| --- |
| license: cc-by-sa-4.0 |
| pretty_name: "The Language–Energy Divide" |
| language: |
| - en |
| - zh |
| - ru |
| - fr |
| - ps |
| - ti |
| - shn |
| - bo |
| multilinguality: |
| - multilingual |
| task_categories: |
| - question-answering |
| - text-generation |
| tags: |
| - energy-efficiency |
| - green-ai |
| - multilingual |
| - llm-inference |
| - belebele |
| - energy-measurement |
| size_categories: |
| - 1K<n<10K |
| configs: |
| - config_name: belebele_qwen3-8b_122lang |
| data_files: results/belebele_canonical_Qwen3-8B_0shot.csv |
| --- |
| |
| <div align="center"> |
|
|
| # 🌍⚡ The Language–Energy Divide |
|
|
| ### Per-language energy measurements & prompts for multilingual LLM inference |
|
|
| [](https://arxiv.org/abs/XXXX.XXXXX) |
| [](https://github.com/MichiganNLP/language-energy-divide) |
|
|
| </div> |
|
|
| This dataset accompanies the paper **"The Language–Energy Divide: Measuring Energy Costs of |
| Multilingual LLM Inference."** It releases the **per-language energy measurements** and the |
| **prompts** used in the study, so researchers can build on our numbers without re-running the |
| full sweep (≈ the entire measurement campaign). |
|
|
| > ⚡ **Per-token energy** varies by up to **8.3×** across languages · 🔋 **total energy** per |
| > request set varies by up to **179×** (English 17.6 kJ → Pashto 3,147 kJ) · 📉 the most |
| > energy-expensive languages are also the **least accurate**. |
|
|
| All energy is measured with the [ML.ENERGY Benchmark](https://arxiv.org/abs/2505.06371) (vLLM |
| serving + the [Zeus](https://ml.energy/zeus) library) and reported as **steady-state** |
| per-token energy. Hardware: **NVIDIA L40S** (main / cross-model / cross-task) and |
| **RTX 6000 Pro Blackwell** (batch-size sweep). |
|
|
| ## 📦 Contents |
|
|
| ### 📊 `results/` — energy measurements |
|
|
| | File | Description | |
| |------|-------------| |
| | `belebele_canonical_Qwen3-8B_0shot.csv` | ⭐ **Main result.** Qwen3-8B, Belebele, **122 languages**: energy/token, output tokens, total energy, accuracy (strict & lenient), resource level. | |
| | `belebele_canonical_Qwen3-14B.csv`, `…Qwen3-8B.csv`, `…Llama-3.1-8B.csv`, `…Llama-3.1-70B.csv` | Per-language energy for additional models (cross-model study). | |
| | `cross_model_12lang_table.csv` | Cross-model per-token energy comparison. | |
| | `crosstask_seqs256_summary.csv` | **Cross-task** (Belebele / GSM8K / LM-Arena) per-token energy, 8-language subset. | |
| | `belebele_seqsweep_l40s_0shot.csv` | **Batch-size sweep** on L40S (`max_num_seqs` ∈ {16,32,64,128,256,512}), 8 languages. | |
| | `seqsweep_belebele_v1_qwen8b_RTX6000_all8.csv` | Batch-size sweep on RTX 6000 Pro Blackwell, 8 languages. | |
|
|
| Key columns: `steady_energy_per_token_J` (J/token), `total_output_tokens`, `whole_gpu_energy_J`, |
| `accuracy_strict` / `accuracy_lenient`, `resource_level` (NLLB high/low). |
|
|
| ### 📝 `prompts/` — the data used |
|
|
| | Path | Description | |
| |------|-------------| |
| | `belebele_instructions_all_languages.json` | Per-language **zero-shot CoT instructions + primers** for all 122 Belebele languages (NLLB-200 translated, back-translation QC, hand-curated where needed). | |
| | `belebele_instruction_manual.csv` | Human-readable table of the per-language instruction/primer translations + curation notes. | |
| | `gsm8k/gsm8k_<lang>_scored.jsonl` | **Machine-translated GSM8K** math prompts with back-translation and quality scores (`bertscore_f1`, `comet`). | |
| | `lmarena/<lang>_scored.jsonl` | **Machine-translated LM-Arena** open-chat prompts with back-translation and quality scores. | |
|
|
| Each translated prompt record: `id`, `prompt_en` (source), `prompt_tgt` (translation), |
| `back_en` (back-translation), `bertscore_f1`, `comet`. Translations passing the quality bar |
| (BERTScore ≥ 0.85, COMET ≥ 0.75) were retained. |
|
|
| ## 🌐 The 8-language subset |
|
|
| Four high-resource — **English** (Latin), **Chinese** (Hans), **Russian** (Cyrillic), |
| **French** (Latin) — and four low-resource — **Southern Pashto** (Arabic), **Tigrinya** |
| (Ethiopic), **Shan** (Myanmar), **Tibetan** (Tibetan) — used for the cross-model, cross-GPU, |
| and cross-task experiments. |
|
|
| ## 🧪 Quick start |
|
|
| ```python |
| import pandas as pd |
| df = pd.read_csv("hf://datasets/MichiganNLP/language-energy-divide/results/belebele_canonical_Qwen3-8B_0shot.csv") |
| df.sort_values("steady_energy_per_token_J").head() |
| ``` |
|
|
| ## 📚 Citation & attribution |
|
|
| ```bibtex |
| @article{language-energy-divide, |
| title = {The Language--Energy Divide: Measuring Energy Costs of Multilingual LLM Inference}, |
| author = {Deng, Naihao and Shen, Alissa and Feng, Yiming and Nwatu, Joan and |
| Chung, Jae-Won and Chowdhury, Mosharaf and Chen, Yulong and Mihalcea, Rada}, |
| year = {2026} |
| } |
| ``` |
|
|
| Built on: **Belebele** ([Bandarkar et al., 2024](https://aclanthology.org/2024.acl-long.44/), |
| CC-BY-SA-4.0), **GSM8K** ([Cobbe et al., 2021](https://arxiv.org/abs/2110.14168), MIT), |
| **LM-Arena / Chatbot Arena** ([Chiang et al., 2024](https://arxiv.org/abs/2403.04132)), |
| translation via **NLLB-200** ([NLLB Team, 2022](https://arxiv.org/abs/2207.04672)), energy |
| measurement via the **ML.ENERGY Benchmark** ([Chung et al., 2025](https://arxiv.org/abs/2505.06371)). |
| Translated prompts derive from these sources; please also cite them and respect their licenses. |
|
|
| ## ⚠️ Note on translations |
|
|
| GSM8K and LM-Arena prompts, and the Belebele per-language instructions/primers, are |
| machine-translated (with QC), not natively authored. They may not fully reflect real-world |
| usage of each language. Measured disparities are best read as a **lower bound** on what |
| speakers of underserved languages face with natively authored inputs. |
|
|
