Скачать или смотреть Challenges and Applications of Large Language Models

This paper aims to identify the challenges and application successes of large language models (LLMs) in order to help researchers understand the current state of the field and be more productive.

[Truncated]

00:00 Section: 1 Introduction
03:09 Section: Benchmark Data Contamination
06:36 Section: Pre-Training Domain Mixtures
10:07 Section: 2.2 Tokenizer-Reliance
12:45 Section: Subword-Level Inputs
16:42 Section: 2.3 High Pre-Training Costs
21:18 Section: Pre-Training Objectives
25:20 Section: Parallelism Strategies
29:03 Section: Miscellaneous
32:03 Section: Parameter-efficient fine-tuning
36:22 Section: Compute Requirements
39:30 Section: Quantization
43:14 Section: Mixture-of-Experts

https://arxiv.org/abs//2307.10169

YouTube:    / @arxivpapers  

PODCASTS:
Apple Podcasts: https://podcasts.apple.com/us/podcast...
Spotify: https://podcasters.spotify.com/pod/sh...