Use Metrics to Scale Model Serving Deployments in Kubernetes

Authors: Xiake Sun, Su Yang, Tianmeng Chen, Tong Qiu

openvino.genai/samples/cpp/rag_sample at openvino_genai_serving · sammysun0711/openvino.genai (github.com)

OpenVINO GenAI Server (OGS) Update:

-Update LLM: stream generation, reset handle, multi-round chat, model cache config

-Support VLM

-Support Reranker for RAG sample

-Support BLIP image embedding for photo search with DB

-Support C++ GUI with imgui for photo search

Now we scale the text embedding to image embedding for RAG sample and support multi-Vector Retriever for RAG.

1. Multi-Vector Retriever for RAG on text: QA over Document
2. Multi-Vector Retriever for RAG on image: Photo search with DB retrieval

‍

Here is a photo search sample with image embedding.

Usage 2: Photo Search with DB retrieval

Steps:

1.use python client to create image vector DB (PostgreSQL)

2.use GUI to search image

Here is a sample image to demonstrate GUI usage on client platform. we search the bus photo with top 10 similar images from the 100 images which are embedded into Vector DB.

Usage 3: Chat with images via MiniCPM-V

Once we have created a multimodal vector DB through image embedding, we can further communicate with the image through VLM.

We integrate the C++ GenAI sample visual_language_chat with openbmb/MiniCPM-V-2_6.

Here is the demo image on client platform.

‍

Authors: Tianmeng Chen, Xiake Sun

Introduction

Stable Diffusion is a generative artificial intelligence model that produces unique images from text and image prompts. ControlNet is a neural network that controls image generation in Stable Diffusion by adding extra conditions. The specific structure of Stable Diffusion + ControlNet is shown below:

In many cases, ControlNet is used in conjunction with other models or frameworks, such as OpenPose, Canny, Line Art, Depth, etc. An example of Stable Diffusion + ControlNet + OpenPose:

OpenPose identifies the key points of the human body from the left image to get the pose image, and then inputs the Pose image to ControlNet and Stable Diffusion to get the right image. In this way, ControlNet can control the generation of Stable Diffusion.

In this blog, we focus on enabling the stable diffusion pipeline with ControlNet in Optimum-intel. Some details can be found in this open PR.

How to enable StableDiffusionControlNet pipeline in Optimum-Intel

The important code is in optimum/intel/openvino/modelling_diffusion.py and optimum/exporters/openvino/model_configs.py. There is the diffusion pipeline related code in file modelling_diffusion.py, you can find several Class: OVStableDiffusionPipelineBase, OVStableDiffusionPipeline, OVStableDiffusionXLPipelineBase, OVStableDiffusionXLPipeline, and so on. What we need to do is mimic these base classes to add the OVStableDiffusionControlNetPipelineBase, StableDiffusionContrlNetPipelineMixin, and OVStableDiffusionControlNetPipeline. A few of the important parts are as follows:

_from_pretrained function in class OVStableDiffusionControlNetPipelineBase: initial whole pipeline from local or download.

_from_transformers function in class OVStableDiffusionControlNetPipelineBase: convert torch model to OpenVINO IR model.

_reshape_unet_controlnet and _reshape_controlnet in class OVStableDiffusionControlNetPipelineBase: reshape dynamic OpenVINO IR model to static in order to decrease cost.

__call__ function in class StableDiffusionContrlNetPipelineMixin: do the inference in the pipeline.

In model_configs.py, we define UNetControlNetOpenVINOConfig by inheriting UNetOnnxConfig, which includes UNetControlNet inputs and outputs.

By now we have completed the rough code, after which some very detailed code additions are needed, so I won't go into that here.

How to use StableDiffusionControlNet pipeline via Optimum-Intel

The next step is how to use the code, examples of which can be found in this repository.

Installation and update of environments and dependencies from source. Make sure your python version is greater that 3.10 and your optimum-intel and optimum version is up to date accounding to the requirements.txt.

# %python -m venv stable-diffusion-controlnet
# %source stable-diffusion-controlnet/bin/activate
%pip install -r requirements.txt

‍

At first, we should convert pytorch model to openvino IR with dynamic shape. Now import related packages.

from optimum.intel import OVStableDiffusionControlNetPipeline
import os
from diffusers import UniPCMultistepScheduler

‍

Set pytroch models of stable diffusion 1.5 and controlnet path if you have them in local, else you can run pipeline from download.

