apunkt/plano
1
0
Fork 0
mirror of https://github.com/katanemo/plano.git synced 2026-06-29 15:49:40 +02:00
Code Issues Projects Releases Packages Wiki Activity

use archfc hosted on aws (#409)

Browse source
This commit is contained in:
Adil Hafeez 2025-02-13 11:03:34 -08:00 • committed by GitHub
parent 28ee14a75c
commit 4ec03af16e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 10 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

2
model_server/src/commons/globals.py
View file

@ -15,7 +15,7 @@ logger = get_model_server_logger()
# Define the client # Define the client
ARCH_ENDPOINT = os.getenv("ARCH_ENDPOINT", "https://api.fc.archgw.com/v1") ARCH_ENDPOINT = os.getenv("ARCH_ENDPOINT", "https://archfc.katanemo.dev/v1")
ARCH_API_KEY = "EMPTY" ARCH_API_KEY = "EMPTY"
ARCH_CLIENT = OpenAI(base_url=ARCH_ENDPOINT, api_key=ARCH_API_KEY) ARCH_CLIENT = OpenAI(base_url=ARCH_ENDPOINT, api_key=ARCH_API_KEY)

4
model_server/src/main.py
View file

@ -4,7 +4,7 @@ import time
import logging import logging
import src.commons.utils as utils import src.commons.utils as utils
from src.commons.globals import handler_map from src.commons.globals import ARCH_ENDPOINT, handler_map
from src.core.utils.model_utils import ( from src.core.utils.model_utils import (
ChatMessage, ChatMessage,
ChatCompletionResponse, ChatCompletionResponse,
@ -51,6 +51,8 @@ logging.getLogger("opentelemetry.exporter.otlp.proto.grpc.exporter").setLevel(
app = FastAPI() app = FastAPI()
FastAPIInstrumentor().instrument_app(app) FastAPIInstrumentor().instrument_app(app)
logger.info(f"using archfc endpoint: {ARCH_ENDPOINT}")
@app.get("/healthz") @app.get("/healthz")
async def healthz(): async def healthz():

5
tests/rest/api_model_server.rest
View file

@ -1,6 +1,5 @@
@model_server_endpoint = http://localhost:51000 @model_server_endpoint = http://localhost:51000
@archfc_endpoint = https://api.fc.archgw.com @archfc_endpoint = https://archfc.katanemo.dev
### talk to function calling endpoint ### talk to function calling endpoint
POST {{model_server_endpoint}}/function_calling HTTP/1.1 POST {{model_server_endpoint}}/function_calling HTTP/1.1
@ -119,7 +118,7 @@ Content-Type: application/json
} }
### talk to Arch-Intent directly for completion ### talk to Arch-Intent directly for completion
POST {{archfc_endpoint}}/v1/chat/completions HTTP/1.1 POST {{{{archfc_endpoint}}}}/v1/chat/completions HTTP/1.1
Content-Type: application/json Content-Type: application/json
{ {

6
tests/rest/insurance_agent.rest
View file

@ -1,5 +1,5 @@
@model_server_endpoint = http://localhost:51000 @model_server_endpoint = http://localhost:51000
@archfc_endpoint = https://api.fc.archgw.com @archfc_endpoint = https://archfc.katanemo.dev
### multi turn conversation with intent, except parameter gathering ### multi turn conversation with intent, except parameter gathering
@ -55,7 +55,7 @@ Content-Type: application/json
] ]
} }
### talk to Arch-Intent directly for completion ### talk to Arch-Intent directly for completion
POST https://api.fc.archgw.com/v1/chat/completions HTTP/1.1 POST https://archfc.katanemo.dev/v1/chat/completions HTTP/1.1
Content-Type: application/json Content-Type: application/json
{ {
@ -126,7 +126,7 @@ Content-Type: application/json
] ]
} }
### talk to Arch-Intent directly for completion, expect No ### talk to Arch-Intent directly for completion, expect No
POST https://api.fc.archgw.com/v1/chat/completions HTTP/1.1 POST https://archfc.katanemo.dev/v1/chat/completions HTTP/1.1
Content-Type: application/json Content-Type: application/json
{ {

2
tests/rest/network_agent.rest
View file

@ -1,5 +1,5 @@
@model_server_endpoint = http://localhost:51000 @model_server_endpoint = http://localhost:51000
@archfc_endpoint = https://api.fc.archgw.com @archfc_endpoint = https://archfc.katanemo.dev
### single turn function calling all parameters insurance agent summary ### single turn function calling all parameters insurance agent summary