Update README.md

README.md

@@ -51,11 +51,34 @@ tags:
 - GPT
 - Gemini
 - Phi
+- Reasoning
+- o3-mini
 pretty_name: timeseriesexam1
 size_categories:
 - n<1K
 ---
 
+## Changelog
+### [v1.1] - 2025-03-12
+
+**Enhancements:**
+- Adjusted generation hyperparameters and templates to eliminate scenarios that might lead to ambiguous or incorrect responses.
+- Improved data formatting for consistency.
+- Updated time-series sample length to 1024 to capture more diverse and complex features.
+
+**Updated Model Evaluations:**
+- The following table shows the updated evaluation on models (tokenization method):
+  | Model    | Tokenization Method | Accuracy |
+  |----------|---------------------|----------|
+  | gpt-4o   | image               | 75.2%    |
+  | gpt-4o   | plain_text          | 51.7%    |
+  | 4o-mini  | plain_text          | 46.6%    |
+  | o3-mini  | plain_text          | 59.0%    |
+
+**Additional Information:**
+- **Note:** The previous version ([v1.0](https://huggingface.co/datasets/AutonLab/TimeSeriesExam1/resolve/9f23771ca10d66607ee7abca0c5dcbae57349ac2/qa_dataset.json)) is still available for reference.
+- **Research code for exam generation via templates is available on [GitHub](https://github.com/moment-timeseries-foundation-model/TimeSeriesExam/tree/exam_generation).**
+
 # Dataset Card for TimeSeriesExam-1
 
 This dataset provides Question-Answer (QA) pairs for the paper [TimeSeriesExam: A Time Series Understanding Exam](https://arxiv.org/pdf/2410.14752). Example inference code can be found [here](https://github.com/moment-timeseries-foundation-model/TimeSeriesExam).