
Executable file
18
tests/test-prompt-french-question
Executable file
|
|
@ -0,0 +1,18 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
import pulsar
|
||||
from trustgraph.clients.prompt_client import PromptClient
|
||||
|
||||
p = PromptClient(pulsar_host="pulsar://localhost:6650")
|
||||
|
||||
question = """What is the square root of 16?"""
|
||||
|
||||
resp = p.request(
|
||||
id="french-question",
|
||||
terms = {
|
||||
"question": question
|
||||
}
|
||||
)
|
||||
|
||||
print(resp)
|
||||
|
||||
44
tests/test-prompt-knowledge
Executable file
44
tests/test-prompt-knowledge
Executable file
|
|
@ -0,0 +1,44 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
import json
|
||||
from trustgraph.clients.prompt_client import PromptClient
|
||||
|
||||
p = PromptClient(pulsar_host="pulsar://localhost:6650")
|
||||
|
||||
knowledge = [
|
||||
("accident", "evoked", "a wide range of deeply felt public responses"),
|
||||
("Space Shuttle concept", "had", "genesis"),
|
||||
("Commission", "had", "a mandate to develop recommendations for corrective or other action based upon the Commission's findings and determinations"),
|
||||
("Commission", "established", "teams of persons"),
|
||||
("Space Shuttle Challenger", "http://www.w3.org/2004/02/skos/core#definition", "A space shuttle that was destroyed in an accident during mission 51-L."),
|
||||
("The mid fuselage", "contains", "the payload bay"),
|
||||
("Volume I", "contains", "Chapter IX"),
|
||||
("accident", "resulted in", "firm national resolve that those men and women be forever enshrined in the annals of American heroes"),
|
||||
("Volume I", "contains", "Chapter VII"),
|
||||
("Volume I", "contains", "Chapter II"),
|
||||
("Volume I", "contains", "Chapter V"),
|
||||
("Commission", "believes", "its investigation and report have been responsive to the request of the President and hopes that they will serve the best interests of the nation in restoring the United States space program to its preeminent position in the world"),
|
||||
("Commission", "construe", "mandate"),
|
||||
("accident", "became", "a milestone on the way to achieving the full potential that space offers to mankind"),
|
||||
("Volume I", "contains", "The Commission"),
|
||||
("Commission", "http://www.w3.org/2004/02/skos/core#definition", "A group established to investigate the space shuttle accident"),
|
||||
("Volume I", "contains", "Appendix D"),
|
||||
("Commission", "had", "a mandate to review the circumstances surrounding the accident to establish the probable cause or causes of the accident"),
|
||||
("Volume I", "contains", "Recommendations")
|
||||
]
|
||||
|
||||
q = "Tell me some facts in the knowledge graph"
|
||||
|
||||
resp = p.request(
|
||||
id="graph-query",
|
||||
terms = {
|
||||
"name": "Jayney",
|
||||
"knowledge": knowledge,
|
||||
"question": q
|
||||
}
|
||||
)
|
||||
|
||||
print(resp)
|
||||
|
||||
|
||||
|
||||
18
tests/test-prompt-question
Executable file
18
tests/test-prompt-question
Executable file
|
|
@ -0,0 +1,18 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
import pulsar
|
||||
from trustgraph.clients.prompt_client import PromptClient
|
||||
|
||||
p = PromptClient(pulsar_host="pulsar://localhost:6650")
|
||||
|
||||
question = """What is the square root of 16?"""
|
||||
|
||||
resp = p.request(
|
||||
id="question",
|
||||
terms = {
|
||||
"question": question
|
||||
}
|
||||
)
|
||||
|
||||
print(resp)
|
||||
|
||||
19
tests/test-prompt-spanish-question
Executable file
19
tests/test-prompt-spanish-question
Executable file
|
|
@ -0,0 +1,19 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
import pulsar
|
||||
from trustgraph.clients.prompt_client import PromptClient
|
||||
|
||||
p = PromptClient(pulsar_host="pulsar://localhost:6650")
|
||||
|
||||
question = """What is the square root of 16?"""
|
||||
|
||||
resp = p.request(
|
||||
id="question",
|
||||
terms = {
|
||||
"question": question,
|
||||
"attitude": "Spanish-speaking bot"
|
||||
}
|
||||
)
|
||||
|
||||
print(resp)
|
||||
|
||||
|
|
@ -39,8 +39,9 @@ class BaseProcessor:
|
|||
|
||||
def __del__(self):
|
||||
|
||||
if self.client:
|
||||
self.client.close()
|
||||
if hasattr(self, "client"):
|
||||
if self.client:
|
||||
self.client.close()
|
||||
|
||||
@staticmethod
|
||||
def add_args(parser):
|
||||
|
|
|
|||
64
trustgraph-base/trustgraph/clients/agent_client.py
Normal file
64
trustgraph-base/trustgraph/clients/agent_client.py
Normal file
|
|
@ -0,0 +1,64 @@
|
|||
|
||||
import _pulsar
|
||||
|
||||
from .. schema import AgentRequest, AgentResponse
|
||||
from .. schema import agent_request_queue
|
||||
from .. schema import agent_response_queue
|
||||
from . base import BaseClient
|
||||
|
||||
# Ugly
|
||||
ERROR=_pulsar.LoggerLevel.Error
|
||||
WARN=_pulsar.LoggerLevel.Warn
|
||||
INFO=_pulsar.LoggerLevel.Info
|
||||
DEBUG=_pulsar.LoggerLevel.Debug
|
||||
|
||||
class AgentClient(BaseClient):
|
||||
|
||||
def __init__(
|
||||
self, log_level=ERROR,
|
||||
subscriber=None,
|
||||
input_queue=None,
|
||||
output_queue=None,
|
||||
pulsar_host="pulsar://pulsar:6650",
|
||||
):
|
||||
|
||||
if input_queue is None: input_queue = agent_request_queue
|
||||
if output_queue is None: output_queue = agent_response_queue
|
||||
|
||||
super(AgentClient, self).__init__(
|
||||
log_level=log_level,
|
||||
subscriber=subscriber,
|
||||
input_queue=input_queue,
|
||||
output_queue=output_queue,
|
||||
pulsar_host=pulsar_host,
|
||||
input_schema=AgentRequest,
|
||||
output_schema=AgentResponse,
|
||||
)
|
||||
|
||||
def request(
|
||||
self,
|
||||
question,
|
||||
think=None,
|
||||
observe=None,
|
||||
timeout=300
|
||||
):
|
||||
|
||||
def inspect(x):
|
||||
|
||||
if x.thought and think:
|
||||
think(x.thought)
|
||||
return
|
||||
|
||||
if x.observation and observe:
|
||||
observe(x.observation)
|
||||
return
|
||||
|
||||
if x.answer:
|
||||
return True
|
||||
|
||||
return False
|
||||
|
||||
return self.call(
|
||||
question=question, inspect=inspect, timeout=timeout
|
||||
).answer
|
||||
|
||||
|
|
@ -59,10 +59,14 @@ class BaseClient:
|
|||
def call(self, **args):
|
||||
|
||||
timeout = args.get("timeout", DEFAULT_TIMEOUT)
|
||||
inspect = args.get("inspect", lambda x: True)
|
||||
|
||||
if "timeout" in args:
|
||||
del args["timeout"]
|
||||
|
||||
if "inspect" in args:
|
||||
del args["inspect"]
|
||||
|
||||
id = str(uuid.uuid4())
|
||||
|
||||
r = self.input_schema(**args)
|
||||
|
|
@ -103,6 +107,10 @@ class BaseClient:
|
|||
f"{value.error.type}: {value.error.message}"
|
||||
)
|
||||
|
||||
complete = inspect(value)
|
||||
|
||||
if not complete: continue
|
||||
|
||||
resp = msg.value()
|
||||
self.consumer.acknowledge(msg)
|
||||
return resp
|
||||
|
|
|
|||
|
|
@ -38,7 +38,7 @@ class DocumentRagClient(BaseClient):
|
|||
output_schema=DocumentRagResponse,
|
||||
)
|
||||
|
||||
def request(self, query, timeout=500):
|
||||
def request(self, query, timeout=300):
|
||||
|
||||
return self.call(
|
||||
query=query, timeout=timeout
|
||||
|
|
|
|||
|
|
@ -38,8 +38,12 @@ class GraphEmbeddingsClient(BaseClient):
|
|||
output_schema=GraphEmbeddingsResponse,
|
||||
)
|
||||
|
||||
def request(self, vectors, limit=10, timeout=300):
|
||||
def request(
|
||||
self, vectors, user="trustgraph", collection="default",
|
||||
limit=10, timeout=300
|
||||
):
|
||||
return self.call(
|
||||
user=user, collection=collection,
|
||||
vectors=vectors, limit=limit, timeout=timeout
|
||||
).entities
|
||||
|
||||
|
|
|
|||
|
|
@ -38,9 +38,12 @@ class GraphRagClient(BaseClient):
|
|||
output_schema=GraphRagResponse,
|
||||
)
|
||||
|
||||
def request(self, query, timeout=500):
|
||||
def request(
|
||||
self, query, user="trustgraph", collection="default",
|
||||
timeout=500
|
||||
):
|
||||
|
||||
return self.call(
|
||||
query=query, timeout=timeout
|
||||
user=user, collection=collection, query=query, timeout=timeout
