Add missing arguments in pipeline parallel generate method

[notsyncing]

Description

Add two arguments: negative_prompt_ids and negative_prompt_attention_mask to the generate method in pipeline_parallel.py. These two arguments are available since transformers 4.32.0.

1. Why the change?

Hello, I'm using both OpenVINO and ipex-llm in one project, and found the generate method in ipex-llm's pipeline-parallel.py has two arguments less than the generate method in transformers, causing a weird error in an OpenVINO-only code path:

Traceback (most recent call last):
  File "/mnt/data/podman/test.py", line 34, in <module>
    outputs = model.generate(**params)
  File "/mnt/data/podman/.local/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/mnt/data/podman/.local/lib/python3.10/site-packages/optimum/intel/openvino/modeling_decoder.py", line 673, in generate
    result = super().generate(
  File "/mnt/data/podman/.local/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
TypeError: generate() takes from 1 to 9 positional arguments but 11 were given

With this PR, the error is gone.

2. User API changes

No change

3. Summary of the change

Add the two missing arguments to the generate method in pipeline_parallel.py

4. How to test?

Install optimum[openvino]==1.22.0, then use the following script to reproduce and check:

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from optimum.intel import OVModelForCausalLM
import ipex_llm.transformers   # This line will lead to the error above

device = "GPU.0" # the device to load the model onto

model = OVModelForCausalLM.from_pretrained(
    "Qwen/Qwen2-7B-Instruct",  # or any smaller model
    export=True,
    use_cache=False,
    device=device
)
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2-7B-Instruct")

messages = [
    {"role": "user", "content": "你好"}
]

inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to(model.device)

params = {
    "inputs": inputs,
    "max_new_tokens": 512,
    "do_sample": True,
    "temperature": 1,
    "top_p": 0.95
}

outputs = model.generate(**params)
res = tokenizer.decode(outputs[0][len(inputs[0]):], skip_special_tokens=True)
print(res)

This does not happen on optimum[openvino]<=1.20.0.