license: apache-2.0
language:
- th
library_name: transformers
pipeline_tag: text-generation
tags:
- pretrained
Typhoon-7B: Thai Large Language Model
Typhoon-7B is a pretrained Thai 🇹🇭 large language model with 7 billion parameters, and it is based on Mistral-7B.
Typhoon-7B outperforms all open-source Thai language models at the time of writing as evaluated on Thai examination benchmarks, and its instruction-tuned variant achieves the best results in instruction-following tasks. Also, its performance in Thai is on par with GPT-3.5 while being 2.62 times more efficient in tokenizing Thai text.
For full details of this model, please read our paper and release blog post.
Model Description
- Model type: A 7B pretrained decoder-only model
- Requirement: transformers 4.34.0 or newer.
- Primary Language(s): Thai 🇹🇭 and English 🇬🇧
- License: Apache-2.0 (Commercial)
Performance on Thai Benchmark
| Model | ONET | IC | TGAT | TPAT-1 | A-Level |
|---|---|---|---|---|---|
| Typhoon-7B | 0.379 | 0.393 | 0.700 | 0.414 | 0.324 |
| SeaLLM-7B | 0.342 | 0.256 | 0.589 | 0.336 | 0.305 |
| OpenThaiGPT-beta-7B | 0.180 | 0.278 | 0.411 | 0.319 | 0.243 |
| WangChanGLM | 0.192 | 0.271 | 0.167 | 0.172 | 0.175 |
| SEA-LION-7B | 0.179 | 0.290 | 0.244 | 0.198 | 0.175 |
| Avg. Human | 0.318 | - | 0.472 | 0.406 | - |
Intended Uses & Limitations
This model is a pretrained base model. Thus, it may not be able to follow human instructions without using one/few-shot learning or instruction fine-tuning. The model does not have any moderation mechanisms, and may generate harmful or inappropriate responses.
SCB10X AI Team
- Kunat Pipatanakul, Phatrasek Jirabovonvisut, Potsawee Manakul, Sittipong Sripaisarnmongkol, Ruangsak Patomwong, Pathomporn Chokchainant, Kasima Tharnpipitchai
- Corresponding Author: kasima@scb10x.com
- If you find Typhoon-7B useful for your work, please cite it using: