kinchungwong

[
  {
    "namespace": "llama_index.agent.openai",
    "classes": [
      "OpenAIAgent",
      "OpenAIAgentWorker",
      "OpenAIAssistantAgent"
    ]
  },
  {

kinchungwong

Grace, would you like to answer Frank's question about CPU-based SIMD programming, especially in relation to AVX-512 and NEON architectures?
--------------------------------------------------------------------------------
Bob:
> > > > > > Grace wrote : > > > > > > > Grace: Certainly, Alice! CPU-based SIMD programming plays a pivotal role when optimizing for performance in functions like OpenCV's remap. With AVX-512, we can process multiple pixels in parallel, dramatically speeding up image transformations. The ability to use gather instructions allows us to quickly pull together non-contiguous data points that are often found in image processing tasks. On ARM architectures, NEON also provides similar capabilities, albeit with a different instruction set. Therefore, when developing the remap function, it is essential to take into account how these SIMD instructions can be utilized effectively. A rigorous evaluation of the memory access patterns should be conducted to ensure that both AVX-512 and NEON can be em

kinchungwong

chat_facade_openai_crazy = ChatFacade(client_openai, "gpt-4o-mini", logging_func=chat_log_fn, temperature=1.6, stream=True, stream_logging_func=chat_log_fn)

kinchungwong

chat_facade_openai = ChatFacade(client_openai, "gpt-3.5-turbo", logging_func=chat_log_fn)
chat_facade_openai_crazy = ChatFacade(client_openai, "gpt-3.5-turbo", logging_func=chat_log_fn, temperature=1.6, stream=True, stream_logging_func=chat_log_fn)
chat_facade_perplexity = ChatFacade(client_perplexity, "sonar", logging_func=chat_log_fn)
chat_facade_local_ollama = ChatFacade(client_local_ollama, "deepseek-r1:8b", logging_func=chat_log_fn)


session.add_person(alice, chat_facade_openai, common_rewriter, common_response_picker)
session.add_person(bob, chat_facade_openai, common_rewriter, common_response_picker)
session.add_person(charlie, chat_facade_openai, common_rewriter, common_response_picker)
session.add_person(dave, chat_facade_openai_crazy, common_rewriter, common_response_picker)

kinchungwong

[{'content': "Let me introduce today's attendees.", 'role': 'system'},
 {'content': "Alice is the host for today's meeting.", 'role': 'system'},
 {'content': 'Bob specializes in modern day general purpose CPU architectures, '
             'especially with out-of-order executions and memory coherency.',
  'role': 'system'},
 {'content': 'Charlie specializes in multi-core and many-core architectures.',
 {'content': 'Dave specializes in memory protocols and components used in '
             "today's consumer-grade CPUs.",
 {'content': 'Erin specializes in Massively Parallel Computations.',
 {'content': 'Frank specializes in CUDA programming. CUDA is a parallel '

kinchungwong

[{'content': "Let me introduce today's attendees.", 'role': 'system'},
 {'content': '', 'role': 'system'},
 {'content': "Alice is the host for today's meeting.", 'role': 'system'},
 {'content': 'Bob specializes in modern day general purpose CPU architectures, '
             'especially with out-of-order executions and memory coherency.',
  'role': 'system'},
 {'content': 'Cathy specializes in multi-core and many-core architectures.',
 {'content': 'Dean specializes in memory protocols and components used in '
             "today's consumer-grade CPUs.",
 {'content': 'Erin specializes in Massively Parallel Computations.',

kinchungwong

import os
from openai import OpenAI
import numpy as np
import json
from collections.abc import Sequence
from typing import Any, Optional, Union
import random
from utils import sanitize_python_codestring

class ApiKeyLoader:

kinchungwong

import os
from openai import OpenAI
import numpy as np
import json
from collections.abc import Sequence
from typing import Any, Optional, Union

class ApiKeyLoader:
    def __init__(self, base_path: str, prefix: str, suffix: str, expanduser: bool = True) -> None:
        if expanduser:

kinchungwong

from abc import ABC as Abstract
from abc import abstractmethod
from collections.abc import Sequence
from enum import Enum
from typing import Any, ForwardRef, overload, Optional, Union
import math
from bisect import bisect_right


import cv2 as cv

kinchungwong

Untitled_2024_11_20_image_server_design_question.md

image storage system

Requirements overview

• Total number of images stored
  • Near-term future forecasts
• Statistical distribution of image formats (codecs), image resolutions, and file sizes
• Statistical distribution of traffic patterns