SD15_PYTORCH_MODEL_DIR="stable-diffusion-v1-5"
CONTROLNET_PYTORCH_MODEL_DIR="control_v11p_sd15_openpose"


if os.path.exists(SD15_PYTORCH_MODEL_DIR) and os.path.exists(CONTROLNET_PYTORCH_MODEL_DIR):
    scheduler = UniPCMultistepScheduler.from_config("scheduler_config.json")
    ov_pipe = OVStableDiffusionControlNetPipeline.from_pretrained(SD15_PYTORCH_MODEL_DIR, controlnet_model_id=CONTROLNET_PYTORCH_MODEL_DIR, compile=False, export=True, scheduler=scheduler,device="GPU.1")
    ov_pipe.save_pretrained(save_directory="./ov_models_dynamic")
    print("Dynamic model is saved in ./ov_models_dynamic")  

else:
    scheduler = UniPCMultistepScheduler.from_config("scheduler_config.json")
    ov_pipe = OVStableDiffusionControlNetPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", controlnet_model_id="lllyasviel/control_v11p_sd15_openpose", compile=False, export=True, scheduler=scheduler, device="GPU.1")
    ov_pipe.save_pretrained(save_directory="./ov_models_dynamic")
    print("Dynamic model is saved in ./ov_models_dynamic")

‍

Now you will have openvino IR models file under **ov_models_dynamic ** folder.

from optimum.intel import OVStableDiffusionControlNetPipeline
from controlnet_aux import OpenposeDetector
from pathlib import Path
import numpy as np
import os
from PIL import Image
from diffusers import UniPCMultistepScheduler
import requests
import torch

‍

We recommand to use static shape model to decrease GPU memory cost. Set your STATIC_SHAPE and DEVICE_NAME.

NEED_STATIC = True
STATIC_SHAPE = [1024,1024]
DEVICE_NAME = "GPU.1"

‍

Load openvino model files, if is static, reshape dynamic models to fixed shape.

if NEED_STATIC:
    print("Using static models")
    scheduler = UniPCMultistepScheduler.from_config("scheduler_config.json")
    ov_config ={"CACHE_DIR": "", 'INFERENCE_PRECISION_HINT': 'f16'}
    if not os.path.exists("ov_models_static"):
        if os.path.exists("ov_models_dynamic"):
            print("load dynamic models from local ov files and reshape to static")
            ov_pipe = OVStableDiffusionControlNetPipeline.from_pretrained(Path("ov_models_dynamic"), scheduler=scheduler, device=DEVICE_NAME, compile=True, ov_config=ov_config, height=STATIC_SHAPE[0], width=STATIC_SHAPE[1])
            ov_pipe.reshape(batch_size=1 ,height=STATIC_SHAPE[0], width=STATIC_SHAPE[1], num_images_per_prompt=1)
            ov_pipe.save_pretrained(save_directory="./ov_models_static")
            print("Static model is saved in ./ov_models_static")  
        else:
            raise ValueError("No ov_models_dynamic exists, please trt ov_model_export.py first")
    else:
        print("load static models from local ov files")
        ov_pipe = OVStableDiffusionControlNetPipeline.from_pretrained(Path("ov_models_static"), scheduler=scheduler, device=DEVICE_NAME, compile=True, ov_config=ov_config, height=STATIC_SHAPE[0], width=STATIC_SHAPE[1])
else:
    scheduler = UniPCMultistepScheduler.from_config("scheduler_config.json")
    ov_config ={"CACHE_DIR": "", 'INFERENCE_PRECISION_HINT': 'f16'}
    print("load dynamic models from local ov files")
    ov_pipe = OVStableDiffusionControlNetPipeline.from_pretrained(Path("ov_models_dynamic"), scheduler=scheduler, device=DEVICE_NAME, compile=True, ov_config=ov_config)
     

‍

Set seed for Numpy and torch to make result reproducible.

seed = 42
torch.manual_seed(seed)           
torch.cuda.manual_seed(seed)       
torch.cuda.manual_seed_all(seed)
np.random.seed(seed)

‍

Load image for ControlNet, or you can use your own image, or generate image with OpenPose OpenVINO model, notice that OpenPose model is not supported by OVStableDiffusionControlNetPipeline yet, so you need to convert it to openvino model first manually. Here we use directly the result from OpenPose:

pose = Image.open(Path("pose_1024.png"))

‍

Set prompt, negative_prompt, image inputs.

prompt = "Dancing Darth Vader, best quality, extremely detailed"
negative_prompt = "monochrome, lowres, bad anatomy, worst quality, low quality"

result = ov_pipe(prompt=prompt, image=pose, num_inference_steps=20, negative_prompt=negative_prompt, height=STATIC_SHAPE[0], width=STATIC_SHAPE[1])

result[0].save("result_1024.png")

‍

Authors :  Qiu Tong, Zhao Hongbo

MeloTTS released by MyShell.ai, is a high-quality, multilingual Text-to-Speech (TTS) library that supports English, Chinese (mixed English), and various other languages. The strengths of the model lie in its lightweight design, which is well-suited for applications on AIPC systems, coupled with its impressive performance. In this article, I will guide you through the process of converting the model to be compatible with OpenVINO toolkits, enabling it to run on various devices such as CPUs, GPUs and NPUs. Additionally, I will provide a concise overview of the model's inference procedure.

‍

Overview of Model Inference Procedure and Pipeline

For each language type, the pipeline requires only two models (two inference procedures). For instance, English language generation necessitates just the 'bert-base-uncased' model and its corresponding MeloTTS-English. Similarly, for Chinese language generation (which includes mixed English), the pipeline needs only the 'bert-base-multilingual-uncased' and MeloTTS-Chinese. This greatly streamlines the pipeline compared to other TTS frameworks, and the compact size of the models makes them suitable for deployment on edge devices.