|
||||
).response
|
||||
|
||||
|
|
|
|||
|
|
@ -35,6 +35,8 @@ class LlmClient(BaseClient):
|
|||
output_schema=TextCompletionResponse,
|
||||
)
|
||||
|
||||
def request(self, prompt, timeout=300):
|
||||
return self.call(prompt=prompt, timeout=timeout).response
|
||||
def request(self, system, prompt, timeout=300):
|
||||
return self.call(
|
||||
system=system, prompt=prompt, timeout=timeout
|
||||
).response
|
||||
|
||||
|
|
|
|||
|
|
@ -1,7 +1,9 @@
|
|||
|
||||
import _pulsar
|
||||
import json
|
||||
import dataclasses
|
||||
|
||||
from .. schema import PromptRequest, PromptResponse, Fact, RowSchema, Field
|
||||
from .. schema import PromptRequest, PromptResponse
|
||||
from .. schema import prompt_request_queue
|
||||
from .. schema import prompt_response_queue
|
||||
from . base import BaseClient
|
||||
|
|
@ -12,6 +14,23 @@ WARN=_pulsar.LoggerLevel.Warn
|
|||
INFO=_pulsar.LoggerLevel.Info
|
||||
DEBUG=_pulsar.LoggerLevel.Debug
|
||||
|
||||
@dataclasses.dataclass
|
||||
class Definition:
|
||||
name: str
|
||||
definition: str
|
||||
|
||||
@dataclasses.dataclass
|
||||
class Relationship:
|
||||
s: str
|
||||
p: str
|
||||
o: str
|
||||
o_entity: str
|
||||
|
||||
@dataclasses.dataclass
|
||||
class Topic:
|
||||
name: str
|
||||
definition: str
|
||||
|
||||
class PromptClient(BaseClient):
|
||||
|
||||
def __init__(
|
||||
|
|
@ -38,63 +57,116 @@ class PromptClient(BaseClient):
|
|||
output_schema=PromptResponse,
|
||||
)
|
||||
|
||||
def request(self, id, variables, timeout=300):
|
||||
|
||||
resp = self.call(
|
||||
id=id,
|
||||
terms={
|
||||
k: json.dumps(v)
|
||||
for k, v in variables.items()
|
||||
},
|
||||
timeout=timeout
|
||||
)
|
||||
|
||||
if resp.text: return resp.text
|
||||
|
||||
return json.loads(resp.object)
|
||||
|
||||
def request_definitions(self, chunk, timeout=300):
|
||||
|
||||
return self.call(
|
||||
kind="extract-definitions", chunk=chunk,
|
||||
defs = self.request(
|
||||
id="extract-definitions",
|
||||
variables={
|
||||
"text": chunk
|
||||
},
|
||||
timeout=timeout
|
||||
).definitions
|
||||
|
||||
def request_topics(self, chunk, timeout=300):
|
||||
)
|
||||
|
||||
return self.call(
|
||||
kind="extract-topics", chunk=chunk,
|
||||
timeout=timeout
|
||||
).topics
|
||||
return [
|
||||
Definition(name=d["entity"], definition=d["definition"])
|
||||
for d in defs
|
||||
]
|
||||
|
||||
def request_relationships(self, chunk, timeout=300):
|
||||
|
||||
return self.call(
|
||||
kind="extract-relationships", chunk=chunk,
|
||||
rels = self.request(
|
||||
id="extract-relationships",
|
||||
variables={
|
||||
"text": chunk
|
||||
},
|
||||
timeout=timeout
|
||||
).relationships
|
||||
)
|
||||
|
||||
return [
|
||||
Relationship(
|
||||
s=d["subject"],
|
||||
p=d["predicate"],
|
||||
o=d["object"],
|
||||
o_entity=d["object-entity"]
|
||||
)
|
||||
for d in rels
|
||||
]
|
||||
|
||||
def request_topics(self, chunk, timeout=300):
|
||||
|
||||
topics = self.request(
|
||||
id="extract-topics",
|
||||
variables={
|
||||
"text": chunk
|
||||
},
|
||||
timeout=timeout
|
||||
)
|
||||
|
||||
return [
|
||||
Topic(name=d["topic"], definition=d["definition"])
|
||||
for d in topics
|
||||
]
|
||||
|
||||
def request_rows(self, schema, chunk, timeout=300):
|
||||
|
||||
return self.call(
|
||||
kind="extract-rows", chunk=chunk,
|
||||
row_schema=RowSchema(
|
||||
name=schema.name,
|
||||
description=schema.description,
|
||||
fields=[
|
||||
Field(
|
||||
name=f.name, type=str(f.type), size=f.size,
|
||||
primary=f.primary, description=f.description,
|
||||
)
|
||||
for f in schema.fields
|
||||
]
|
||||
),
|
||||
return self.request(
|
||||
id="extract-rows",
|
||||
variables={
|
||||
"chunk": chunk,
|
||||
"row-schema": {
|
||||
"name": schema.name,
|
||||
"description": schema.description,
|
||||
"fields": [
|
||||
{
|
||||
"name": f.name, "type": str(f.type),
|
||||
"size": f.size, "primary": f.primary,
|
||||
"description": f.description,
|
||||
}
|
||||
for f in schema.fields
|
||||
]
|
||||
}
|
||||
},
|
||||
timeout=timeout
|
||||
).rows
|
||||
)
|
||||
|
||||
def request_kg_prompt(self, query, kg, timeout=300):
|
||||
|
||||
return self.call(
|
||||
kind="kg-prompt",
|
||||
query=query,
|
||||
kg=[
|
||||
Fact(s=v[0], p=v[1], o=v[2])
|
||||
for v in kg
|
||||
],
|
||||
return self.request(
|
||||
id="kg-prompt",
|
||||
variables={
|
||||
"query": query,
|
||||
"knowledge": [
|
||||
{ "s": v[0], "p": v[1], "o": v[2] }
|
||||
for v in kg
|
||||
]
|
||||
},
|
||||
timeout=timeout
|
||||
).answer
|
||||
)
|
||||
|
||||
def request_document_prompt(self, query, documents, timeout=300):
|
||||
|
||||
return self.call(
|
||||
kind="document-prompt",
|
||||
query=query,
|
||||
documents=documents,
|
||||
return self.request(
|
||||
id="document-prompt",
|
||||
variables={
|
||||
"query": query,
|
||||
"documents": documents,
|
||||
},
|
||||
timeout=timeout
|
||||
).answer
|
||||
)
|
||||
|
||||
|
||||
|
|
|
|||
|
|
@ -48,11 +48,18 @@ class TriplesQueryClient(BaseClient):
|
|||
|
||||
return Value(value=ent, is_uri=False)
|
||||
|
||||
def request(self, s, p, o, limit=10, timeout=60):
|
||||
def request(
|
||||
self,
|
||||
s, p, o,
|
||||
user="trustgraph", collection="default",
|
||||
limit=10, timeout=120,
|
||||
):
|
||||
return self.call(
|
||||
s=self.create_value(s),
|
||||
p=self.create_value(p),
|
||||
o=self.create_value(o),
|
||||
user=user,
|
||||
collection=collection,
|
||||
limit=limit,
|
||||
timeout=timeout,
|
||||
).triples
|
||||
|
|
|
|||
6
trustgraph-base/trustgraph/knowledge/__init__.py
Normal file
6
trustgraph-base/trustgraph/knowledge/__init__.py
Normal file
|
|
@ -0,0 +1,6 @@
|
|||
|
||||
from . identifier import *
|
||||
from . publication import *
|
||||
from . document import *
|
||||
from . organization import *
|
||||
|
||||
25
trustgraph-base/trustgraph/knowledge/defs.py
Normal file
25
trustgraph-base/trustgraph/knowledge/defs.py
Normal file
|
|
@ -0,0 +1,25 @@
|
|||
|
||||
IS_A = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#type'
|
||||
LABEL = 'http://www.w3.org/2000/01/rdf-schema#label'
|
||||
|
||||
DIGITAL_DOCUMENT = 'https://schema.org/DigitalDocument'
|
||||
PUBLICATION_EVENT = 'https://schema.org/PublicationEvent'
|
||||
ORGANIZATION = 'https://schema.org/Organization'
|
||||
|
||||
NAME = 'https://schema.org/name'
|
||||
DESCRIPTION = 'https://schema.org/description'
|
||||
COPYRIGHT_NOTICE = 'https://schema.org/copyrightNotice'
|
||||
COPYRIGHT_HOLDER = 'https://schema.org/copyrightHolder'
|
||||
COPYRIGHT_YEAR = 'https://schema.org/copyrightYear'
|
||||
LICENSE = 'https://schema.org/license'
|
||||
PUBLICATION = 'https://schema.org/publication'
|
||||
START_DATE = 'https://schema.org/startDate'
|
||||
END_DATE = 'https://schema.org/endDate'
|
||||
PUBLISHED_BY = 'https://schema.org/publishedBy'
|
||||
DATE_PUBLISHED = 'https://schema.org/datePublished'
|
||||
PUBLICATION = 'https://schema.org/publication'
|
||||
DATE_PUBLISHED = 'https://schema.org/datePublished'
|
||||
URL = 'https://schema.org/url'
|
||||
IDENTIFIER = 'https://schema.org/identifier'
|
||||
KEYWORD = 'https://schema.org/keywords'
|
||||
|
||||
120
trustgraph-base/trustgraph/knowledge/document.py
Normal file
120
trustgraph-base/trustgraph/knowledge/document.py
Normal file
|
|
@ -0,0 +1,120 @@
|
|||
|
||||
from . defs import *
|
||||
from .. schema import Triple, Value
|
||||
|
||||
class DigitalDocument:
|
||||
|
||||
def __init__(
|
||||
self, id, name=None, description=None, copyright_notice=None,
|
||||
copyright_holder=None, copyright_year=None, license=None,
|
||||
identifier=None,
|
||||
publication=None, url=None, keywords=[]
|
||||
):
|
||||
|
||||
self.id = id
|
||||
self.name = name
|
||||
self.description = description
|
||||
self.copyright_notice = copyright_notice
|
||||
self.copyright_holder = copyright_holder
|
||||
self.copyright_year = copyright_year
|
||||
self.license = license
|
||||
self.publication = publication
|
||||
self.url = url
|
||||
self.identifier = identifier
|
||||
self.keywords = keywords
|
||||
|
||||
def emit(self, emit):
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=IS_A, is_uri=True),
|
||||
o=Value(value=DIGITAL_DOCUMENT, is_uri=True)
|
||||
))
|
||||
|
||||
if self.name:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=LABEL, is_uri=True),
|
||||
o=Value(value=self.name, is_uri=False)
|
||||
))
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=NAME, is_uri=True),
|
||||
o=Value(value=self.name, is_uri=False)
|
||||
))
|
||||
|
||||
if self.identifier:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=id, is_uri=True),
|
||||
p=Value(value=IDENTIFIER, is_uri=True),
|
||||
o=Value(value=self.identifier, is_uri=False)
|
||||
))
|
||||
|
||||
if self.description:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=DESCRIPTION, is_uri=True),
|
||||
o=Value(value=self.description, is_uri=False)
|
||||
))
|
||||
|
||||
if self.copyright_notice:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=COPYRIGHT_NOTICE, is_uri=True),
|
||||
o=Value(value=self.copyright_notice, is_uri=False)
|
||||
))
|
||||
|
||||
if self.copyright_holder:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=COPYRIGHT_HOLDER, is_uri=True),
|
||||
o=Value(value=self.copyright_holder, is_uri=False)
|
||||
))
|
||||
|
||||
if self.copyright_year:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=COPYRIGHT_YEAR, is_uri=True),
|
||||
o=Value(value=self.copyright_year, is_uri=False)
|
||||
))
|
||||
|
||||
if self.license:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=LICENSE, is_uri=True),
|
||||
o=Value(value=self.license, is_uri=False)
|
||||
))
|
||||
|
||||
if self.keywords:
|
||||
for k in self.keywords:
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=KEYWORD, is_uri=True),
|
||||
o=Value(value=k, is_uri=False)
|
||||
))
|
||||
|
||||
if self.publication:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=PUBLICATION, is_uri=True),
|
||||
o=Value(value=self.publication.id, is_uri=True)
|
||||
))
|
||||
|
||||
self.publication.emit(emit)
|
||||
|
||||
if self.url:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=URL, is_uri=True),
|
||||
o=Value(value=self.url, is_uri=True)
|
||||
))
|
||||
23
trustgraph-base/trustgraph/knowledge/identifier.py
Normal file
23
trustgraph-base/trustgraph/knowledge/identifier.py
Normal file
|
|
@ -0,0 +1,23 @@
|
|||
|
||||
import uuid
|
||||
import hashlib
|
||||
|
||||
def hash(data):
|
||||
|
||||
if isinstance(data, str):
|
||||
data = data.encode("utf-8")
|
||||
|
||||
# Create a SHA256 hash from the data
|
||||
id = hashlib.sha256(data).hexdigest()
|
||||
|
||||
# Convert into a UUID, 64-byte hash becomes 32-byte UUID
|
||||
id = str(uuid.UUID(id[::2]))
|
||||
|
||||
return id
|
||||
|
||||
def to_uri(pref, id):
|
||||
return f"https://trustgraph.ai/{pref}/{id}"
|
||||
|
||||
PREF_PUBEV = "pubev"
|
||||
PREF_ORG = "org"
|
||||
PREF_DOC = "doc"
|
||||
40
trustgraph-base/trustgraph/knowledge/organization.py
Normal file
40
trustgraph-base/trustgraph/knowledge/organization.py
Normal file
|
|
@ -0,0 +1,40 @@
|
|||
|
||||
from . defs import *
|
||||
from .. schema import Triple, Value
|
||||
|
||||
class Organization:
|
||||
def __init__(self, id, name=None, description=None):
|
||||
self.id = id
|
||||
self.name = name
|
||||
self.description = description
|
||||
|
||||
def emit(self, emit):
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=IS_A, is_uri=True),
|
||||
o=Value(value=ORGANIZATION, is_uri=True)
|
||||
))
|
||||
|
||||
if self.name:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=LABEL, is_uri=True),
|
||||
o=Value(value=self.name, is_uri=False)
|
||||
))
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=NAME, is_uri=True),
|
||||
o=Value(value=self.name, is_uri=False)
|
||||
))
|
||||
|
||||
if self.description:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=DESCRIPTION, is_uri=True),
|
||||
o=Value(value=self.description, is_uri=False)
|
||||
))
|
||||
|
||||
69
trustgraph-base/trustgraph/knowledge/publication.py
Normal file
69
trustgraph-base/trustgraph/knowledge/publication.py
Normal file
|
|
@ -0,0 +1,69 @@
|
|||
|
||||
from . defs import *
|
||||
from .. schema import Triple, Value
|
||||
|
||||
class PublicationEvent:
|
||||
def __init__(
|
||||
self, id, organization=None, name=None, description=None,
|
||||
start_date=None, end_date=None,
|
||||
):
|
||||
self.id = id
|
||||
self.organization = organization
|
||||
self.name = name
|
||||
self.description = description
|
||||
self.start_date = start_date
|
||||
self.end_date = end_date
|
||||
|
||||
def emit(self, emit):
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=IS_A, is_uri=True),
|
||||
o=Value(value=PUBLICATION_EVENT, is_uri=True)))
|
||||
|
||||
if self.name:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=LABEL, is_uri=True),
|
||||
o=Value(value=self.name, is_uri=False)
|
||||
))
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=NAME, is_uri=True),
|
||||
o=Value(value=self.name, is_uri=False)
|
||||
))
|
||||
|
||||
if self.description:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=DESCRIPTION, is_uri=True),
|
||||
o=Value(value=self.description, is_uri=False)
|
||||
))
|
||||
|
||||
if self.organization:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=PUBLISHED_BY, is_uri=True),
|
||||
o=Value(value=self.organization.id, is_uri=True)
|
||||
))
|
||||
|
||||
self.organization.emit(emit)
|
||||
|
||||
if self.start_date:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=START_DATE, is_uri=True),
|
||||
o=Value(value=self.start_date, is_uri=False)
|
||||
))
|
||||
|
||||
if self.end_date:
|
||||
|
||||
emit(Triple(
|
||||
s=Value(value=self.id, is_uri=True),
|
||||
p=Value(value=END_DATE, is_uri=True),
|
||||
o=Value(value=self.end_date, is_uri=False)))
|
||||
|
|
@ -1,6 +1,7 @@
|
|||
|
||||
RDF_LABEL = "http://www.w3.org/2000/01/rdf-schema#label"
|
||||
DEFINITION = "http://www.w3.org/2004/02/skos/core#definition"
|
||||
SUBJECT_OF = "https://schema.org/subjectOf"
|
||||
|
||||
TRUSTGRAPH_ENTITIES = "http://trustgraph.ai/e/"
|
||||
|
||||
|
|
|
|||
|
|
@ -7,6 +7,6 @@ from . object import *
|
|||
from . topic import *
|
||||
from . graph import *
|
||||
from . retrieval import *
|
||||
|
||||
|
||||
from . metadata import *
|
||||
from . agent import *
|
||||
|
||||
|
|
|
|||
37
trustgraph-base/trustgraph/schema/agent.py
Normal file
37
trustgraph-base/trustgraph/schema/agent.py
Normal file
|
|
@ -0,0 +1,37 @@
|
|||
|
||||
from pulsar.schema import Record, String, Array, Map
|
||||
|
||||
from . topic import topic
|
||||
from . types import Error
|
||||
|
||||
############################################################################
|
||||
|
||||
# Prompt services, abstract the prompt generation
|
||||
|
||||
class AgentStep(Record):
|
||||
thought = String()
|
||||
action = String()
|
||||
arguments = Map(String())
|
||||
observation = String()
|
||||
|
||||
class AgentRequest(Record):
|
||||
question = String()
|
||||
plan = String()
|
||||
state = String()
|
||||
history = Array(AgentStep())
|
||||
|
||||
class AgentResponse(Record):
|
||||
answer = String()
|
||||
error = Error()
|
||||
thought = String()
|
||||
observation = String()
|
||||
|
||||
agent_request_queue = topic(
|
||||
'agent', kind='non-persistent', namespace='request'
|
||||
)
|
||||
agent_response_queue = topic(
|
||||
'agent', kind='non-persistent', namespace='response'
|
||||
)
|
||||
|
||||
############################################################################
|
||||
|
||||
|
|
@ -2,17 +2,13 @@
|
|||
from pulsar.schema import Record, Bytes, String, Boolean, Integer, Array, Double
|
||||
from . topic import topic
|
||||
from . types import Error
|
||||
|
||||
class Source(Record):
|
||||
source = String()
|
||||
id = String()
|
||||
title = String()
