Datasets:
Tasks:
Question Answering
Modalities:
Text
Formats:
csv
Languages:
Chinese
Size:
10K - 100K
ArXiv:
License:
docs: update README for v1.1 - version history, leaderboard, About iKala section
Browse files
README.md
CHANGED
|
@@ -549,11 +549,12 @@ configs:
|
|
| 549 |
# TMMLU+ : Large scale traditional chinese massive multitask language understanding
|
| 550 |
|
| 551 |
<p align="center">
|
| 552 |
-
<img src="https://huggingface.co/datasets/ikala/tmmluplus/resolve/main/cover.png" alt="A close-up image of a neat paper note with a white background. The text 'TMMLU+' is written horizontally across the center of the note in bold, black.
|
| 553 |
</p>
|
| 554 |
-
We present TMMLU+, a traditional Chinese massive multitask language understanding dataset. TMMLU+ is a multiple-choice question-answering dataset featuring 66 subjects, ranging from elementary to professional level.
|
| 555 |
|
| 556 |
-
|
|
|
|
|
|
|
| 557 |
|
| 558 |
|
| 559 |
```python
|
|
@@ -598,62 +599,75 @@ Statistic on all four categories : STEM, Social Science, Humanities, Other
|
|
| 598 |
| **Total** | 20118 | 330 | 2242 |
|
| 599 |
|
| 600 |
|
| 601 |
-
##
|
| 602 |
-
|
| 603 |
-
|
|
| 604 |
-
|---
|
| 605 |
-
|
|
| 606 |
-
|[
|
| 607 |
-
|
| 608 |
-
|
| 609 |
-
|
| 610 |
-
|
| 611 |
-
|
| 612 |
-
|
| 613 |
-
|
| 614 |
-
|
|
| 615 |
-
|
|
| 616 |
-
|
|
| 617 |
-
|
|
| 618 |
-
|
|
| 619 |
-
|
|
| 620 |
-
|
|
| 621 |
-
|
| 622 |
-
|
| 623 |
-
|
| 624 |
-
|
| 625 |
-
|
| 626 |
-
|
| 627 |
-
|
| 628 |
-
|
| 629 |
-
|
| 630 |
-
|
| 631 |
-
|
| 632 |
-
|
| 633 |
-
|
| 634 |
-
|
| 635 |
-
|
| 636 |
-
|
| 637 |
-
|
|
| 638 |
-
|
|
| 639 |
-
|
|
| 640 |
-
|
|
| 641 |
-
|
|
| 642 |
-
|
|
| 643 |
-
|
|
| 644 |
-
|
|
| 645 |
-
|
|
| 646 |
-
|
|
| 647 |
-
|
|
| 648 |
-
|
|
| 649 |
-
|
|
| 650 |
-
|
|
| 651 |
-
|
|
| 652 |
-
|
| 653 |
-
|
| 654 |
-
|
| 655 |
-
|
| 656 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 657 |
|
| 658 |
```
|
| 659 |
@article{ikala2024improved,
|
|
@@ -664,3 +678,18 @@ Results via [ievals](https://github.com/iKala/ievals) ( settings : 0-shot direct
|
|
| 664 |
}
|
| 665 |
```
|
| 666 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 549 |
# TMMLU+ : Large scale traditional chinese massive multitask language understanding
|
| 550 |
|
| 551 |
<p align="center">
|
| 552 |
+
<img src="https://huggingface.co/datasets/ikala/tmmluplus/resolve/main/cover.png" alt="A close-up image of a neat paper note with a white background. The text 'TMMLU+' is written horizontally across the center of the note in bold, black." style="max-width: 400" width=400 />
|
| 553 |
</p>
|
|
|
|
| 554 |
|
| 555 |
+
iKala presents **TMMLU+**, a large-scale benchmark for evaluating LLM capabilities in **Traditional Chinese, with content primarily reflecting Taiwan's linguistic, educational, and professional contexts**. It covers **66 subjects**, from elementary to professional domains, and is approximately **six times larger** than [TMMLU](https://github.com/mtkresearch/TCEval) with broader, more balanced coverage.
|
| 556 |
+
|
| 557 |
+
**TMMLU+ v1.1** improves benchmark quality through systematic review: outdated legal and regulatory content was updated, incomplete or invalid questions were corrected or removed, and ambiguous items were reviewed by domain experts. Questions without a single defensible answer were excluded.
