Zhitong_SDU_WT_LLM

LoRA fine-tuned amazon/chronos-2 on cell-level 4G/5G multi-service traffic data (8 metrics per cell per network) from the Shandong University WT (无线/网通) dataset.

Usage

from chronos import Chronos2Pipeline

pipeline = Chronos2Pipeline.from_pretrained("Zhitong_SDU_WT_LLM",
                                            device_map="cuda")
forecast = pipeline.predict(context, prediction_length=24)

Training

base model: amazon/chronos-2
finetune mode: LoRA (r=16, " alpha=32)
context length: 336
prediction length: 24
training series: 800
num_steps: 30, batch_size: 8, lr: 1e-05

Downloads last month: 16

Safetensors

Model size

0.1B params

Tensor type

F32

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for ctzhang/Zhitong

Base model

amazon/chronos-2

Adapter

(26)

this model