Add gigachat

assistant_dists/universal_prompted_assistant/dev.yml

@@ -71,6 +71,12 @@ services:
       - "./common:/src/common"
     ports:
       - 8180:8180
+  gigachat-api:
+    volumes:
+      - "./services/gigachat_api_lm:/src"
+      - "./common:/src/common"
+    ports:
+      - 8187:8187
   anthropic-api-claude-v1:
     volumes:
       - "./services/anthropic_api_lm:/src"

assistant_dists/universal_prompted_assistant/docker-compose.override.yml

@@ -6,7 +6,7 @@ services:
         sentence-ranker:8128, 
         transformers-lm-gptjt:8161, openai-api-chatgpt:8145, openai-api-davinci3:8131,
         openai-api-gpt4:8159, openai-api-gpt4-32k:8160, openai-api-chatgpt-16k:8167,
-        openai-api-gpt4-turbo:8180, dff-universal-prompted-skill:8147"
+        openai-api-gpt4-turbo:8180, gigachat-api:8187, dff-universal-prompted-skill:8147"
       WAIT_HOSTS_TIMEOUT: ${WAIT_TIMEOUT:-1000}
 
   sentseg:
@@ -219,6 +219,25 @@ services:
         reservations:
           memory: 100M
 
+  gigachat-api:
+    env_file: [ .env ]
+    build:
+      args:
+        SERVICE_PORT: 8187
+        SERVICE_NAME: gigachat_api
+        PRETRAINED_MODEL_NAME_OR_PATH: GigaChat:1.3.23.1
+      context: .
+      dockerfile: ./services/gigachat_api_lm/Dockerfile
+    command: flask run -h 0.0.0.0 -p 8187
+    environment:
+      - FLASK_APP=server
+    deploy:
+      resources:
+        limits:
+          memory: 500M
+        reservations:
+          memory: 100M
+
   anthropic-api-claude-v1:
     env_file: [ .env ]
     build:

assistant_dists/universal_ru_prompted_assistant/dev.yml

@@ -67,6 +67,12 @@ services:
       - "./common:/src/common"
     ports:
       - 8180:8180
+  gigachat-api:
+    volumes:
+      - "./services/gigachat_api_lm:/src"
+      - "./common:/src/common"
+    ports:
+      - 8187:8187
   dff-universal-ru-prompted-skill:
     volumes:
       - "./skills/dff_universal_prompted_skill:/src"

assistant_dists/universal_ru_prompted_assistant/docker-compose.override.yml

@@ -213,6 +213,25 @@ services:
         reservations:
           memory: 100M
 
+  gigachat-api:
+    env_file: [ .env ]
+    build:
+      args:
+        SERVICE_PORT: 8187
+        SERVICE_NAME: gigachat_api
+        PRETRAINED_MODEL_NAME_OR_PATH: GigaChat:1.3.23.1
+      context: .
+      dockerfile: ./services/gigachat_api_lm/Dockerfile
+    command: flask run -h 0.0.0.0 -p 8187
+    environment:
+      - FLASK_APP=server
+    deploy:
+      resources:
+        limits:
+          memory: 500M
+        reservations:
+          memory: 100M
+
   dff-universal-ru-prompted-skill:
     env_file: [ .env_ru ]
     build:

common/generative_configs/gigachat.json

@@ -0,0 +1,7 @@
+{
+  "max_tokens": 256,
+  "temperature": 0.4,
+  "top_p": 1.0,
+  "frequency_penalty": 0,
+  "presence_penalty": 0
+}

components.tsv

@@ -190,7 +190,7 @@
 8184	external-fake-server
 8185	transformers-mistral-7b-128k
 8186	dff-document-qa-transformers-llm-skill
-8187
+8187    gigachat-api
 8188
 8189
 8190

components/0bBDINLSJDnjn1pzf8sdA.yml

@@ -0,0 +1,25 @@
+name: gigachat
+display_name: GigaChat
+component_type: Generative
+model_type: NN-based
+is_customizable: false
+author: publisher@deeppavlov.ai
+description: GigaChat is a service that can interact with the user in a dialogue format, write code, 
+  create texts and pictures at the user’s request. At the same time, 
+  GigaChat considers controversial issues or provocations.
+ram_usage: 100M
+gpu_usage: null
+group: services
+connector:
+  protocol: http
+  timeout: 120.0
+  url: http://gigachat-api:8187/respond
+dialog_formatter: null
+response_formatter: null
+previous_services: null
+required_previous_services: null
+state_manager_method: null
+tags: null
+endpoint: respond
+service: services/gigachat_api_lm/service_configs/gigachat-api
+date_created: '2023-12-25T09:45:32'

services/gigachat_api_lm/Dockerfile

@@ -0,0 +1,14 @@
+FROM python:3.10
+
+WORKDIR /src
+
+COPY ./services/gigachat_api_lm/requirements.txt /src/requirements.txt
+RUN pip install -r /src/requirements.txt
+
+ARG PRETRAINED_MODEL_NAME_OR_PATH
+ENV PRETRAINED_MODEL_NAME_OR_PATH ${PRETRAINED_MODEL_NAME_OR_PATH}
+
+COPY services/gigachat_api_lm /src
+COPY common /src/common
+
+CMD gunicorn --workers=1 server:app -b 0.0.0.0:${SERVICE_PORT} --timeout=300

services/gigachat_api_lm/requirements.txt

@@ -0,0 +1,9 @@
+flask==1.1.1
+itsdangerous==2.0.1
+gunicorn==19.9.0
+requests==2.22.0
+sentry-sdk[flask]==0.14.1
+healthcheck==1.3.3
+jinja2<=3.0.3
+Werkzeug<=2.0.3
+gigachat==0.1.10