|
| 558 |
|
| 559 |
|
| 560 |
```python
|
|
|
|
| 599 |
| **Total** | 20118 | 330 | 2242 |
|
| 600 |
|
| 601 |
|
| 602 |
+
## Dataset Versions
|
| 603 |
+
|
| 604 |
+
| Version | Tag | Description |
|
| 605 |
+
|---|---|---|
|
| 606 |
+
| v1.0 | [`v1.0`](https://huggingface.co/datasets/ikala/tmmluplus/tree/v1.0) | Original release, unmodified |
|
| 607 |
+
| v1.1 | [`v1.1`](https://huggingface.co/datasets/ikala/tmmluplus/tree/v1.1) | Verified and corrected release (see below) — **this page reflects v1.1** |
|
| 608 |
+
|
| 609 |
+
v1.1 was produced by individually re-verifying every question flagged as potentially problematic by four rule-based scans (6,116 candidate questions out of 22,742), followed by a human review pass on the 691 questions where the verification suggested a change.
|
| 610 |
+
|
| 611 |
+
- **197** questions had their answer corrected — 196 with the answer key changed (`change_answer`), plus 1 whose stem was rewritten to fix a corrupted/duplicated fragment (all 4 options and the answer were kept)
|
| 612 |
+
- **539** questions were removed entirely (255 `change_content` — defective stem/options; 233 `expert_review` — unresolved ambiguity even after review; 51 `remove` — flagged in an earlier review pass)
|
| 613 |
+
- **22,203** questions remain in v1.1 (out of 22,742 in v1.0)
|
| 614 |
+
|
| 615 |
+
| Category | Test | Dev | Validation |
|
| 616 |
+
|-----------------------------------|-------|------|------------|
|
| 617 |
+
| STEM | 3369 | 69 | 382 |
|
| 618 |
+
| Social Sciences | 5864 | 89 | 657 |
|
| 619 |
+
| Humanities | 1723 | 34 | 189 |
|
| 620 |
+
| Other (Business, Health, Misc.) | 8724 | 129 | 974 |
|
| 621 |
+
| **Total** | **19680** | **321** | **2202** |
|
| 622 |
+
|
| 623 |
+
*(For v1.0 figures, see the "Statistic on all four categories" table above, or load `revision="v1.0"`.)*
|
| 624 |
+
|
| 625 |
+
To load a specific version:
|
| 626 |
+
|
| 627 |
+
```python
|
| 628 |
+
from datasets import load_dataset
|
| 629 |
+
load_dataset('ikala/tmmluplus', 'accounting', revision='v1.0') # original
|
| 630 |
+
load_dataset('ikala/tmmluplus', 'accounting', revision='v1.1') # verified/corrected
|
| 631 |
+
```
|
| 632 |
+
|
| 633 |
+
|
| 634 |
+
## Leaderboard
|
| 635 |
+
|
| 636 |
+
Scores below are computed against the v1.1 question set for the 21 models that have completed evaluation on all 66 subjects. Category and Total scores are the average of each category's own accuracy (STEM/Social Science/Humanities/Other weighted equally, matching the [`ievals`](https://github.com/iKala/ievals) methodology), not a per-question average. For v1.0 scores, load `revision="v1.0"` and re-run evaluation against that question set.
