Generate text with distributed **Llama 2 (70B)**, **Stable Beluga 2**, **Falcon**, **Guanaco-65B** or **BLOOM-176B** and fine‑tune them for your own tasks — right from your desktop computer or Google Colab:
Generate text with distributed **Llama 2** (70B), **Falcon** (40B+), **BLOOM** (176B) (or their derivatives), and fine‑tune them for your own tasks — right from your desktop computer or Google Colab:
```python
```python
from transformers import AutoTokenizer
from transformers import AutoTokenizer
from petals import AutoDistributedModelForCausalLM
from petals import AutoDistributedModelForCausalLM
# Choose any model available at https://health.petals.dev
# Choose any model available at https://health.petals.dev
model_name = "petals-team/StableBeluga2"
model_name = "petals-team/StableBeluga2" # This one is fine-tuned Llama 2 (70B)
# Connect to a distributed network hosting model layers
# Connect to a distributed network hosting model layers
@ -31,9 +31,9 @@ print(tokenizer.decode(outputs[0])) # A cat sat on a mat...
🚀 <b><ahref="https://colab.research.google.com/drive/1uCphNY7gfAUkdDrTx21dZZwCOUDCMPw8?usp=sharing">Try now in Colab</a></b>
🚀 <b><ahref="https://colab.research.google.com/drive/1uCphNY7gfAUkdDrTx21dZZwCOUDCMPw8?usp=sharing">Try now in Colab</a></b>
</p>
</p>
🦙 **Want to run Llama 2?** Request access to its weights at the ♾️ [Meta AI website](https://ai.meta.com/resources/models-and-libraries/llama-downloads/) and 🤗 [Model Hub](https://huggingface.co/meta-llama/Llama-2-70b-hf), then run `huggingface-cli login` in the terminal before loading the model. Or just try it in our [chatbot app](https://chat.petals.dev).
🔏 **Privacy.** Your data will be processed with the help of other people in the public swarm. Learn more about privacy [here](https://github.com/bigscience-workshop/petals/wiki/Security,-privacy,-and-AI-safety). For sensitive data, you can set up a [private swarm](https://github.com/bigscience-workshop/petals/wiki/Launch-your-own-swarm) among people you trust.
🔏 **Privacy.** Your data will be processed by other people in the public swarm. Learn more about privacy [here](https://github.com/bigscience-workshop/petals/wiki/Security,-privacy,-and-AI-safety). For sensitive data, you can set up a [private swarm](https://github.com/bigscience-workshop/petals/wiki/Launch-your-own-swarm) among people you trust.
🦙 **Want to run Llama 2?** Request access to its weights at the ♾️ [Meta AI website](https://ai.meta.com/resources/models-and-libraries/llama-downloads/) and 🤗 [Model Hub](https://huggingface.co/meta-llama/Llama-2-70b-hf), then run `huggingface-cli login` in the terminal before loading the model. Or just try it in our [chatbot app](https://chat.petals.dev).
💬 **Any questions?** Ping us in [our Discord](https://discord.gg/KdThf2bWVU)!
💬 **Any questions?** Ping us in [our Discord](https://discord.gg/KdThf2bWVU)!
- Petals runs large language models like [Llama](https://github.com/facebookresearch/llama/blob/main/MODEL_CARD.md) and [BLOOM](https://huggingface.co/bigscience/bloom) **collaboratively** — you load a small part of the model, then join people serving the other parts to run inference or fine-tuning.
- You load a small part of the model, then join a [network](https://health.petals.dev) of people serving the other parts. Single‑batch inference runs at up to **6 tokens/sec** for **Llama 2** (70B) and up to **4 tokens/sec** for **Falcon** (180B) — enough for [chatbots](https://chat.petals.dev) and interactive apps.
- Single-batch inference runs at **up to 6 steps/sec** for **Llama 2** (70B) and ≈ 1 step/sec for BLOOM-176B. This is [up to 10x faster](https://github.com/bigscience-workshop/petals#benchmarks) than offloading, enough to build [chatbots](https://chat.petals.dev) and other interactive apps. Parallel inference reaches hundreds of tokens/sec.
