Improve language tag
#1
by lbourdois - opened
README.md
CHANGED
|
@@ -1,111 +1,123 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
language:
|
| 4 |
-
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
-
|
| 9 |
-
|
| 10 |
-
-
|
| 11 |
-
-
|
| 12 |
-
-
|
| 13 |
-
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
#
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
)
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
#
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
##
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 111 |
```
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- zho
|
| 5 |
+
- eng
|
| 6 |
+
- fra
|
| 7 |
+
- spa
|
| 8 |
+
- por
|
| 9 |
+
- deu
|
| 10 |
+
- ita
|
| 11 |
+
- rus
|
| 12 |
+
- jpn
|
| 13 |
+
- kor
|
| 14 |
+
- vie
|
| 15 |
+
- tha
|
| 16 |
+
- ara
|
| 17 |
+
pipeline_tag: text-generation
|
| 18 |
+
library_name: transformers
|
| 19 |
+
base_model:
|
| 20 |
+
- Qwen/Qwen2.5-1.5B-Instruct
|
| 21 |
+
datasets:
|
| 22 |
+
- tokyotech-llm/lmsys-chat-1m-synth
|
| 23 |
+
- tokyotech-llm/swallow-magpie-ultra-v0.1
|
| 24 |
+
- tokyotech-llm/swallow-swallow-gemma-magpie-v0.1
|
| 25 |
+
---
|
| 26 |
+
# TinySwallow-1.5B-Instruct
|
| 27 |
+
|
| 28 |
+
🤗 [Models](https://huggingface.co/SakanaAI) | 📚 [Paper](https://arxiv.org/abs/2501.16937) | 📝 [Blog](https://sakana.ai/taid-jp/) | 🐦 [Twitter](https://twitter.com/SakanaAILabs)
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
**TinySwallow-1.5B-Instruct** is an instruction-tuned version of [TinySwallow-1.5B](https://huggingface.co/SakanaAI/TinySwallow-1.5B), created through *TAID (Temporally Adaptive Interpolated Distillation)*, our new knowledge distillation method.
|
| 32 |
+
We used [Qwen2.5-32B-Instruct](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct) as the teacher model and [Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct) as the student model.
|
| 33 |
+
The model has been further instruction-tuned to enhance its ability to follow instructions and engage in conversations in Japanese.
|
| 34 |
+
|
| 35 |
+
## Usage
|
| 36 |
+
Use the code below to get started with the model.
|
| 37 |
+
|
| 38 |
+
<details>
|
| 39 |
+
<summary> Click to expand </summary>
|
| 40 |
+
|
| 41 |
+
```python
|
| 42 |
+
import torch
|
| 43 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
| 44 |
+
|
| 45 |
+
|
| 46 |
+
# 1. load model
|
| 47 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 48 |
+
repo_id = "SakanaAI/TinySwallow-1.5B-Instruct"
|
| 49 |
+
model = AutoModelForCausalLM.from_pretrained(repo_id)
|
| 50 |
+
tokenizer = AutoTokenizer.from_pretrained(repo_id)
|
| 51 |
+
model.to(device)
|
| 52 |
+
|
| 53 |
+
# 2. prepare inputs
|
| 54 |
+
text = "知識蒸留について簡単に教えてください。"
|
| 55 |
+
messages = [{"role": "user", "content": text}]
|
| 56 |
+
input_ids = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt")
|
| 57 |
+
|
| 58 |
+
# 3. generate
|
| 59 |
+
output_ids = model.generate(
|
| 60 |
+
input_ids.to(device),
|
| 61 |
+
max_new_tokens=1024,
|
| 62 |
+
)
|
| 63 |
+
output_ids = output_ids[:, input_ids.shape[1] :]
|
| 64 |
+
generated_text = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0]
|
| 65 |
+
print(generated_text)
|
| 66 |
+
```
|
| 67 |
+
|
| 68 |
+
</details>
|
| 69 |
+
|
| 70 |
+
## Model Details
|
| 71 |
+
|
| 72 |
+
- **Model type:** Autoregressive Language Model
|
| 73 |
+
- **Language(s):** Japanese
|
| 74 |
+
- **Paper:** https://arxiv.org/abs/2501.16937
|
| 75 |
+
- **Blog:** https://sakana.ai/taid-jp/
|
| 76 |
+
- **Training Datasets:**
|
| 77 |
+
- [Gemma-2-LMSYS-Chat-1M-Synth](https://huggingface.co/datasets/tokyotech-llm/lmsys-chat-1m-synth/blob/main/README_gemma.md)
|
| 78 |
+
- [tokyotech-llm/swallow-magpie-ultra-v0.1](https://huggingface.co/datasets/tokyotech-llm/swallow-magpie-ultra-v0.1)
|
| 79 |
+
- [tokyotech-llm/swallow-gemma-magpie-v0.1](https://huggingface.co/datasets/tokyotech-llm/swallow-gemma-magpie-v0.1)
|
| 80 |
+
|
| 81 |
+
## Uses
|
| 82 |
+
This model is provided for research and development purposes only and should be considered as an experimental prototype.