|
||||
from . metadata import Metadata
|
||||
|
||||
############################################################################
|
||||
|
||||
# PDF docs etc.
|
||||
class Document(Record):
|
||||
source = Source()
|
||||
metadata = Metadata()
|
||||
data = Bytes()
|
||||
|
||||
document_ingest_queue = topic('document-load')
|
||||
|
|
@ -22,7 +18,7 @@ document_ingest_queue = topic('document-load')
|
|||
# Text documents / text from PDF
|
||||
|
||||
class TextDocument(Record):
|
||||
source = Source()
|
||||
metadata = Metadata()
|
||||
text = Bytes()
|
||||
|
||||
text_ingest_queue = topic('text-document-load')
|
||||
|
|
@ -32,7 +28,7 @@ text_ingest_queue = topic('text-document-load')
|
|||
# Chunks of text
|
||||
|
||||
class Chunk(Record):
|
||||
source = Source()
|
||||
metadata = Metadata()
|
||||
chunk = Bytes()
|
||||
|
||||
chunk_ingest_queue = topic('chunk-load')
|
||||
|
|
@ -42,7 +38,7 @@ chunk_ingest_queue = topic('chunk-load')
|
|||
# Chunk embeddings are an embeddings associated with a text chunk
|
||||
|
||||
class ChunkEmbeddings(Record):
|
||||
source = Source()
|
||||
metadata = Metadata()
|
||||
vectors = Array(Array(Double()))
|
||||
chunk = Bytes()
|
||||
|
||||
|
|
|
|||
|
|
@ -1,16 +1,16 @@
|
|||
|
||||
from pulsar.schema import Record, Bytes, String, Boolean, Integer, Array, Double
|
||||
|
||||
from . documents import Source
|
||||
from . types import Error, Value
|
||||
from . types import Error, Value, Triple
|
||||
from . topic import topic
|
||||
from . metadata import Metadata
|
||||
|
||||
############################################################################
|
||||
|
||||
# Graph embeddings are embeddings associated with a graph entity
|
||||
|
||||
class GraphEmbeddings(Record):
|
||||
source = Source()
|
||||
metadata = Metadata()
|
||||
vectors = Array(Array(Double()))
|
||||
entity = Value()
|
||||
|
||||
|
|
@ -23,6 +23,8 @@ graph_embeddings_store_queue = topic('graph-embeddings-store')
|
|||
class GraphEmbeddingsRequest(Record):
|
||||
vectors = Array(Array(Double()))
|
||||
limit = Integer()
|
||||
user = String()
|
||||
collection = String()
|
||||
|
||||
class GraphEmbeddingsResponse(Record):
|
||||
error = Error()
|
||||
|
|
@ -39,11 +41,9 @@ graph_embeddings_response_queue = topic(
|
|||
|
||||
# Graph triples
|
||||
|
||||
class Triple(Record):
|
||||
source = Source()
|
||||
s = Value()
|
||||
p = Value()
|
||||
o = Value()
|
||||
class Triples(Record):
|
||||
metadata = Metadata()
|
||||
triples = Array(Triple())
|
||||
|
||||
triples_store_queue = topic('triples-store')
|
||||
|
||||
|
|
@ -56,6 +56,8 @@ class TriplesQueryRequest(Record):
|
|||
p = Value()
|
||||
o = Value()
|
||||
limit = Integer()
|
||||
user = String()
|
||||
collection = String()
|
||||
|
||||
class TriplesQueryResponse(Record):
|
||||
error = Error()
|
||||
|
|
|
|||
16
trustgraph-base/trustgraph/schema/metadata.py
Normal file
16
trustgraph-base/trustgraph/schema/metadata.py
Normal file
|
|
@ -0,0 +1,16 @@
|
|||
|
||||
from pulsar.schema import Record, String, Array
|
||||
from . types import Triple
|
||||
|
||||
class Metadata(Record):
|
||||
|
||||
# Source identifier
|
||||
id = String()
|
||||
|
||||
# Subgraph
|
||||
metadata = Array(Triple())
|
||||
|
||||
# Collection management
|
||||
user = String()
|
||||
collection = String()
|
||||
|
||||
|
|
@ -9,6 +9,7 @@ from . types import Error
|
|||
# LLM text completion
|
||||
|
||||
class TextCompletionRequest(Record):
|
||||
system = String()
|
||||
prompt = String()
|
||||
|
||||
class TextCompletionResponse(Record):
|
||||
|
|
|
|||
|
|
@ -2,7 +2,7 @@
|
|||
from pulsar.schema import Record, Bytes, String, Boolean, Integer, Array
|
||||
from pulsar.schema import Double, Map
|
||||
|
||||
from . documents import Source
|
||||
from . metadata import Metadata
|
||||
from . types import Value, RowSchema
|
||||
from . topic import topic
|
||||
|
||||
|
|
@ -12,7 +12,7 @@ from . topic import topic
|
|||
# object
|
||||
|
||||
class ObjectEmbeddings(Record):
|
||||
source = Source()
|
||||
metadata = Metadata()
|
||||
vectors = Array(Array(Double()))
|
||||
name = String()
|
||||
key_name = String()
|
||||
|
|
@ -25,7 +25,7 @@ object_embeddings_store_queue = topic('object-embeddings-store')
|
|||
# Stores rows of information
|
||||
|
||||
class Rows(Record):
|
||||
source = Source()
|
||||
metadata = Metadata()
|
||||
row_schema = RowSchema()
|
||||
rows = Array(Map(String()))
|
||||
|
||||
|
|
|
|||
|
|
@ -39,20 +39,21 @@ class Fact(Record):
|
|||
# schema, chunk -> rows
|
||||
|
||||
class PromptRequest(Record):
|
||||
kind = String()
|
||||
chunk = String()
|
||||
query = String()
|
||||
kg = Array(Fact())
|
||||
documents = Array(Bytes())
|
||||
row_schema = RowSchema()
|
||||
id = String()
|
||||
|
||||
# JSON encoded values
|
||||
terms = Map(String())
|
||||
|
||||
class PromptResponse(Record):
|
||||
|
||||
# Error case
|
||||
error = Error()
|
||||
answer = String()
|
||||
definitions = Array(Definition())
|
||||
topics = Array(Topic())
|
||||
relationships = Array(Relationship())
|
||||
rows = Array(Map(String()))
|
||||
|
||||
# Just plain text
|
||||
text = String()
|
||||
|
||||
# JSON encoded
|
||||
object = String()
|
||||
|
||||
prompt_request_queue = topic(
|
||||
'prompt', kind='non-persistent', namespace='request'
|
||||
|
|
|
|||
|
|
@ -9,6 +9,8 @@ from . types import Error, Value
|
|||
|
||||
class GraphRagQuery(Record):
|
||||
query = String()
|
||||
user = String()
|
||||
collection = String()
|
||||
|
||||
class GraphRagResponse(Record):
|
||||
error = Error()
|
||||
|
|
@ -27,6 +29,8 @@ graph_rag_response_queue = topic(
|
|||
|
||||
class DocumentRagQuery(Record):
|
||||
query = String()
|
||||
user = String()
|
||||
collection = String()
|
||||
|
||||
class DocumentRagResponse(Record):
|
||||
error = Error()
|
||||
|
|
|
|||
|
|
@ -10,6 +10,11 @@ class Value(Record):
|
|||
is_uri = Boolean()
|
||||
type = String()
|
||||
|
||||
class Triple(Record):
|
||||
s = Value()
|
||||
p = Value()
|
||||
o = Value()
|
||||
|
||||
class Field(Record):
|
||||
name = String()
|
||||
# int, string, long, bool, float, double
|
||||
|
|
|
|||
|
|
@ -34,7 +34,7 @@ setuptools.setup(
|
|||
python_requires='>=3.8',
|
||||
download_url = "https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v" + version + ".tar.gz",
|
||||
install_requires=[
|
||||
"trustgraph-base<0.12",
|
||||
"trustgraph-base>=0.15,<0.16",
|
||||
"pulsar-client",
|
||||
"prometheus-client",
|
||||
"boto3",
|
||||
|
|
|
|||
|
|
@ -7,6 +7,7 @@ Input is prompt, output is response. Mistral is default.
|
|||
import boto3
|
||||
import json
|
||||
from prometheus_client import Histogram
|
||||
import os
|
||||
|
||||
from .... schema import TextCompletionRequest, TextCompletionResponse, Error
|
||||
from .... schema import text_completion_request_queue
|
||||
|
|
@ -21,10 +22,11 @@ default_input_queue = text_completion_request_queue
|
|||
default_output_queue = text_completion_response_queue
|
||||
default_subscriber = module
|
||||
default_model = 'mistral.mistral-large-2407-v1:0'
|
||||
default_region = 'us-west-2'
|
||||
default_temperature = 0.0
|
||||
default_max_output = 2048
|
||||
|
||||
default_aws_id_key = os.getenv("AWS_ID_KEY", None)
|
||||
default_aws_secret = os.getenv("AWS_SECRET", None)
|
||||
default_aws_region = os.getenv("AWS_REGION", 'us-west-2')
|
||||
|
||||
class Processor(ConsumerProducer):
|
||||
|
||||
|
|
@ -34,12 +36,21 @@ class Processor(ConsumerProducer):
|
|||
output_queue = params.get("output_queue", default_output_queue)
|
||||
subscriber = params.get("subscriber", default_subscriber)
|
||||
model = params.get("model", default_model)
|
||||
aws_id = params.get("aws_id_key")
|
||||
aws_secret = params.get("aws_secret")
|
||||
aws_region = params.get("aws_region", default_region)
|
||||
aws_id_key = params.get("aws_id_key", default_aws_id_key)
|
||||
aws_secret = params.get("aws_secret", default_aws_secret)
|
||||
aws_region = params.get("aws_region", default_aws_region)
|
||||
temperature = params.get("temperature", default_temperature)
|
||||
max_output = params.get("max_output", default_max_output)
|
||||
|
||||
if aws_id_key is None:
|
||||
raise RuntimeError("AWS ID not specified")
|
||||
|
||||
if aws_secret is None:
|
||||
raise RuntimeError("AWS secret not specified")
|
||||
|
||||
if aws_region is None:
|
||||
raise RuntimeError("AWS region not specified")
|
||||
|
||||
super(Processor, self).__init__(
|
||||
**params | {
|
||||
"input_queue": input_queue,
|
||||
|
|
@ -71,7 +82,7 @@ class Processor(ConsumerProducer):
|
|||
self.max_output = max_output
|
||||
|
||||
self.session = boto3.Session(
|
||||
aws_access_key_id=aws_id,
|
||||
aws_access_key_id=aws_id_key,
|
||||
aws_secret_access_key=aws_secret,
|
||||
region_name=aws_region
|
||||
)
|
||||
|
|
@ -90,7 +101,7 @@ class Processor(ConsumerProducer):
|
|||
|
||||
print(f"Handling prompt {id}...", flush=True)
|
||||
|
||||
prompt = v.prompt
|
||||
prompt = v.system + "\n\n" + v.prompt
|
||||
|
||||
try:
|
||||
|
||||
|
|
@ -289,17 +300,20 @@ class Processor(ConsumerProducer):
|
|||
|
||||
parser.add_argument(
|
||||
'-z', '--aws-id-key',
|
||||
default=default_aws_id_key,
|
||||
help=f'AWS ID Key'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-k', '--aws-secret',
|
||||
default=default_aws_secret,
|
||||
help=f'AWS Secret Key'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-r', '--aws-region',
|
||||
help=f'AWS Region (default: us-west-2)'
|
||||
default=default_aws_region,
|
||||
help=f'AWS Region'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
|
|
@ -320,4 +334,3 @@ def run():
|
|||
|
||||
Processor.start(module, __doc__)
|
||||
|
||||
|
||||
|
|
|
|||
|
|
@ -9,12 +9,17 @@ import os
|
|||
from trustgraph.clients.triples_query_client import TriplesQueryClient
|
||||
|
||||
default_pulsar_host = os.getenv("PULSAR_HOST", 'pulsar://localhost:6650')
|
||||
default_user = 'trustgraph'
|
||||
default_collection = 'default'
|
||||
|
||||
def show_graph(pulsar):
|
||||
def show_graph(pulsar, user, collection):
|
||||
|
||||
tq = TriplesQueryClient(pulsar_host=pulsar)
|
||||
|
||||
rows = tq.request(None, None, None, limit=10_000_000)
|
||||
rows = tq.request(
|
||||
user=user, collection=collection,
|
||||
s=None, p=None, o=None, limit=10_000_000
|
||||
)
|
||||
|
||||
for row in rows:
|
||||
print(row.s.value, row.p.value, row.o.value)
|
||||
|
|
@ -22,7 +27,7 @@ def show_graph(pulsar):
|
|||
def main():
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
prog='graph-show',
|
||||
prog='tg-graph-show',
|
||||
description=__doc__,
|
||||
)
|
||||
|
||||
|
|
@ -32,11 +37,26 @@ def main():
|
|||
help=f'Pulsar host (default: {default_pulsar_host})',
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-u', '--user',
|
||||
default=default_user,
|
||||
help=f'User ID (default: {default_user})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-c', '--collection',
|
||||
default=default_collection,
|
||||
help=f'Collection ID (default: {default_collection})'
|
||||
)
|
||||
|
||||
args = parser.parse_args()
|
||||
|
||||
try:
|
||||
|
||||
show_graph(args.pulsar_host)
|
||||
show_graph(
|
||||
pulsar=args.pulsar_host, user=args.user,
|
||||
collection=args.collection,
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
|
||||
|
|
|
|||
|
|
@ -1,7 +1,8 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
"""
|
||||
Connects to the graph query service and dumps all graph edges.
|
||||
Connects to the graph query service and dumps all graph edges in Turtle
|
||||
format.
|
||||
"""
|
||||
|
||||
import argparse
|
||||
|
|
@ -50,7 +51,7 @@ def show_graph(pulsar):
|
|||
def main():
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
prog='graph-show',
|
||||
prog='tg-graph-to-turtle',
|
||||
description=__doc__,
|
||||
)
|
||||
|
||||
|
|
|
|||
|
|
@ -1,7 +1,7 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
"""
|
||||
Initialises Pulsar with Trustgraph tenant / namespaces & policy
|
||||
Initialises Pulsar with Trustgraph tenant / namespaces & policy.
|
||||
"""
|
||||
|
||||
import requests
|
||||
|
|
|
|||
123
trustgraph-cli/scripts/tg-invoke-agent
Executable file
123
trustgraph-cli/scripts/tg-invoke-agent
Executable file
|
|
@ -0,0 +1,123 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
"""
|
||||
Uses the GraphRAG service to answer a query
|
||||
"""
|
||||
|
||||
import argparse
|
||||
import os
|
||||
import textwrap
|
||||
|
||||
from trustgraph.clients.agent_client import AgentClient
|
||||
|
||||
default_pulsar_host = os.getenv("PULSAR_HOST", 'pulsar://localhost:6650')
|
||||
default_user = 'trustgraph'
|
||||
default_collection = 'default'
|
||||
|
||||
def wrap(text, width=75):
|
||||
if text is None: text = "n/a"
|
||||
out = textwrap.wrap(
|
||||
text, width=width
|
||||
)
|
||||
return "\n".join(out)
|
||||
|
||||
def output(text, prefix="> ", width=78):
|
||||
out = textwrap.indent(
|
||||
text, prefix=prefix
|
||||
)
|
||||
print(out)
|
||||
|
||||
def query(
|
||||
pulsar_host, query, user, collection,
|
||||
plan=None, state=None, verbose=False
|
||||
):
|
||||
|
||||
am = AgentClient(pulsar_host=pulsar_host)
|
||||
|
||||
if verbose:
|
||||
output(wrap(query), "\U00002753 ")
|
||||
print()
|
||||
|
||||
def think(x):
|
||||
if verbose:
|
||||
output(wrap(x), "\U0001f914 ")
|
||||
print()
|
||||
|
||||
def observe(x):
|
||||
if verbose:
|
||||
output(wrap(x), "\U0001f4a1 ")
|
||||
print()
|
||||
|
||||
resp = am.request(
|
||||
question=query, think=think, observe=observe,
|
||||
)
|
||||
|
||||
print(resp)
|
||||
|
||||
def main():
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
prog='tg-invoke-agent',
|
||||
description=__doc__,
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-p', '--pulsar-host',
|
||||
default=default_pulsar_host,
|
||||
help=f'Pulsar host (default: {default_pulsar_host})',
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-q', '--query',
|
||||
required=True,
|
||||
help=f'Query to execute',
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-u', '--user',
|
||||
default=default_user,
|
||||
help=f'User ID (default: {default_user})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-c', '--collection',
|
||||
default=default_collection,
|
||||
help=f'Collection ID (default: {default_collection})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-l', '--plan',
|
||||
help=f'Agent plan (default: unspecified)'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-s', '--state',
|
||||
help=f'Agent initial state (default: unspecified)'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-v', '--verbose',
|
||||
action="store_true",
|
||||
help=f'Output thinking/observations'
|
||||
)
|
||||
|
||||
args = parser.parse_args()
|
||||
|
||||
try:
|
||||
|
||||
query(
|
||||
pulsar_host=args.pulsar_host,
|
||||
query=args.query,
|
||||
user=args.user,
|
||||
collection=args.collection,
|
||||
plan=args.plan,
|
||||
state=args.state,
|
||||
verbose=args.verbose,
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
|
||||
print("Exception:", e, flush=True)
|
||||
|
||||
main()
|
||||
|
||||
63
trustgraph-cli/scripts/tg-invoke-llm
Executable file
63
trustgraph-cli/scripts/tg-invoke-llm
Executable file
|
|
@ -0,0 +1,63 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
"""
|
||||
Invokes the text completion service by specifying an LLM system prompt
|
||||
and user prompt. Both arguments are required.
