koaning

Marimo notebook assistant

I am a specialized AI assistant designed to help create data science notebooks using marimo. I focus on creating clear, efficient, and reproducible data analysis workflows with marimo's reactive programming model.

<assistant_info>

• I specialize in data science and analytics using marimo notebooks
• I provide complete, runnable code that follows best practices
• I emphasize reproducibility and clear documentation
• I focus on creating interactive data visualizations and analysis
• I understand marimo's reactive programming model

boxabirds

# Project Policy

This policy provides a single, authoritative, and machine-readable source of truth for AI coding agents and humans, ensuring that all work is governed by clear, unambiguous rules and workflows. It aims to eliminate ambiguity, reduce supervision needs, and facilitate automation while maintaining accountability and compliance with best practices.

# 1. Introduction

> Rationale: Sets the context, actors, and compliance requirements for the policy, ensuring all participants understand their roles and responsibilities.

## 1.1 Actors

boxabirds

# Written by Julian Harris https://makingaiagents.substack.com
# julian.harris+maia@gmail.co
#                                  Apache License
#                            Version 2.0, January 2004
#                         http://www.apache.org/licenses/

#    TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION

#    1. Definitions.

kalomaze

the generic basics of preference reward modeling

The Bradley-Terry model works like this:

• It's based on a chosen/rejected split
• The model is trained on binary judgements of specific content/samples as being either 'preferred' or 'dispreferred'
• The log ratio between preferred and dispreferred can be used as the natural reward signal

leehanchung

flowchart TD
    %% Main Entry Point and Initialization
    Start([Application Start]) --> ParseArgs[Parse Command Line Arguments]
    ParseArgs --> CheckStdin{Check stdin}
    CheckStdin -->|Input Piped| ProcessPiped[Process Piped Input]
    CheckStdin -->|Interactive Mode| CheckAuth{Authentication Status}
    
    %% Authentication Flow
    CheckAuth -->|Not Authenticated| OAuthFlow[OAuth Authentication]

kalomaze

# the "verifiers" repository is a clean implementation of templated GRPO reinforcement learning training environments
# this is a generic set of "install from scratch" commands complete with a deepspeed z3 config that i have been using when i spin up nodes
# it will run on the gsm8k example w/ default batch size & generation size (8), and the 8th GPU is used for vllm generations
# qwen 14b full finetuning will run on this configuration too without LoRA or CUDA OOM, at least for the gsm8k task's context sizes + generation lengths
# hyperparameters are controlled by `verifiers/utils/config_utils.py`; i have been preferring extreme grad clipping (between 0.001 and 0.01) and low beta (under 0.01)

# NOTE FEB 27: examples have moved into `verifiers/examples` not `/examples`

cd /root
mkdir boom

echohive42

{
  "timestamp": "2025-01-25T21:47:37.634489",
  "question": "why does language exist?",
  "n_approaches": 5,
  "approach_list": [
    "Anthropological perspective",
    "Evolutionary biology",
    "Cognitive science",
    "Theoretical linguistics",
    "Sociocultural analysis"

kalomaze

class RescaleDescentTrainer(Trainer):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        # Initialize all buffers
        self.tokens_buffer = []          # for raw token loss
        self.weighted_tokens_buffer = [] # for entropy weighted token loss
        self.unigram_rate_buffer = []
        self.bigram_rate_buffer = []
        self.trigram_rate_buffer = []
        self.weighted_unigram_buffer = []

pyros-projects

Technical Project Planning Meta-Prompt

You are an expert software architect and technical project planner. Your task is to create a comprehensive technical implementation plan for a software project based on the provided inputs.

User Input

do you know googly python-fire? Python Fire is a library for automatically generating command line interfaces (CLIs) from absolutely any Python object. I want a similar library, but instead of a CLI it generates amazing web apps for any python project!

Output Format

pyros-projects

Technical Project Planning Meta-Prompt

You are an expert software architect and technical project planner. Your task is to create a comprehensive technical implementation plan for a software project based on the provided inputs.

User Input

You are an expert developer capable of building a repository analysis application. Your goal is to create a comprehensive plan for implementing a web application that retrieves a GitHub repository identifier (e.g., 127.0.0.1:5001/kohya-ss/sd-scripts) and produces an interactive report usable as context for LLMs.

The web application should accept a GitHub repository ID in the specified URL format and generate an interactive repository report. This report should provide details such as code file descriptions and allow filtering by file extension. The app should collect all relevant information from the repository that an LLM would typically need to make informed decisions. It is up to you to decide what type of information to include, while also offering users the