# KimMeen/Time-LLM

[ICLR 2024] Official implementation of " 🦙 Time-LLM: Time Series Forecasting by Reprogramming Large Language Models"

Repository: https://github.com/KimMeen/Time-LLM
Canonical: https://ross.abutalabs.com/products/time-llm
Homepage: https://arxiv.org/abs/2310.01728
Language: Python
License: Apache-2.0
License Family: permissive
Topics: cross-modal-learning, cross-modality, deep-learning, language-model, large-language-models, machine-learning, multimodal-deep-learning, multimodal-time-series, prompt-tuning, time-series, time-series-analysis, time-series-forecast, time-series-forecasting
Last push: 2025-10-15T11:40:18+00:00

## Health v2 (maintenance only)
Score: 47/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 47, release rhythm 35, longevity 68
- inputs: {"age_days": 957, "days_push": 322, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2685, forks 466 (observed 2026-08-28T04:07:10.206464+00:00)

## What it is
Time-LLM is the official PyTorch implementation of an ICLR 2024 paper that reprograms frozen large language models (Llama, GPT-2, BERT) for general time series forecasting. It aligns time series data with text prototypes and uses Prompt-as-Prefix to leverage LLM reasoning without modifying the backbone model.

## Use cases
- forecast time series with large language models
- reprogram an LLM for time series prediction
- reproduce the Time-LLM ICLR 2024 paper results
- use Llama or GPT-2 as a time series forecasting backbone
- apply prompt-based forecasting to energy or weather data
- compare LLM-based forecasting baselines

## When to choose
- you want to leverage pretrained LLMs for forecasting without training the backbone
- you need a research-grade implementation of the Time-LLM method
- you want to experiment with cross-modal reprogramming and prompt-as-prefix techniques

## When to avoid
- you need lightweight, low-latency production forecasting without GPU-heavy LLM inference
- you want simple classical or dedicated time series models
- you need a maintained library API rather than research code

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, llm-inference
- domain: time-series, large-language-models, machine-learning, deep-learning
- platform: python
- tags: time-series-forecasting, llm-reprogramming, prompt-as-prefix, iclr-2024, multimodal, research-code

## Member repositories
- KimMeen/Time-LLM (main) score 47

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:10.206464+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:17:49.055326+00:00, confidence not recorded.
  - readme: https://github.com/KimMeen/Time-LLM (fetched 2026-08-28T04:07:10.206464+00:00, sha ab00d1958eec)
  - homepage: https://arxiv.org/abs/2310.01728 (fetched 2026-08-29T09:59:58.181156+00:00, sha 3b1a83d06abb)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T09:59:58.190060+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T09:59:58.193163+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T09:59:58.194783+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T09:59:58.191595+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.