Add model card for TIC-FM

README.md

@@ -0,0 +1,57 @@
+---
+pipeline_tag: other
+---
+
+# TIC-FM
+
+This repository contains the pre-trained weights and configuration for **TIC-FM** (Time Series In-Context Foundation Model), presented in the paper [Rethinking Zero-Shot Time Series Classification: From Task-specific Classifiers to In-Context Inference](https://huggingface.co/papers/2602.00620).
+
+- **Repository:** [GitHub - fangjuntao/TIC-FM](https://github.com/fangjuntao/TIC-FM)
+- **Paper:** [Hugging Face Papers](https://huggingface.co/papers/2602.00620)
+
+## Model Description
+
+TIC-FM is an in-context learning framework that treats the labeled training set as context and predicts labels for test instances in a single forward pass without requiring parameter updates or fine-tuning. It pairs a time series encoder and a lightweight projection adapter with a split-masked latent memory Transformer.
+
+## Quick Start & Evaluation
+
+To set up the environment and run evaluations with this checkpoint, clone the official repository and set up the conda environment:
+
+```bash
+git clone https://github.com/fangjuntao/TIC-FM.git
+cd TIC-FM
+conda create -n TICFS python=3.9 -y
+conda activate TICFS
+pip install torch numpy
+```
+
+To run the evaluation script on the UCR datasets using the downloaded checkpoint:
+
+```bash
+python scripts/eval_TSEncoder_orion_icl_classifier_ucr_full.py \
+  --suite ucr \
+  --mode classifier_v2 \
+  --ucr_path /path/to/UCRdata/ \
+  --full_ckpt checkpoints/TSEncoder_orion_icl_full.pt \
+  --model_hparams_json checkpoints/TSEncoder_orion_icl_model_hparams.json
+```
+
+To evaluate a single dataset:
+
+```bash
+python scripts/eval_TSEncoder_orion_icl_classifier_ucr_full.py \
+  --suite ucr \
+  --dataset ECG200 \
+  --mode direct
+```
+
+## Citation
+
+```bibtex
+@article{fang2026rethinking,
+  title={Rethinking Zero-Shot Time Series Classification: From Task-specific Classifiers to In-Context Inference},
+  author={Fang, Juntao and others},
+  journal={arXiv preprint arXiv:2602.00620},
+  year={2026}
+}
+```