|
| 83 |
+
It is not intended for commercial use or deployment in mission-critical environments.
|
| 84 |
+
Use of this model is at the user's own risk, and its performance and outcomes are not guaranteed.
|
| 85 |
+
Sakana AI shall not be liable for any direct, indirect, special, incidental, or consequential damages, or any loss arising from the use of this model, regardless of the results obtained.
|
| 86 |
+
Users must fully understand the risks associated with the use of this model and use it at their own discretion.
|
| 87 |
+
|
| 88 |
+
|
| 89 |
+
## Acknowledgement
|
| 90 |
+
|
| 91 |
+
We would like to thank the developers of the source models for their contributions and for making their work available.
|
| 92 |
+
|
| 93 |
+
## Authors
|
| 94 |
+
|
| 95 |
+
* [Sakana AI](https://sakana.ai/)
|
| 96 |
+
* [Makoto Shing](https://huggingface.co/mkshing)
|
| 97 |
+
* [Taishi Nakamura](https://x.com/Setuna7777_2)
|
| 98 |
+
* [Kou Misaki](https://huggingface.co/takkyu2)
|
| 99 |
+
* [Takuya Akiba](https://huggingface.co/iwiwi)
|
| 100 |
+
* [Swallow Team](https://swallow-llm.github.io/index.en.html)
|
| 101 |
+
* [Naoki Okazaki](https://www.chokkan.org/index.ja.html)
|
| 102 |
+
* [Youmi Ma](https://www.nlp.c.titech.ac.jp/member/youmi.en.html)
|
| 103 |
+
* [Kakeru Hattori](https://aya-se.vercel.app/)
|
| 104 |
+
* [Kazuki Fujii](https://x.com/okoge_kaz)
|
| 105 |
+
* [Sakae Mizuki](https://s-mizuki-nlp.github.io/)
|
| 106 |
+
|
| 107 |
+
## License
|
| 108 |
+
This model is derived from Qwen ([Apache 2.0](https://www.apache.org/licenses/LICENSE-2.0)) and trained on Gemma data ([Gemma Terms](https://ai.google.dev/gemma/terms), [Prohibited Use](https://ai.google.dev/gemma/prohibited_use_policy)). Use (including commercial) is permitted if you comply with both licenses/policies above.
|
| 109 |
+
|
| 110 |
+
|
| 111 |
+
## Citation
|
| 112 |
+
|
| 113 |
+
```bibtex
|
| 114 |
+
@misc{sakana2025taid,
|
| 115 |
+
title = {TAID: Temporally Adaptive Interpolated Distillation for Efficient Knowledge Transfer in Language Models},
|
| 116 |
+
author. = {Makoto Shing and Kou Misaki and Han Bao and Sho Yokoi and Takuya Akiba},
|
| 117 |
+
year = {2025},
|
| 118 |
+
eprint = {2501.16937},
|
| 119 |
+
archivePrefix = {arXiv},
|
| 120 |
+
primaryClass = {cs.LG},
|
| 121 |
+
url = {https://arxiv.org/abs/2501.16937}
|
| 122 |
+
}
|
| 123 |
```
|