mirror of
https://github.com/mudler/LocalAI.git
synced 2025-05-20 02:24:59 +00:00
chore(autogptq): drop archived backend (#5214)
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
This commit is contained in:
Ettore Di Giacinto
GitHub
parent
8abecb4a18
commit
61cc76c455
23 changed files with 5 additions and 322 deletions
|
|
@ -190,11 +190,7 @@ message ModelOptions {
|
|||
int32 NGQA = 20;
|
||||
string ModelFile = 21;
|
||||
|
||||
// AutoGPTQ
|
||||
string Device = 22;
|
||||
bool UseTriton = 23;
|
||||
string ModelBaseName = 24;
|
||||
bool UseFastTokenizer = 25;
|
||||
|
||||
|
||||
// Diffusers
|
||||
string PipelineType = 26;
|
||||
|
|
|
|||
|
|
@ -1,17 +0,0 @@
|
|||
.PHONY: autogptq
|
||||
autogptq: protogen
|
||||
bash install.sh
|
||||
|
||||
.PHONY: protogen
|
||||
protogen: backend_pb2_grpc.py backend_pb2.py
|
||||
|
||||
.PHONY: protogen-clean
|
||||
protogen-clean:
|
||||
$(RM) backend_pb2_grpc.py backend_pb2.py
|
||||
|
||||
backend_pb2_grpc.py backend_pb2.py:
|
||||
python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto
|
||||
|
||||
.PHONY: clean
|
||||
clean: protogen-clean
|
||||
rm -rf venv __pycache__
|
||||
|
|
@ -1,5 +0,0 @@
|
|||
# Creating a separate environment for the autogptq project
|
||||
|
||||
```
|
||||
make autogptq
|
||||
```
|
||||
|
|
@ -1,158 +0,0 @@
|
|||
#!/usr/bin/env python3
|
||||
from concurrent import futures
|
||||
import argparse
|
||||
import signal
|
||||
import sys
|
||||
import os
|
||||
import time
|
||||
import base64
|
||||
|
||||
import grpc
|
||||
import backend_pb2
|
||||
import backend_pb2_grpc
|
||||
|
||||
from auto_gptq import AutoGPTQForCausalLM
|
||||
from transformers import AutoTokenizer, AutoModelForCausalLM
|
||||
from transformers import TextGenerationPipeline
|
||||
|
||||
_ONE_DAY_IN_SECONDS = 60 * 60 * 24
|
||||
|
||||
# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
|
||||
MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
|
||||
|
||||
# Implement the BackendServicer class with the service methods
|
||||
class BackendServicer(backend_pb2_grpc.BackendServicer):
|
||||
def Health(self, request, context):
|
||||
return backend_pb2.Reply(message=bytes("OK", 'utf-8'))
|
||||
def LoadModel(self, request, context):
|
||||
try:
|
||||
device = "cuda:0"
|
||||
if request.Device != "":
|
||||
device = request.Device
|
||||
|
||||
# support loading local model files
|
||||
model_path = os.path.join(os.environ.get('MODELS_PATH', './'), request.Model)
|
||||
tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=True, trust_remote_code=request.TrustRemoteCode)
|
||||
|
||||
# support model `Qwen/Qwen-VL-Chat-Int4`
|
||||
if "qwen-vl" in request.Model.lower():
|
||||
self.model_name = "Qwen-VL-Chat"
|
||||
model = AutoModelForCausalLM.from_pretrained(model_path,
|
||||
trust_remote_code=request.TrustRemoteCode,
|
||||
device_map="auto").eval()
|
||||
else:
|
||||
model = AutoGPTQForCausalLM.from_quantized(model_path,
|
||||
model_basename=request.ModelBaseName,
|
||||
use_safetensors=True,
|
||||
trust_remote_code=request.TrustRemoteCode,
|
||||
device=device,
|
||||
use_triton=request.UseTriton,
|
||||
quantize_config=None)
|
||||
|
||||
self.model = model
|
||||
self.tokenizer = tokenizer
|
||||
except Exception as err:
|
||||
return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
|
||||
return backend_pb2.Result(message="Model loaded successfully", success=True)
|
||||
|
||||
def Predict(self, request, context):
|
||||
penalty = 1.0
|
||||
if request.Penalty != 0.0:
|
||||
penalty = request.Penalty
|
||||
tokens = 512
|
||||
if request.Tokens != 0:
|
||||
tokens = request.Tokens
|
||||
top_p = 0.95
|
||||
if request.TopP != 0.0:
|
||||
top_p = request.TopP
|
||||
|
||||
|
||||
prompt_images = self.recompile_vl_prompt(request)
|
||||
compiled_prompt = prompt_images[0]
|
||||
print(f"Prompt: {compiled_prompt}", file=sys.stderr)
|
||||
|
||||
# Implement Predict RPC
|
||||
pipeline = TextGenerationPipeline(
|
||||
model=self.model,
|
||||
tokenizer=self.tokenizer,
|
||||
max_new_tokens=tokens,
|
||||
temperature=request.Temperature,
|
||||
top_p=top_p,
|
||||
repetition_penalty=penalty,
|
||||
)
|
||||
t = pipeline(compiled_prompt)[0]["generated_text"]
|
||||
print(f"generated_text: {t}", file=sys.stderr)
|
||||
|
||||
if compiled_prompt in t:
|
||||
t = t.replace(compiled_prompt, "")
|
||||
# house keeping. Remove the image files from /tmp folder
|
||||
for img_path in prompt_images[1]:
|
||||
try:
|
||||
os.remove(img_path)
|
||||
except Exception as e:
|
||||
print(f"Error removing image file: {img_path}, {e}", file=sys.stderr)
|
||||
|
||||
return backend_pb2.Result(message=bytes(t, encoding='utf-8'))
|
||||
|
||||
def PredictStream(self, request, context):
|
||||
# Implement PredictStream RPC
|
||||
#for reply in some_data_generator():
|
||||
# yield reply
|
||||
# Not implemented yet
|
||||
return self.Predict(request, context)
|
||||
|
||||
def recompile_vl_prompt(self, request):
|
||||
prompt = request.Prompt
|
||||
image_paths = []
|
||||
|
||||
if "qwen-vl" in self.model_name.lower():
|
||||
# request.Images is an array which contains base64 encoded images. Iterate the request.Images array, decode and save each image to /tmp folder with a random filename.
