hugobowne

# Dropout in Neural Networks

This summary provides an overview of recent research findings and perspectives on dropout as a regularization technique in neural networks based on the latest arXiv papers.

## Summary of Work

Several recent studies explore the role and effects of dropout in neural networks across a range of contexts including Bayesian deep learning, model robustness, and uncertainty estimation. Dropout is commonly used to prevent overfitting by randomly dropping units during training, which helps the model generalize better.

One notable approach is the combination of dropout with Bayesian inference methods, such as Monte Carlo (MC) dropout, to quantify predictive uncertainty and improve calibration of probabilistic forecasts. For instance, an ensemble technique utilizing initial weights combined with MC dropout was shown to enhance both forecast skill and uncertainty calibration in convective initiation nowcasting.

hugobowne

# Small and Large Language Models

## Research Question
Understanding the distinctions, collaboration, performance, and application of small language models (SLMs) versus large language models (LLMs).

## Summary of Work

1. **Dynamic Scoring with Enhanced Semantics for Training-Free HOI Detection** explores Vision-Language models leveraging small and large features for interaction detection without training, emphasizing multi-head attention to fuse visual and textual features.

2. **DynaSearcher** proposes a dynamic knowledge graph augmented search agent using multi-reward reinforcement learning. It uses small models effectively for retrieval tasks, achieving results comparable to large models but with fewer resources.

hugobowne

# Small and Large Language Models

This summary provides insights from recent papers related to small and large language models (LLMs) and their applications, performance, and efficiency improvements.

## Summary of Work

1. **Dynamic Scoring with Enhanced Semantics for HOI Detection**
   - Focus on Vision-Language Models (VLMs) improving human-object interaction detection without heavy training.
   - Utilizes small visual cues and textual features for robust multimodal understanding.

hugobowne

# Summary of Research on Small Large Language Models

This summary provides an overview of recent research related to the application and development of small large language models (LLMs) or techniques leveraging small models in conjunction with larger ones to improve efficiency, accuracy, or scalability.

## Summary of Work

1. **Dynamic Scoring with Enhanced Semantics for Training-Free Human-Object Interaction Detection**
   - This work explores the usage of vision-language models which can include smaller models for enhanced semantic understanding without extensive training, showing competitive results especially on rare interactions.

2. **DynaSearcher: Dynamic Knowledge Graph Augmented Search Agent via Multi-Reward Reinforcement Learning**

hugobowne

# Should I Prioritize Recall or Precision in RAG Pipelines?

## Short Description

This report reviews recent research on the trade-offs and impacts of prioritizing recall versus precision when building Retrieval-Augmented Generation (RAG) pipelines. RAG combines retrieval of relevant documents with generative models and is widely used in question answering and other NLP applications that require external knowledge integration.

## Summary of Work

Recent papers emphasize hybrid and graph-enhanced retrieval mechanisms that improve both recall and precision simultaneously. For example, BifrostRAG introduces a dual-graph retrieval structure for multi-hop QA, achieving around 92.8% precision and 85.5% recall, showing that balancing both can significantly enhance results.

hugobowne

hugobowne

import pandas as pd
import pandas_profiling

# Create small df
data = {"name": ["Hugo", "Ville"], "city": ["Sydney", "SF"]}
df = pd.DataFrame(data)

# Create report and save to file
profile = pandas_profiling.ProfileReport(df)
profile.to_file("df_report.html")

hugobowne

# scheme.py
#
# Challenge:  Can you implement a mini-scheme interpreter (program that's running another program) capable of
# executing the following code (now at bottom of file):

def seval(sexp, env):
    if isinstance(sexp, (int, float)):
        return sexp
    elif isinstance(sexp, str):  #Symbols
        return env[sexp]  #Evaluate symbol names in the 'env'

hugobowne

class MyStreamListener(tweepy.StreamListener):
    def __init__(self, api=None):
        super(MyStreamListener, self).__init__()
        self.num_tweets = 0
        self.file = open("tweets.txt", "w")

    def on_status(self, status):
        tweet = status._json
        self.file.write( json.dumps(tweet) + '\n' )
        self.num_tweets += 1

hugobowne

These are the materials for my workshop on creating interactive data visualizations with D3! We will be using the following two tools to works through these exercises:

• bl.ocks.org: D3.js Gist renderer and online community
• Building Blocks: In-browser interactive D3.js bl.ock editor

And please do not hesitate to reach out to me directly via email at [email protected] or over twitter @clearspandex

Throughout this workshop, you will learn how to make an interactive map of AirBnB listings in SF to better understand the companies impact on the city.