razhangwei

• quite powerful as it simplies the writing the system prompt

• often need an extra step of post processing, which can be done by LLM or traditional programs.

• structured output vs json mode:

  • structured output: 100% gurantee schema; schema doesn't consume tokens

• json: scheme usually goes to system prompts, costing tokens, doesn't guarantee json or scheme correctness.

razhangwei

pipx install llm
llm install llm-gemini 
llm -m gemini-1.5-flash-latest

razhangwei

## sudo visudo -f /etc/sudoers.d/custom_config


# /etc/sudoers.d/custom_config

# Extend the sudo ticket lifetime to 2 hours (7200 seconds)
Defaults timestamp_timeout=7200

# Allow your user to run the script without a password
wei ALL=(ALL) NOPASSWD: /Users/wei/start_nano_transcriber.sh

razhangwei

class FeedbackManager:
    """
    Manage user feedback by printing a message and printing dots after it.

    This class takes care of printing a message and then appending dots to it
    until told to clear. It also handles clearing the feedback by deleting the
    message and dots.

    The printing of dots is done in a separate thread to avoid blocking.
    """

razhangwei

class AudioRecorder:
    def __init__(self):

        self.is_recording = False
        self.recording_queue = queue.Queue()
        self.stop_recording = threading.Event()
        self.recording_thread = None

    def start_recording(self):
        if not self.is_recording:

razhangwei

Jinja

Go

Delimiters:

{{ and }} are used to enclose template actions.

Conditionals:

razhangwei

import torch
import torch.nn as nn
import torch.nn.functional as F

class NF4Quantizer(nn.Module):
    def __init__(self):
        super().__init__()
        self.nf4_values = torch.tensor([
            -1.0, -0.6961928009986877, -0.5250730514526367, -0.39491748809814453,
            -0.28444138169288635, -0.18477343022823334, -0.09105003625154495, 0.0,

razhangwei

1. install with pipx
2. ` pipx run cookiecutter gh:audreyfeldroy/cookiecutter-pypackage`

razhangwei

from pynput import keyboard
from pynput.keyboard import Key, Controller

# For simulating keyboard presses
keyboard_controller = Controller()

def on_press(key):
    try:
        print(f'Alphanumeric key pressed: {key.char}')
    except AttributeError:

razhangwei

import math
import torch
import torch.nn as nn

class RotaryPositionEmbedding(nn.Module):
    """
    Implements Rotary Position Embedding (RoPE) as a PyTorch module.
    
    Args:
        dim (int): Dimension of the embedding (must be even)