|
||||
# Then, save the image file paths to an array "image_paths".
|
||||
# read "request.Prompt", replace "[img-%d]" with the image file paths in the order they appear in "image_paths". Save the new prompt to "prompt".
|
||||
for i, img in enumerate(request.Images):
|
||||
timestamp = str(int(time.time() * 1000)) # Generate timestamp
|
||||
img_path = f"/tmp/vl-{timestamp}.jpg" # Use timestamp in filename
|
||||
with open(img_path, "wb") as f:
|
||||
f.write(base64.b64decode(img))
|
||||
image_paths.append(img_path)
|
||||
prompt = prompt.replace(f"[img-{i}]", "<img>" + img_path + "</img>,")
|
||||
else:
|
||||
prompt = request.Prompt
|
||||
return (prompt, image_paths)
|
||||
|
||||
def serve(address):
|
||||
server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS),
|
||||
options=[
|
||||
('grpc.max_message_length', 50 * 1024 * 1024), # 50MB
|
||||
('grpc.max_send_message_length', 50 * 1024 * 1024), # 50MB
|
||||
('grpc.max_receive_message_length', 50 * 1024 * 1024), # 50MB
|
||||
])
|
||||
backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
|
||||
server.add_insecure_port(address)
|
||||
server.start()
|
||||
print("Server started. Listening on: " + address, file=sys.stderr)
|
||||
|
||||
# Define the signal handler function
|
||||
def signal_handler(sig, frame):
|
||||
print("Received termination signal. Shutting down...")
|
||||
server.stop(0)
|
||||
sys.exit(0)
|
||||
|
||||
# Set the signal handlers for SIGINT and SIGTERM
|
||||
signal.signal(signal.SIGINT, signal_handler)
|
||||
signal.signal(signal.SIGTERM, signal_handler)
|
||||
|
||||
try:
|
||||
while True:
|
||||
time.sleep(_ONE_DAY_IN_SECONDS)
|
||||
except KeyboardInterrupt:
|
||||
server.stop(0)
|
||||
|
||||
if __name__ == "__main__":
|
||||
parser = argparse.ArgumentParser(description="Run the gRPC server.")
|
||||
parser.add_argument(
|
||||
"--addr", default="localhost:50051", help="The address to bind the server to."
|
||||
)
|
||||
args = parser.parse_args()
|
||||
|
||||
serve(args.addr)
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
#!/bin/bash
|
||||
set -e
|
||||
|
||||
source $(dirname $0)/../common/libbackend.sh
|
||||
|
||||
# This is here because the Intel pip index is broken and returns 200 status codes for every package name, it just doesn't return any package links.
|
||||
# This makes uv think that the package exists in the Intel pip index, and by default it stops looking at other pip indexes once it finds a match.
|
||||
# We need uv to continue falling through to the pypi default index to find optimum[openvino] in the pypi index
|
||||
# the --upgrade actually allows us to *downgrade* torch to the version provided in the Intel pip index
|
||||
if [ "x${BUILD_PROFILE}" == "xintel" ]; then
|
||||
EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
|
||||
fi
|
||||
|
||||
installRequirements
|
||||
|
|
@ -1,2 +0,0 @@
|
|||
--extra-index-url https://download.pytorch.org/whl/cu118
|
||||
torch==2.4.1+cu118
|
||||
|
|
@ -1 +0,0 @@
|
|||
torch==2.4.1
|
||||
|
|
@ -1,2 +0,0 @@
|
|||
--extra-index-url https://download.pytorch.org/whl/rocm6.0
|
||||
torch==2.4.1+rocm6.0
|
||||
|
|
@ -1,6 +0,0 @@
|
|||
--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
|
||||
intel-extension-for-pytorch==2.3.110+xpu
|
||||
torch==2.3.1+cxx11.abi
|
||||
oneccl_bind_pt==2.3.100+xpu
|
||||
optimum[openvino]
|
||||
setuptools
|
||||
|
|
@ -1,6 +0,0 @@
|
|||
accelerate
|
||||
auto-gptq==0.7.1
|
||||
grpcio==1.71.0
|
||||
protobuf
|
||||
certifi
|
||||
transformers
|
||||
|
|
@ -1,4 +0,0 @@
|
|||
#!/bin/bash
|
||||
source $(dirname $0)/../common/libbackend.sh
|
||||
|
||||
startBackend $@
|
||||
|
|
@ -1,6 +0,0 @@
|
|||
#!/bin/bash
|
||||
set -e
|
||||
|
||||
source $(dirname $0)/../common/libbackend.sh
|
||||
|
||||
runUnittests
|
||||
Loading…
Add table
Add a link
Reference in a new issue