|
| 637 |
+
|
| 638 |
+
| Model | STEM | Social Science | Humanities | Other | Total |
|
| 639 |
+
|---|---|---|---|---|---|
|
| 640 |
+
| claude-opus-5 | **98.71** | **96.09** | **94.60** | **95.25** | **96.16** |
|
| 641 |
+
| gemini-3.7-flash (reasoning) | 93.37 | 92.09 | 88.91 | 87.64 | 90.50 |
|
| 642 |
+
| gpt-5.6-sol | 93.28 | 91.54 | 82.30 | 86.39 | 88.38 |
|
| 643 |
+
| claude-fable-5 | 90.61 | 83.41 | 90.02 | 84.22 | 87.06 |
|
| 644 |
+
| deepseek/deepseek-v4-pro-0813 | 92.89 | 89.24 | 79.63 | 83.92 | 86.42 |
|
| 645 |
+
| claude-sonnet-5 | 89.21 | 85.79 | 81.72 | 81.90 | 84.66 |
|
| 646 |
+
| gemini-3.1-pro-preview (reasoning) | 87.14 | 83.22 | 84.91 | 81.76 | 84.26 |
|
| 647 |
+
| deepseek/deepseek-v4-flash | 89.42 | 86.72 | 77.42 | 80.42 | 83.50 |
|
| 648 |
+
| tencent/hy3 | 89.81 | 88.25 | 74.75 | 80.50 | 83.33 |
|
| 649 |
+
| gpt-5.6-terra | 90.22 | 86.95 | 72.49 | 80.07 | 82.43 |
|
| 650 |
+
| moonshotai/kimi-k3 | 86.12 | 82.55 | 77.31 | 78.03 | 81.00 |
|
| 651 |
+
| qwen/qwen3.7-max | 85.07 | 84.02 | 76.03 | 77.25 | 80.59 |
|
| 652 |
+
| qwen/qwen3.8-27b | 93.10 | 83.99 | 63.49 | 79.61 | 80.05 |
|
| 653 |
+
| gpt-5.6-luna | 87.59 | 83.65 | 70.81 | 76.48 | 79.63 |
|
| 654 |
+
| z-ai/glm-5.2 | 85.31 | 83.34 | 70.28 | 75.42 | 78.59 |
|
| 655 |
+
| xiaomi/mimo-v2.5 | 87.44 | 83.53 | 67.32 | 74.19 | 78.12 |
|
| 656 |
+
| minimax/minimax-m3 | 86.21 | 79.57 | 65.76 | 73.91 | 76.36 |
|
| 657 |
+
| x-ai/grok-4.3 | 81.32 | 81.65 | 67.67 | 74.79 | 76.36 |
|
| 658 |
+
| claude-haiku-4-5 | 82.88 | 74.54 | 59.37 | 70.14 | 71.73 |
|
| 659 |
+
| google/gemma-4-31b-it | 81.26 | 73.29 | 58.15 | 66.66 | 69.84 |
|
| 660 |
+
| nvidia/nemotron-3-ultra-550b-a55b | 72.77 | 73.99 | 59.20 | 63.96 | 67.48 |
|
| 661 |
+
|
| 662 |
+
*Note: all models were called with default API parameters (no reasoning effort or thinking mode explicitly configured). Models marked (reasoning) reported a separate reasoning/thinking token count from the API under this default before producing their final answer; other models answered directly without an exposed reasoning trace.*
|
| 663 |
+
|
| 664 |
+
|
| 665 |
+
## Licensing Information
|
| 666 |
+
|
| 667 |
+
This dataset is released under the [MIT License](https://opensource.org/licenses/MIT). You are free to use, copy, modify, and redistribute it, including for commercial purposes, provided the original copyright notice is retained.
|
| 668 |
+
|
| 669 |
+
|
| 670 |
+
## Citation
|
| 671 |
|
| 672 |
```
|
| 673 |
@article{ikala2024improved,
|
|
|
|
| 678 |
}
|
| 679 |
```
|
| 680 |
|
| 681 |
+
|
| 682 |
+
<img src="https://huggingface.co/datasets/ikala/tmmluplus/resolve/main/ikala_logo.png" alt="iKala logo" width="40" style="border-radius: 6px;" />
|
| 683 |
+
|
| 684 |
+
## About iKala
|
| 685 |
+
|
| 686 |
+
iKala helps enterprises make better, faster decisions by embedding AI and data at the core of their business. We support AI transformation by helping organizations move from data to decisions, delivering full AI solutions that combine their first-party data with iKala's intelligence built on billions of global social signals.
|
| 687 |
+
|
| 688 |
+
Headquartered in Taiwan with a global footprint, iKala serves over 1,000 enterprises and 50,000 brands across more than 190 countries, including Fortune 500 companies.
|
| 689 |
+
|
| 690 |
+
<img src="https://huggingface.co/datasets/ikala/tmmluplus/resolve/main/ikala_logo.png" alt="iKala logo" width="20" style="vertical-align: middle; border-radius: 4px;" /> iKala Official Website: [ikala.ai](https://ikala.ai)
|
| 691 |
+
<br>
|
| 692 |
+
<img src="https://huggingface.co/datasets/ikala/tmmluplus/resolve/main/kolr_logo.png" alt="Kolr logo" width="20" style="vertical-align: middle; border-radius: 4px;" /> Kolr Official Website: [kolr.ai](https://kolr.ai)
|
| 693 |
+
<br>
|
| 694 |
+
<img src="https://huggingface.co/datasets/ikala/tmmluplus/resolve/main/kuroma_logo.png" alt="Kuroma logo" width="20" style="vertical-align: middle; border-radius: 4px;" /> Kuroma Official Website: [kuroma.ai](https://kuroma.ai)
|
| 695 |
+
|