|
||||
"""
|
||||
|
||||
import argparse
|
||||
import os
|
||||
import json
|
||||
from trustgraph.clients.llm_client import LlmClient
|
||||
|
||||
default_pulsar_host = os.getenv("PULSAR_HOST", 'pulsar://localhost:6650')
|
||||
|
||||
def query(pulsar_host, system, prompt):
|
||||
|
||||
cli = LlmClient(pulsar_host=pulsar_host)
|
||||
|
||||
resp = cli.request(system=system, prompt=prompt)
|
||||
|
||||
print(resp)
|
||||
|
||||
def main():
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
prog='tg-invoke-llm',
|
||||
description=__doc__,
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-p', '--pulsar-host',
|
||||
default=default_pulsar_host,
|
||||
help=f'Pulsar host (default: {default_pulsar_host})',
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'system',
|
||||
nargs=1,
|
||||
help='LLM system prompt e.g. You are a helpful assistant',
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'prompt',
|
||||
nargs=1,
|
||||
help='LLM prompt e.g. What is 2 + 2?',
|
||||
)
|
||||
|
||||
args = parser.parse_args()
|
||||
|
||||
try:
|
||||
|
||||
query(
|
||||
pulsar_host=args.pulsar_host,
|
||||
system=args.system[0],
|
||||
prompt=args.prompt[0],
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
|
||||
print("Exception:", e, flush=True)
|
||||
|
||||
main()
|
||||
|
||||
83
trustgraph-cli/scripts/tg-invoke-prompt
Executable file
83
trustgraph-cli/scripts/tg-invoke-prompt
Executable file
|
|
@ -0,0 +1,83 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
"""
|
||||
Invokes the LLM prompt service by specifying the prompt template to use
|
||||
and values for the variables in the prompt template. The
|
||||
prompt template is identified by its template identifier e.g.
|
||||
question, extract-definitions. Template variable values are specified
|
||||
using key=value arguments on the command line, and these replace
|
||||
{{key}} placeholders in the template.
|
||||
"""
|
||||
|
||||
import argparse
|
||||
import os
|
||||
import json
|
||||
from trustgraph.clients.prompt_client import PromptClient
|
||||
|
||||
default_pulsar_host = os.getenv("PULSAR_HOST", 'pulsar://localhost:6650')
|
||||
|
||||
def query(pulsar_host, template_id, variables):
|
||||
|
||||
cli = PromptClient(pulsar_host=pulsar_host)
|
||||
|
||||
resp = cli.request(id=template_id, variables=variables)
|
||||
|
||||
if isinstance(resp, str):
|
||||
print(resp)
|
||||
else:
|
||||
print(json.dumps(resp, indent=4))
|
||||
|
||||
def main():
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
prog='tg-invoke-prompt',
|
||||
description=__doc__,
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-p', '--pulsar-host',
|
||||
default=default_pulsar_host,
|
||||
help=f'Pulsar host (default: {default_pulsar_host})',
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'id',
|
||||
metavar='template-id',
|
||||
nargs=1,
|
||||
help=f'Prompt identifier e.g. question, extract-definitions',
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'variable',
|
||||
nargs='*',
|
||||
metavar="variable=value",
|
||||
help='''Prompt template terms of the form variable=value, can be
|
||||
specified multiple times''',
|
||||
)
|
||||
|
||||
args = parser.parse_args()
|
||||
|
||||
variables = {}
|
||||
|
||||
for variable in args.variable:
|
||||
|
||||
toks = variable.split("=", 1)
|
||||
if len(toks) != 2:
|
||||
raise RuntimeError(f"Malformed variable: {variable}")
|
||||
|
||||
variables[toks[0]] = toks[1]
|
||||
|
||||
try:
|
||||
|
||||
query(
|
||||
pulsar_host=args.pulsar_host,
|
||||
template_id=args.id[0],
|
||||
variables=variables,
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
|
||||
print("Exception:", e, flush=True)
|
||||
|
||||
main()
|
||||
|
||||
|
|
@ -6,14 +6,23 @@ Loads a PDF document into TrustGraph processing.
|
|||
|
||||
import pulsar
|
||||
from pulsar.schema import JsonSchema
|
||||
from trustgraph.schema import Document, Source, document_ingest_queue
|
||||
import base64
|
||||
import hashlib
|
||||
import argparse
|
||||
import os
|
||||
import time
|
||||
import uuid
|
||||
|
||||
from trustgraph.schema import Document, document_ingest_queue
|
||||
from trustgraph.schema import Metadata
|
||||
from trustgraph.log_level import LogLevel
|
||||
from trustgraph.knowledge import hash, to_uri
|
||||
from trustgraph.knowledge import PREF_PUBEV, PREF_DOC, PREF_ORG
|
||||
from trustgraph.knowledge import Organization, PublicationEvent
|
||||
from trustgraph.knowledge import DigitalDocument
|
||||
|
||||
default_user = 'trustgraph'
|
||||
default_collection = 'default'
|
||||
|
||||
class Loader:
|
||||
|
||||
|
|
@ -21,7 +30,10 @@ class Loader:
|
|||
self,
|
||||
pulsar_host,
|
||||
output_queue,
|
||||
user,
|
||||
collection,
|
||||
log_level,
|
||||
metadata,
|
||||
):
|
||||
|
||||
self.client = pulsar.Client(
|
||||
|
|
@ -35,6 +47,10 @@ class Loader:
|
|||
chunking_enabled=True,
|
||||
)
|
||||
|
||||
self.user = user
|
||||
self.collection = collection
|
||||
self.metadata = metadata
|
||||
|
||||
def load(self, files):
|
||||
|
||||
for file in files:
|
||||
|
|
@ -47,13 +63,25 @@ class Loader:
|
|||
path = file
|
||||
data = open(path, "rb").read()
|
||||
|
||||
id = hashlib.sha256(path.encode("utf-8")).hexdigest()[0:8]
|
||||
# Create a SHA256 hash from the data
|
||||
id = hash(data)
|
||||
|
||||
id = to_uri(PREF_DOC, id)
|
||||
|
||||
triples = []
|
||||
|
||||
def emit(t):
|
||||
triples.append(t)
|
||||
|
||||
self.metadata.id = id
|
||||
self.metadata.emit(emit)
|
||||
|
||||
r = Document(
|
||||
source=Source(
|
||||
source=path,
|
||||
title=path,
|
||||
metadata=Metadata(
|
||||
id=id,
|
||||
metadata=triples,
|
||||
user=self.user,
|
||||
collection=self.collection,
|
||||
),
|
||||
data=base64.b64encode(data),
|
||||
)
|
||||
|
|
@ -71,7 +99,7 @@ class Loader:
|
|||
def main():
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
prog='loader',
|
||||
prog='tg-load-pdf',
|
||||
description=__doc__,
|
||||
)
|
||||
|
||||
|
|
@ -90,6 +118,66 @@ def main():
|
|||
help=f'Output queue (default: {default_output_queue})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-u', '--user',
|
||||
default=default_user,
|
||||
help=f'User ID (default: {default_user})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-c', '--collection',
|
||||
default=default_collection,
|
||||
help=f'Collection ID (default: {default_collection})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--name', help=f'Document name'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--description', help=f'Document description'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--copyright-notice', help=f'Copyright notice'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--copyright-holder', help=f'Copyright holder'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--copyright-year', help=f'Copyright year'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--license', help=f'Copyright license'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--publication-organization', help=f'Publication organization'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--publication-description', help=f'Publication description'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--publication-date', help=f'Publication date'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--url', help=f'Document URL'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--keyword', nargs='+', help=f'Keyword'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--identifier', '--id', help=f'Document ID'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-l', '--log-level',
|
||||
type=LogLevel,
|
||||
|
|
@ -109,10 +197,38 @@ def main():
|
|||
|
||||
try:
|
||||
|
||||
document = DigitalDocument(
|
||||
id,
|
||||
name=args.name,
|
||||
description=args.description,
|
||||
copyright_notice=args.copyright_notice,
|
||||
copyright_holder=args.copyright_holder,
|
||||
copyright_year=args.copyright_year,
|
||||
license=args.license,
|
||||
url=args.url,
|
||||
keywords=args.keyword,
|
||||
)
|
||||
|
||||
if args.publication_organization:
|
||||
org = Organization(
|
||||
id=to_uri(PREF_ORG, hash(args.publication_organization)),
|
||||
name=args.publication_organization,
|
||||
)
|
||||
document.publication = PublicationEvent(
|
||||
id = to_uri(PREF_PUBEV, str(uuid.uuid4())),
|
||||
organization=org,
|
||||
description=args.publication_description,
|
||||
start_date=args.publication_date,
|
||||
end_date=args.publication_date,
|
||||
)
|
||||
|
||||
p = Loader(
|
||||
pulsar_host=args.pulsar_host,
|
||||
output_queue=args.output_queue,
|
||||
user=args.user,
|
||||
collection=args.collection,
|
||||
log_level=args.log_level,
|
||||
metadata=document,
|
||||
)
|
||||
|
||||
p.load(args.files)
|
||||
|
|
|
|||
|
|
@ -6,14 +6,23 @@ Loads a text document into TrustGraph processing.
