https://github.com/KimMeen/Time-LLM

[ICLR 2024] Official implementation of " 🦙 Time-LLM: Time Series Forecasting by Reprogramming Large Language Models"
https://github.com/KimMeen/Time-LLM

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 synced: 3 days ago
JSON representation

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

• Host: GitHub
• URL: https://github.com/KimMeen/Time-LLM
• Owner: KimMeen
• License: apache-2.0
• Created: 2024-01-20T01:26:30.000Z (10 months ago)
• Default Branch: main
• Last Pushed: 2024-08-29T07:24:56.000Z (2 months ago)
• Last Synced: 2024-10-29T15:27:53.044Z (10 days ago)
• 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
• Language: Python
• Homepage: https://arxiv.org/abs/2310.01728
• Size: 1.06 MB
• Stars: 1,378
• Watchers: 16
• Forks: 240
• Open Issues: 27
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• StarryDivineSky - KimMeen/Time-LLM - LLM： 通过重新编程大型语言模型进行时间序列预测。Time-LLM 是一种重编程框架，用于重新用于LLMs一般时间序列预测，同时保持骨干语言模型不变。值得注意的是，我们发现时间序列分析（例如，预测）可以被看作是另一个“语言任务”，可以由现成LLM的。时间-LLM包括两个关键组成部分：（1）将输入时间序列重新编程为对用户LLM来说更自然的文本原型表示，以及（2）通过声明性提示（例如，领域专家知识和任务指令）增强输入上下文以指导LLM推理。（2024 年 3 月）：Time-LLM 已升级为通用框架，用于将各种语言模型重新用于时间序列预测。它现在默认支持 Llama-7B，并包括与另外两个较小的 PLM（GPT-2 和 BERT）的兼容性。只需调整 --llm_model 和 --llm_dim 即可切换主干。Time-LLM 已被包含在 NeuralForecast 中。时间LLM 已被 XiMou Optimization Technology Co.， Ltd. （XMO） 用于太阳能、风能和天气预报。 (时间序列 / 网络服务_其他)