Supplementary Data: Benchmarking Document Reading Systems on Small-Scale Hardware

Read full text Summarise with VIRAT
Authors: Mehmet Emir San
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Supplementary materials for BSc thesis: Benchmarking Document Reading Systems on Small-Scale Hardware. Contains model outputs, LLM-as-a-Judge evaluation results (input and output JSONL files), and performance measurements (TTFT, TPS) for seven models evaluated on the DocBench benchmark. Models include: llama-3-8b, gemma3n:e2b, phi3:3.8b, qwen3:4b (local via Ollama), and vllm-llama-4-scout-17b-16e-Instruct, vllm-meta-llama-llama-3-3-70b-Instruct, vllm-nvidia-llama-3-3-70b-Instruct-fp8 (remote). Raw DocBench PDF documents are excluded as they are available from the original DocBench repository.

Fields of Study & Subjects

Topic Modeling Natural Language Processing Techniques Handwritten Text Recognition Techniques

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
20 March 2026
Language
EN