|
|||
|
||||
import pulsar
|
||||
from pulsar.schema import JsonSchema
|
||||
from trustgraph.schema import TextDocument, Source, text_ingest_queue
|
||||
import base64
|
||||
import hashlib
|
||||
import argparse
|
||||
import os
|
||||
import time
|
||||
import uuid
|
||||
|
||||
from trustgraph.schema import TextDocument, text_ingest_queue
|
||||
from trustgraph.schema import Metadata
|
||||
from trustgraph.log_level import LogLevel
|
||||
from trustgraph.knowledge import hash, to_uri
|
||||
from trustgraph.knowledge import PREF_PUBEV, PREF_DOC, PREF_ORG
|
||||
from trustgraph.knowledge import Organization, PublicationEvent
|
||||
from trustgraph.knowledge import DigitalDocument
|
||||
|
||||
default_user = 'trustgraph'
|
||||
default_collection = 'default'
|
||||
|
||||
class Loader:
|
||||
|
||||
|
|
@ -21,7 +30,10 @@ class Loader:
|
|||
self,
|
||||
pulsar_host,
|
||||
output_queue,
|
||||
user,
|
||||
collection,
|
||||
log_level,
|
||||
metadata,
|
||||
):
|
||||
|
||||
self.client = pulsar.Client(
|
||||
|
|
@ -35,6 +47,10 @@ class Loader:
|
|||
chunking_enabled=True,
|
||||
)
|
||||
|
||||
self.user = user
|
||||
self.collection = collection
|
||||
self.metadata = metadata
|
||||
|
||||
def load(self, files):
|
||||
|
||||
for file in files:
|
||||
|
|
@ -47,13 +63,25 @@ class Loader:
|
|||
path = file
|
||||
data = open(path, "rb").read()
|
||||
|
||||
id = hashlib.sha256(path.encode("utf-8")).hexdigest()[0:8]
|
||||
# Create a SHA256 hash from the data
|
||||
id = hash(data)
|
||||
|
||||
id = to_uri(PREF_DOC, id)
|
||||
|
||||
triples = []
|
||||
|
||||
def emit(t):
|
||||
triples.append(t)
|
||||
|
||||
self.metadata.id = id
|
||||
self.metadata.emit(emit)
|
||||
|
||||
r = TextDocument(
|
||||
source=Source(
|
||||
source=path,
|
||||
title=path,
|
||||
metadata=Metadata(
|
||||
id=id,
|
||||
metadata=triples,
|
||||
user=self.user,
|
||||
collection=self.collection,
|
||||
),
|
||||
text=data,
|
||||
)
|
||||
|
|
@ -71,7 +99,7 @@ class Loader:
|
|||
def main():
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
prog='loader',
|
||||
prog='tg-load-text',
|
||||
description=__doc__,
|
||||
)
|
||||
|
||||
|
|
@ -90,6 +118,66 @@ def main():
|
|||
help=f'Output queue (default: {default_output_queue})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-u', '--user',
|
||||
default=default_user,
|
||||
help=f'User ID (default: {default_user})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-c', '--collection',
|
||||
default=default_collection,
|
||||
help=f'Collection ID (default: {default_collection})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--name', help=f'Document name'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--description', help=f'Document description'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--copyright-notice', help=f'Copyright notice'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--copyright-holder', help=f'Copyright holder'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--copyright-year', help=f'Copyright year'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--license', help=f'Copyright license'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--publication-organization', help=f'Publication organization'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--publication-description', help=f'Publication description'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--publication-date', help=f'Publication date'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--url', help=f'Document URL'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--keyword', nargs='+', help=f'Keyword'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'--identifier', '--id', help=f'Document ID'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-l', '--log-level',
|
||||
type=LogLevel,
|
||||
|
|
@ -109,10 +197,38 @@ def main():
|
|||
|
||||
try:
|
||||
|
||||
document = DigitalDocument(
|
||||
id,
|
||||
name=args.name,
|
||||
description=args.description,
|
||||
copyright_notice=args.copyright_notice,
|
||||
copyright_holder=args.copyright_holder,
|
||||
copyright_year=args.copyright_year,
|
||||
license=args.license,
|
||||
url=args.url,
|
||||
keywords=args.keyword,
|
||||
)
|
||||
|
||||
if args.publication_organization:
|
||||
org = Organization(
|
||||
id=to_uri(PREF_ORG, hash(args.publication_organization)),
|
||||
name=args.publication_organization,
|
||||
)
|
||||
document.publication = PublicationEvent(
|
||||
id = to_uri(PREF_PUBEV, str(uuid.uuid4())),
|
||||
organization=org,
|
||||
description=args.publication_description,
|
||||
start_date=args.publication_date,
|
||||
end_date=args.publication_date,
|
||||
)
|
||||
|
||||
p = Loader(
|
||||
pulsar_host=args.pulsar_host,
|
||||
output_queue=args.output_queue,
|
||||
user=args.user,
|
||||
collection=args.collection,
|
||||
log_level=args.log_level,
|
||||
metadata=document,
|
||||
)
|
||||
|
||||
p.load(args.files)
|
||||
|
|
|
|||
160
trustgraph-cli/scripts/tg-load-turtle
Executable file
160
trustgraph-cli/scripts/tg-load-turtle
Executable file
|
|
@ -0,0 +1,160 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
"""
|
||||
Loads Graph embeddings into TrustGraph processing.
|
||||
"""
|
||||
|
||||
import pulsar
|
||||
from pulsar.schema import JsonSchema
|
||||
from trustgraph.schema import Triples, Triple, Value, Metadata
|
||||
from trustgraph.schema import triples_store_queue
|
||||
import argparse
|
||||
import os
|
||||
import time
|
||||
import pyarrow as pa
|
||||
import rdflib
|
||||
|
||||
from trustgraph.log_level import LogLevel
|
||||
|
||||
default_user = 'trustgraph'
|
||||
default_collection = 'default'
|
||||
default_pulsar_host = os.getenv("PULSAR_HOST", 'pulsar://localhost:6650')
|
||||
default_output_queue = triples_store_queue
|
||||
|
||||
class Loader:
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
pulsar_host,
|
||||
output_queue,
|
||||
log_level,
|
||||
files,
|
||||
user,
|
||||
collection,
|
||||
):
|
||||
|
||||
self.client = pulsar.Client(
|
||||
pulsar_host,
|
||||
logger=pulsar.ConsoleLogger(log_level.to_pulsar())
|
||||
)
|
||||
|
||||
self.producer = self.client.create_producer(
|
||||
topic=output_queue,
|
||||
schema=JsonSchema(Triples),
|
||||
chunking_enabled=True,
|
||||
)
|
||||
|
||||
self.files = files
|
||||
self.user = user
|
||||
self.collection = collection
|
||||
|
||||
def run(self):
|
||||
|
||||
try:
|
||||
|
||||
for file in self.files:
|
||||
self.load_file(file)
|
||||
|
||||
except Exception as e:
|
||||
print(e, flush=True)
|
||||
|
||||
def load_file(self, file):
|
||||
|
||||
g = rdflib.Graph()
|
||||
g.parse(file, format="turtle")
|
||||
|
||||
for e in g:
|
||||
s = Value(value=str(e[0]), is_uri=True)
|
||||
p = Value(value=str(e[1]), is_uri=True)
|
||||
if type(e[2]) == rdflib.term.URIRef:
|
||||
o = Value(value=str(e[2]), is_uri=True)
|
||||
else:
|
||||
o = Value(value=str(e[2]), is_uri=False)
|
||||
|
||||
r = Triples(
|
||||
metadata=Metadata(
|
||||
id=None,
|
||||
metadata=[],
|
||||
user=self.user,
|
||||
collection=self.collection,
|
||||
),
|
||||
triples=[ Triple(s=s, p=p, o=o) ]
|
||||
)
|
||||
|
||||
self.producer.send(r)
|
||||
|
||||
def __del__(self):
|
||||
self.client.close()
|
||||
|
||||
def main():
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
prog='tg-load-turtle',
|
||||
description=__doc__,
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-p', '--pulsar-host',
|
||||
default=default_pulsar_host,
|
||||
help=f'Pulsar host (default: {default_pulsar_host})',
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-o', '--output-queue',
|
||||
default=default_output_queue,
|
||||
help=f'Output queue (default: {default_output_queue})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-u', '--user',
|
||||
default=default_user,
|
||||
help=f'User ID (default: {default_user})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-c', '--collection',
|
||||
default=default_collection,
|
||||
help=f'Collection ID (default: {default_collection})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-l', '--log-level',
|
||||
type=LogLevel,
|
||||
default=LogLevel.ERROR,
|
||||
choices=list(LogLevel),
|
||||
help=f'Output queue (default: info)'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'files', nargs='+',
|
||||
help=f'Turtle files to load'
|
||||
)
|
||||
|
||||
args = parser.parse_args()
|
||||
|
||||
while True:
|
||||
|
||||
try:
|
||||
p = Loader(
|
||||
pulsar_host=args.pulsar_host,
|
||||
output_queue=args.output_queue,
|
||||
log_level=args.log_level,
|
||||
files=args.files,
|
||||
user=args.user,
|
||||
collection=args.collection,
|
||||
)
|
||||
|
||||
p.run()
|
||||
|
||||
print("File loaded.")