services/gigachat_api_lm/server.py

@@ -0,0 +1,116 @@
+import json
+import logging
+import os
+import time
+import sentry_sdk
+
+from gigachat import GigaChat
+from gigachat.models import Chat
+from flask import Flask, request, jsonify
+from sentry_sdk.integrations.flask import FlaskIntegration
+
+
+sentry_sdk.init(dsn=os.getenv("SENTRY_DSN"), integrations=[FlaskIntegration()])
+
+
+logging.basicConfig(
+    format="%(asctime)s - %(name)s - %(levelname)s - %(message)s", level=logging.INFO
+)
+logger = logging.getLogger(__name__)
+
+PRETRAINED_MODEL_NAME_OR_PATH = os.environ.get("PRETRAINED_MODEL_NAME_OR_PATH")
+logger.info(f"PRETRAINED_MODEL_NAME_OR_PATH = {PRETRAINED_MODEL_NAME_OR_PATH}")
+GIGACHAT_ROLES = ["assistant", "user"]
+
+app = Flask(__name__)
+logging.getLogger("werkzeug").setLevel("WARNING")
+DEFAULT_CONFIGS = {
+    "GigaChat:1.3.23.1": json.load(
+        open("common/generative_configs/gigachat.json", "r")
+    ),
+}
+
+
+def generate_responses(
+    context,
+    gigachat_api_key,
+    gigachat_org,
+    prompt,
+    generation_params,
+    continue_last_uttr=False,
+):
+    assert gigachat_api_key, logger.error(
+        "Error: GigaChat API key is not specified in env"
+    )
+    giga = GigaChat(credentials=gigachat_api_key, verify_ssl_certs=False)
+
+    s = len(context) % 2
+    messages = [
+        {"role": "system", "content": prompt},
+    ]
+    messages += [
+        {
+            "role": f"{GIGACHAT_ROLES[(s + uttr_id) % 2]}",
+            "content": uttr,
+        }
+        for uttr_id, uttr in enumerate(context)
+    ]
+    logger.info(f"context inside generate_responses seen as: {messages}")
+
+    payload = Chat(messages=messages, scope=gigachat_org, **generation_params)
+    response = giga.chat(payload)
+
+    outputs = [resp.message.content.strip() for resp in response.choices]
+
+    return outputs
+
+
+@app.route("/ping", methods=["POST"])
+def ping():
+    return "pong"
+
+
+@app.route("/respond", methods=["POST"])
+def respond():
+    st_time = time.time()
+    contexts = request.json.get("dialog_contexts", [])
+    prompts = request.json.get("prompts", [])
+    configs = request.json.get("configs", None)
+    configs = [None] * len(prompts) if configs is None else configs
+    configs = [
+        DEFAULT_CONFIGS[PRETRAINED_MODEL_NAME_OR_PATH] if el is None else el
+        for el in configs
+    ]
+
+    if len(contexts) > 0 and len(prompts) == 0:
+        prompts = [""] * len(contexts)
+
+    gigachat_api_keys = request.json.get("gigachat_credentials", [])
+    gigachat_orgs = request.json.get("gigachat_scopes", None)
+    gigachat_orgs = [None] * len(contexts) if gigachat_orgs is None else gigachat_orgs
+
+    try:
+        responses = []
+        for context, gigachat_api_key, gigachat_org, prompt, config in zip(
+            contexts, gigachat_api_keys, gigachat_orgs, prompts, configs
+        ):
+            curr_responses = []
+            outputs = generate_responses(
+                context, gigachat_api_key, gigachat_org, prompt, config
+            )
+            for response in outputs:
+                if len(response) >= 2:
+                    curr_responses += [response]
+                else:
+                    curr_responses += [""]
+            responses += [curr_responses]
+
+    except Exception as exc:
+        logger.exception(exc)
+        sentry_sdk.capture_exception(exc)
+        responses = [[""]] * len(contexts)
+
+    logger.info(f"gigachat-api result: {responses}")
+    total_time = time.time() - st_time
+    logger.info(f"gigachat-api exec time: {total_time:.3f}s")
+    return jsonify(responses)

services/gigachat_api_lm/service_configs/gigachat-api/environment.yml

@@ -0,0 +1,4 @@
+SERVICE_PORT: 8187
+SERVICE_NAME: gigachat_api
+PRETRAINED_MODEL_NAME_OR_PATH: GigaChat:1.3.23.1
+FLASK_APP: server

services/gigachat_api_lm/service_configs/gigachat-api/service.yml

@@ -0,0 +1,30 @@
+name: gigachat-api
+endpoints:
+- respond
+- generate_goals
+compose:
+  env_file:
+  - .env
+  build:
+    args:
+      SERVICE_PORT: 8187
+      SERVICE_NAME: gigachat_api
+      PRETRAINED_MODEL_NAME_OR_PATH: GigaChat:1.3.23.1
+      FLASK_APP: server
+    context: .
+    dockerfile: ./services/gigachat_api_lm/Dockerfile
+  command: flask run -h 0.0.0.0 -p 8187
+  environment:
+  - FLASK_APP=server
+  deploy:
+    resources:
+      limits:
+        memory: 100M
+      reservations:
+        memory: 100M
+  volumes:
+  - ./services/gigachat_api_lm:/src
+  - ./common:/src/common
+  ports:
+  - 8187:8187
+proxy: null

skills/dff_universal_prompted_skill/scenario/response.py

@@ -41,6 +41,7 @@
     "http://transformers-lm-vicuna13b:8168/respond": [],
     "http://transformers-lm-ruxglm:8171/respond": [],
     "http://transformers-lm-rugpt35:8178/respond": [],
+    "http://gigachat-api:8187/respond": ["GIGACHAT_CREDENTIAL", "GIGACHAT_SCOPE"],
 }