File size: 4,057 Bytes
a291bf0 ccc62ae a291bf0 ea96f9b a291bf0 616fc5a a291bf0 616fc5a a291bf0 616fc5a 22b0ac8 616fc5a 22b0ac8 616fc5a |
1 2 3 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 |
<div align="center">
<picture>
<img alt="specify theme context for images" src="https://raw.githubusercontent.com/01-ai/Yi/main/assets/img/Yi_logo_icon_light.svg" width="150px">
</picture>
</div>
<p align="center">
<a href="https://huggingface.co/01-ai">🤗 HuggingFace</a> •
<a href="https://www.modelscope.cn/organization/01ai/">🤖 ModelScope</a> •
<a href="https://wisemodel.cn/organization/01.AI">✡️ WiseModel</a>
<br/>
<a href="https://github.com/01-ai/Yi-1.5">🐙 GitHub</a> •
<a href="https://discord.gg/hYUwWddeAu">👾 Discord</a> •
<a href="https://twitter.com/01ai_yi">🐤 Twitter</a> •
<a href="https://github.com/01-ai/Yi/issues/43#issuecomment-1827285245">💬 WeChat</a>
<br/>
<a href="https://arxiv.org/abs/2403.04652">📝 Paper</a> •
<a href="https://github.com/01-ai/Yi/tree/main?tab=readme-ov-file#faq">🙌 FAQ</a> •
<a href="https://github.com/01-ai/Yi/tree/main?tab=readme-ov-file#learning-hub">📗 Learning Hub</a>
</p>
# Intro
Yi-1.5 is an upgraded version of Yi. It is continuously pre-trained on Yi with a high-quality corpus of 500B tokens and fine-tuned on 3M diverse fine-tuning samples.
Compared with Yi, Yi-1.5 delivers stronger performance in coding, math, reasoning, and instruction-following capability, while still maintaining excellent capabilities in language understanding, commonsense reasoning, and reading comprehension.
<div align="center">
Model | Context Length | Pre-trained Tokens
| :------------: | :------------: | :------------: |
| Yi-1.5 | 4K | 3.6T
</div>
# Models
- Chat models
<div align="center">
Model | Download
| :------------: | :------------: |
Yi-1.5-34B-Chat | • 🤗 Hugging Face
Yi-1.5-9B-Chat | • 🤗 Hugging Face
Yi-1.5-6B-Chat | • 🤗 Hugging Face
</div>
- Base models
<div align="center">
Model | Download
| :------------: | :------------: |
Yi-1.5-34B | • 🤗 Hugging Face
Yi-1.5-9B | • 🤗 Hugging Face
Yi-1.5-6B | • 🤗 Hugging Face
</div>
# Benchmarks
- Chat models
Waiting for benchmark results.
- Base models
- Yi-1.5-34B excels beyond or is on par with some larger models in overall performance.
| Model | MMLU | CMMLU | BBH | AGIEval | HumanEva(+) | MBPP(+) | GSM8k | MATH |
| --------------- | ---- | ----- | ---- | ------- | ----------- | ---------- | ----- | ---- |
| Mistral 8\*22B | 77.8 | 60.4 | 61.9 | 58.6 | 45.1(34.1) | 71.2(-) | 81.7 | 41.8 |
| DeepSeek-V2 | 78.5 | 84.0 | 78.9 | \- | 48.8(-) | 66.6(-) | 79.2 | 43.6 |
| Qwen1.5-32B | 73.7 | 83 | 66.8 | 72.6 | 37.8(35.4) | 49.4(44.4) | 79.5 | 37.7 |
| Qwen1.5-72B | 77.5 | 84.2 | 65.5 | 71.7 | 41.5(39.0) | 53.4(43.6) | 81.4 | 41.8 |
| Llama3_70B_base | 78.7 | 68.9 | 65.0 | 52.8 | 38.4(34.8) | 69.7(52.9) | 82.4 | 42.4 |
| Yi 1.5-34B | 77.1 | 84.8 | 76.4 | 71.1 | 46.3(40.2) | 65.5(55.4) | 82.7 | 41.0 |
- Yi-1.5-9B is a strong performer among similarly sized open-source models.
| Model | MMLU | CMMLU | BBH | AGIEval | HumanEval(+) | MBPP(+) | GSM8k | Math |
| -------------- | ---- | ----- | ---- | ------- | ------------ | ---------- | ----- | ----- |
| Gemma-7B | 64.3 | 48.4 | 41.1 | 46.0 | 33.5(28.0) | 45.8(32.8) | 55.7 | 24.8 |
| Qwen1.5-7B | 61.0 | 73.4 | 33.4 | 61.6 | 36.0(31.1) | 46.1(37.6) | 70.1 | 20.3 |
| Mistral-7B | 62.5 | 44.6 | 45.0 | 42.4 | 29.3(22.6) | 50.2(32.1) | 47.5 | 15.5 |
| Mistral 8\*7B | 70.6 | 53.0 | 52.4 | 49.5 | 40.2(31.1) | 60.7(31.1) | 65.7 | 28.4 |
| Llama3-8B_Base | 66.6 | 50.9 | 47.9 | 44.7 | 34.7(31.7) | 48.0(44.9) | 54.7 | 21.16 |
| Yi 1.5-6B | 63.5 | 70.8 | 45.7 | 56.0 | 36.5(28.7) | 56.8(46.9) | 62.2 | 28.42 |
| Yi 1.5-9B | 69.5 | 74.8 | 50.9 | 62.7 | 41.4(34.1) | 61.1(53.6) | 73.7 | 32.6 |
# Quick Start
For getting up and running with Yi-1.5 models quickly, see [README](https://github.com/01-ai/Yi-1.5).
|