- You can employ any fine-tuning and sampling methods, execute custom paths through the model, or see its hidden states. You get the comforts of an API with the flexibility of **PyTorch** and **🤗 Transformers**.
- Beyond classic language model APIs — you can employ any fine-tuning and sampling methods, execute custom paths through the model, or see its hidden states. You get the comforts of an API with the flexibility of PyTorch.
<thcolspan="6">Offloading, max. possible speed on 1x A100 <sup>1</sup></th>
</tr>
<tralign="center">
<td>256 Gbit/s</td>
<td></td>
<td>0.18</td>
<td>0.18</td>
<td>2.7</td>
<td>170.3</td>
</tr>
<tralign="center">
<td>128 Gbit/s</td>
<td></td>
<td>0.09</td>
<td>0.09</td>
<td>2.4</td>
<td>152.8</td>
</tr>
<tr>
<thcolspan="6">Petals on 14 heterogeneous servers across Europe and North America <sup>2</sup></th>
</tr>
<tralign="center">
<tdcolspan="2">Real world</td>
<td>0.83</td>
<td>0.79</td>
<td>32.6</td>
<td>179.4</td>
</tr>
<tr>
<thcolspan="6">Petals on 3 servers, with one A100 each <sup>3</sup></th>
</tr>
<tralign="center">
<td>1 Gbit/s</td>
<td>< 5 ms</td>
<td>1.71</td>
<td>1.54</td>
<td>70.0</td>
<td>253.6</td>
</tr>
<tralign="center">
<td>100 Mbit/s</td>
<td>< 5 ms</td>
<td>1.66</td>
<td>1.49</td>
<td>56.4</td>
<td>182.0</td>
</tr>
<tralign="center">
<td>100 Mbit/s</td>
<td>100 ms</td>
<td>1.23</td>
<td>1.11</td>
<td>19.7</td>
<td>112.2</td>
</tr>
</table>
<sup>1</sup>**An upper bound for offloading performance.** We base our offloading numbers on the best possible hardware setup for offloading: CPU RAM offloading via PCIe 4.0 with 16 PCIe lanes per GPU and PCIe switches for pairs of GPUs. We assume zero latency for the upper bound estimation. In 8-bit, the model uses 1 GB of memory per billion parameters. PCIe 4.0 with 16 lanes has a throughput of 256 Gbit/s, so offloading 176B parameters takes 5.5 seconds. The throughput is twice as slow (128 Gbit/s) if we have two GPUs behind the same PCIe switch.
<sup>2</sup>**A real-world distributed setting** with 14 servers holding 2× RTX 3060, 4× 2080Ti, 2× 3090, 2× A4000, and 4× A5000 GPUs. These are personal servers and servers from university labs, spread across Europe and North America and connected to the Internet at speeds of 100–1000 Mbit/s. 4 servers operate from under firewalls.
<sup>3</sup>**An optimistic setup** that requires least communication. The client nodes have 8 CPU cores and no GPU.
We provide more evaluations and discuss these results in more detail in **Section 3.3** of our [paper](https://arxiv.org/pdf/2209.01188.pdf).
## 🛠️ Contributing
Please see our [FAQ](https://github.com/bigscience-workshop/petals/wiki/FAQ:-Frequently-asked-questions#contributing) on contributing.
Please see our [FAQ](https://github.com/bigscience-workshop/petals/wiki/FAQ:-Frequently-asked-questions#contributing) on contributing.
## 📜 Citation
### 📜 Citation
Alexander Borzunov, Dmitry Baranchuk, Tim Dettmers, Max Ryabinin, Younes Belkada, Artem Chumachenko, Pavel Samygin, and Colin Raffel.
Alexander Borzunov, Dmitry Baranchuk, Tim Dettmers, Max Ryabinin, Younes Belkada, Artem Chumachenko, Pavel Samygin, and Colin Raffel.
[Petals: Collaborative Inference and Fine-tuning of Large Models.](https://arxiv.org/abs/2209.01188)
[Petals: Collaborative Inference and Fine-tuning of Large Models.](https://arxiv.org/abs/2209.01188)
Alexander Borzunov
8 months ago
committed by
GitHub