MeloTTS is based on Variational Inference with adversarial learning for end-to-end Text-to-Speech (VITS). The inference process is illustrated in the figure. It encompasses a text encoder, a stochastic duration predictor, a decoder.

The text encoder accepts phones, tones and a hidden layer from a BERT model as input. It then produces the text encoder's output along with its mean value and a logarithmic variance. To align the input texts with the target speech, the outputs from the encoders are processed using a stochastic duration predictor, which generates an alignment matrix. This matrix is then used to expand the mean value and a logarithmic variance (assuming a Gaussian distribution) to obtain the results for the latent variables .Subsequently, the inverse flow transformation is applied to obtain the distribution of final latent variable z, which represents the spectrogram. In the decoder, by upsampling the spectrogram, the final audio waveform is obtained.

‍

def ov_infer(self, phones=None, phones_length=None, speaker_id=None, tones=None, lang_ids=None, bert=None, ja_bert=None, sdp_ratio=0.2, noise_scale=0.6, noise_scale_w=0.8, speed=1.0):

‍

The inference entry is the function. In practical inference, phone refers to the distinct speech sounds, while tone refers to the vocal pitch contour. For Chinese, a phone corresponds to pinyin, and a tone corresponds to one of the four tones. In English, phones are the consonants and vowels, and tones relate to stress patterns. Here, noise_scale and noise_scale_w do not refer to actual noise. Both noise_scale_w and noise_scale are components within the Stochastic Duration Predictor, used to introduce randomness in order to enhance the expressiveness of the model.

‍

Note that MeloTTS does not include a voice cloning component, unlike the majority of other TTS  models, which makes it more lightweight. If voice cloning is required, please refer to OpenVoice.

‍

Enable Model for OpenVINO

As previously mentioned, the pipeline requires just two models for each language. Taking English as an example, we must first convert both 'bert-base-uncased' and 'MeloTTS-English' into the OpenVINO IR format.

        example_input={
                "x": x_tst,
                "x_lengths": x_tst_lengths,
                "sid": speakers,
                "tone": tones,
                "language": lang_ids,
                "bert": bert,
                "ja_bert": ja_bert,
                "noise_scale": noise_scale,
                "length_scale": length_scale,
                "noise_scale_w": noise_scale_w,
                "sdp_ratio": sdp_ratio,
            }
            
        ov_model = ov.convert_model(
            self.model,
            example_input=example_input,
        )
        get_input_names = lambda: ["phones", "phones_length", "speakers",
                                  "tones", "lang_ids", "bert", "ja_bert",
                                  "noise_scale", "length_scale", "noise_scale_w", "sdp_ratio"]
        for input, input_name in zip(ov_model.inputs, get_input_names()):
            input.get_tensor().set_names({input_name})
        outputs_name = ['audio']
        for output, output_name in zip(ov_model.outputs, outputs_name):
            output.get_tensor().set_names({output_name})
        """
        reshape model
        Set the batch size of all input tensors to 1
        """   
        shapes = {}     
        for input_layer  in ov_model.inputs:
            shapes[input_layer] = input_layer.partial_shape
            shapes[input_layer][0] = 1
        ov_model.reshape(shapes)

        ov.save_model(ov_model, Path(ov_model_path))

‍

For instance, we convert the MeloTTS-English model from the pytorch format directly by utilizing the openvino.convert_model API along with pseudo input data.

Note that the input and output layers (it is optional) are renamed to facilitate subsequent development. Furthermore, the batch dimension for all inputs is fixed at 1, as multiple batches are not required here (this is also optional).

‍

We further quantized both the BERT and TTS models to int8 using pseudo data. We observed that our method of quantizing the TTS model introduces a slight distortion to the current sound. To suppress this, we implemented DeepFilterNet, which is also very lightweight.

‍

More about model conversion and int8 quantization please refer to MeloTTS-OV .

‍

Run BERT part on NPU

To enhance performance and reduce CPU offloading, we can shift the execution of the BERT model to the NPU on Meteor Lake.

To adapt the model for the NPU, we've converted the model to accept static shape inputs a and pad each input during inference.

def reshape_for_npu(model, bert_static_shape = 32):
        # change dynamic shape to static shape
        shapes = dict()
        for input_layer  in model.inputs:
            shapes[input_layer] = bert_static_shape
        model.reshape(shapes)
        ov.save_model(model, Path(ov_model_save_path))
        print(f"save static model in {Path(ov_model_save_path)}")

def main():
    core = Core()
    model = core.read_model(ov_model_path)
    reshape_for_npu(model, bert_static_shape=bert_static_shape)

‍

Simple Demo

Here are the audio files generated by the int8 quantized model from OpenVINO.

https://github.com/zhaohb/MeloTTS-OV/tree/speech-enhancement-and-npu/demo

‍

‍

‍