|
||||
break
|
||||
|
||||
except Exception as e:
|
||||
|
||||
print("Exception:", e, flush=True)
|
||||
print("Will retry...", flush=True)
|
||||
|
||||
time.sleep(10)
|
||||
|
||||
main()
|
||||
|
||||
|
|
@ -9,17 +9,19 @@ import os
|
|||
from trustgraph.clients.document_rag_client import DocumentRagClient
|
||||
|
||||
default_pulsar_host = os.getenv("PULSAR_HOST", 'pulsar://localhost:6650')
|
||||
default_user = 'trustgraph'
|
||||
default_collection = 'default'
|
||||
|
||||
def query(pulsar, query):
|
||||
def query(pulsar_host, query, user, collection):
|
||||
|
||||
rag = DocumentRagClient(pulsar_host=pulsar)
|
||||
resp = rag.request(query)
|
||||
resp = rag.request(user=user, collection=collection, query=query)
|
||||
print(resp)
|
||||
|
||||
def main():
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
prog='graph-show',
|
||||
prog='tg-query-document-rag',
|
||||
description=__doc__,
|
||||
)
|
||||
|
||||
|
|
@ -35,11 +37,28 @@ def main():
|
|||
help=f'Query to execute',
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-u', '--user',
|
||||
default=default_user,
|
||||
help=f'User ID (default: {default_user})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-c', '--collection',
|
||||
default=default_collection,
|
||||
help=f'Collection ID (default: {default_collection})'
|
||||
)
|
||||
|
||||
args = parser.parse_args()
|
||||
|
||||
try:
|
||||
|
||||
query(args.pulsar_host, args.query)
|
||||
query(
|
||||
pulsar_host=args.pulsar_host,
|
||||
query=args.query,
|
||||
user=args.user,
|
||||
collection=args.collection,
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
|
||||
|
|
|
|||
|
|
@ -9,17 +9,19 @@ import os
|
|||
from trustgraph.clients.graph_rag_client import GraphRagClient
|
||||
|
||||
default_pulsar_host = os.getenv("PULSAR_HOST", 'pulsar://localhost:6650')
|
||||
default_user = 'trustgraph'
|
||||
default_collection = 'default'
|
||||
|
||||
def query(pulsar, query):
|
||||
def query(pulsar_host, query, user, collection):
|
||||
|
||||
rag = GraphRagClient(pulsar_host=pulsar)
|
||||
resp = rag.request(query)
|
||||
rag = GraphRagClient(pulsar_host=pulsar_host)
|
||||
resp = rag.request(user=user, collection=collection, query=query)
|
||||
print(resp)
|
||||
|
||||
def main():
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
prog='graph-show',
|
||||
prog='tg-graph-query-rag',
|
||||
description=__doc__,
|
||||
)
|
||||
|
||||
|
|
@ -35,11 +37,28 @@ def main():
|
|||
help=f'Query to execute',
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-u', '--user',
|
||||
default=default_user,
|
||||
help=f'User ID (default: {default_user})'
|
||||
)
|
||||
|
||||
parser.add_argument(
|
||||
'-c', '--collection',
|
||||
default=default_collection,
|
||||
help=f'Collection ID (default: {default_collection})'
|
||||
)
|
||||
|
||||
args = parser.parse_args()
|
||||
|
||||
try:
|
||||
|
||||
query(args.pulsar_host, args.query)
|
||||
query(
|
||||
pulsar_host=args.pulsar_host,
|
||||
query=args.query,
|
||||
user=args.user,
|
||||
collection=args.collection,
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
|
||||
|
|
|
|||
|
|
@ -34,7 +34,7 @@ setuptools.setup(
|
|||
python_requires='>=3.8',
|
||||
download_url = "https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v" + version + ".tar.gz",
|
||||
install_requires=[
|
||||
"trustgraph-base<0.12",
|
||||
"trustgraph-base>=0.15,<0.16",
|
||||
"requests",
|
||||
"pulsar-client",
|
||||
"rdflib",
|
||||
|
|
@ -46,9 +46,13 @@ setuptools.setup(
|
|||
"scripts/tg-init-pulsar-manager",
|
||||
"scripts/tg-load-pdf",
|
||||
"scripts/tg-load-text",
|
||||
"scripts/tg-load-turtle",
|
||||
"scripts/tg-query-document-rag",
|
||||
"scripts/tg-query-graph-rag",
|
||||
"scripts/tg-init-pulsar",
|
||||
"scripts/tg-processor-state",
|
||||
"scripts/tg-invoke-agent",
|
||||
"scripts/tg-invoke-prompt",
|
||||
"scripts/tg-invoke-llm",
|
||||
]
|
||||
)
|
||||
|
|
|
|||
|
|
@ -34,8 +34,8 @@ setuptools.setup(
|
|||
python_requires='>=3.8',
|
||||
download_url = "https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v" + version + ".tar.gz",
|
||||
install_requires=[
|
||||
"trustgraph-base<0.12",
|
||||
"trustgraph-flow<0.12",
|
||||
"trustgraph-base>=0.15,<0.16",
|
||||
"trustgraph-flow>=0.15,<0.16",
|
||||
"torch",
|
||||
"urllib3",
|
||||
"transformers",
|
||||
|
|
|
|||
6
trustgraph-flow/scripts/agent-manager-react
Normal file
6
trustgraph-flow/scripts/agent-manager-react
Normal file
|
|
@ -0,0 +1,6 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
from trustgraph.agent.react import run
|
||||
|
||||
run()
|
||||
|
||||
6
trustgraph-flow/scripts/text-completion-azure-openai
Executable file
6
trustgraph-flow/scripts/text-completion-azure-openai
Executable file
|
|
@ -0,0 +1,6 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
from trustgraph.model.text_completion.azure_openai import run
|
||||
|
||||
run()
|
||||
|
||||
6
trustgraph-flow/scripts/text-completion-googleaistudio
Executable file
6
trustgraph-flow/scripts/text-completion-googleaistudio
Executable file
|
|
@ -0,0 +1,6 @@
|
|||
#!/usr/bin/env python3
|
||||
|
||||
from trustgraph.model.text_completion.googleaistudio import run
|
||||
|
||||
run()
|
||||
|
||||
|
|
@ -34,7 +34,7 @@ setuptools.setup(
|
|||
python_requires='>=3.8',
|
||||
download_url = "https://github.com/trustgraph-ai/trustgraph/archive/refs/tags/v" + version + ".tar.gz",
|
||||
install_requires=[
|
||||
"trustgraph-base<0.12",
|
||||
"trustgraph-base>=0.15,<0.16",
|
||||
"urllib3",
|
||||
"rdflib",
|
||||
"pymilvus",
|
||||
|
|
@ -55,8 +55,12 @@ setuptools.setup(
|
|||
"openai",
|
||||
"neo4j",
|
||||
"tiktoken",
|
||||
"google-generativeai",
|
||||
"ibis",
|
||||
"jsonschema",
|
||||
],
|
||||
scripts=[
|
||||
"scripts/agent-manager-react",
|
||||
"scripts/chunker-recursive",
|
||||
"scripts/chunker-token",
|
||||
"scripts/de-query-milvus",
|
||||
|
|
@ -83,8 +87,10 @@ setuptools.setup(
|
|||
"scripts/rows-write-cassandra",
|
||||
"scripts/run-processing",
|
||||
"scripts/text-completion-azure",
|
||||
"scripts/text-completion-azure-openai",
|
||||
"scripts/text-completion-claude",
|
||||
"scripts/text-completion-cohere",
|
||||
"scripts/text-completion-googleaistudio",
|
||||
"scripts/text-completion-llamafile",
|
||||
"scripts/text-completion-ollama",
|
||||
"scripts/text-completion-openai",
|
||||
|
|
|
|||
0
trustgraph-flow/trustgraph/agent/__init__.py
Normal file
0
trustgraph-flow/trustgraph/agent/__init__.py
Normal file
19
trustgraph-flow/trustgraph/agent/react/README.md
Normal file
19
trustgraph-flow/trustgraph/agent/react/README.md
Normal file
|
|
@ -0,0 +1,19 @@
|
|||
|
||||
agent-manager-react \
|
||||
-p pulsar://localhost:6650 \
|
||||
--tool-type \
|
||||
shuttle=knowledge-query:query \
|
||||
cats=knowledge-query:query \
|
||||
compute=text-completion:computation \
|
||||
--tool-description \
|
||||
shuttle="Query a knowledge base with information about the space shuttle. The query should be a simple natural language question" \
|
||||
cats="Query a knowledge base with information about Mark's cats. The query should be a simple natural language question" \
|
||||
compute="A computation engine which can answer questions about maths and computation" \
|
||||
--tool-argument \
|
||||
cats="query:string:The search query string" \
|
||||
shuttle="query:string:The search query string" \
|
||||
compute="computation:string:The computation to solve"
|
||||
|
||||
|
||||
--context 'The space shuttle challenger final mission was 58-L'
|
||||
|
||||
Some files were not shown because too many files have changed in this diff Show more
Loading…
Add table
Add a link
Reference in a new issue