Update to petals.dev (#390)

Since `petals.ml` DNS record is still unavailable, we're switching everything to https://petals.dev

Co-authored-by: Aleksandr Borzunov <hxrussia@gmail.com>

README.md

@@ -65,8 +65,8 @@ Basic tutorials:
 
 Useful tools and advanced guides:
 
-- [Chatbot web app](http://chat.petals.ml) (connects to Petals via an HTTP/WebSocket endpoint): [source code](https://github.com/borzunov/chat.petals.ml)
+- [Chatbot web app](https://chat.petals.dev) (connects to Petals via an HTTP/WebSocket endpoint): [source code](https://github.com/borzunov/chat.petals.dev)
-- [Monitor](http://health.petals.ml) for the public swarm: [source code](https://github.com/borzunov/health.petals.ml)
+- [Monitor](https://health.petals.dev) for the public swarm: [source code](https://github.com/borzunov/health.petals.dev)
 - Launch your own swarm: [guide](https://github.com/bigscience-workshop/petals/wiki/Launch-your-own-swarm)
 - Run a custom foundation model: [guide](https://github.com/bigscience-workshop/petals/wiki/Run-a-custom-model-with-Petals)
 
@@ -78,7 +78,7 @@ Learning more:
 ## How does it work?
 
 - 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 team up with people serving the other parts to run inference or fine-tuning.
-- 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 for [chatbots](http://chat.petals.ml) and other interactive apps. Parallel inference reaches hundreds of tokens/sec.
+- 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 for [chatbots](https://chat.petals.dev) and other interactive apps. Parallel inference reaches hundreds of tokens/sec.
 - 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.
 
 <p align="center">
@@ -218,5 +218,5 @@ _arXiv preprint arXiv:2209.01188,_ 2022.
     This project is a part of the <a href="https://bigscience.huggingface.co/">BigScience</a> research workshop.
 </p>
 <p align="center">
-    <img src="https://petals.ml/bigscience.png" width="150">
+    <img src="https://petals.dev/bigscience.png" width="150">
 </p>

examples/prompt-tuning-sst2.ipynb

@@ -330,7 +330,7 @@
     "id": "51770911"
    },
    "source": [
-    "Our model has been trained! You can now upload it to the Hub for later use, try out different models [served in the public swarm](http://health.petals.ml/), or [join Petals with your own GPU](https://github.com/bigscience-workshop/petals#connect-your-gpu-and-increase-petals-capacity)!"
+    "Our model has been trained! You can now upload it to the Hub for later use, try out different models [served in the public swarm](https://health.petals.dev/), or [join Petals with your own GPU](https://github.com/bigscience-workshop/petals#connect-your-gpu-and-increase-petals-capacity)!"
    ]
   },
   {

src/petals/cli/run_server.py

@@ -152,7 +152,7 @@ def main():
                         "weight matrix. See https://huggingface.co/transformers/v4.9.0/parallelism.html#tensor-parallelism")
 
     parser.add_argument("--skip_reachability_check", action='store_true',
-                        help="Skip checking this server's reachability via health.petals.ml "
+                        help="Skip checking this server's reachability via health.petals.dev "
                              "when connecting to the public swarm. If you connect to a private swarm, "
                              "the check is skipped by default. Use this option only if you know what you are doing")
 

src/petals/client/routing/sequence_manager.py

@@ -513,7 +513,7 @@ class MissingBlocksError(RuntimeError):
     def __init__(self, block_indices: Union[int, Sequence[int]]):
         super().__init__(
             f"No servers holding blocks {block_indices} are online. "
-            f"You can check the public swarm's state at http://health.petals.ml "
+            f"You can check the public swarm's state at https://health.petals.dev "
             f"If there are not enough servers, please connect your GPU: "
             f"https://github.com/bigscience-workshop/petals#connect-your-gpu-and-increase-petals-capacity "
         )

src/petals/constants.py

@@ -1,17 +1,18 @@
 import torch
 
 PUBLIC_INITIAL_PEERS = [
-    # Temporary IPs until DNS issues get resolved
+    # IPv4 DNS addresses
-    "/ip4/159.223.29.252/tcp/31337/p2p/QmedTaZXmULqwspJXz44SsPZyTNKxhnnFvYRajfH7MGhCY",
+    "/dns/bootstrap1.petals.dev/tcp/31337/p2p/QmedTaZXmULqwspJXz44SsPZyTNKxhnnFvYRajfH7MGhCY",
-    "/ip4/24.144.96.147/tcp/31338/p2p/QmQGTqmM7NKjV6ggU1ZCap8zWiyKR89RViDXiqehSiCpY5",
+    "/dns/bootstrap2.petals.dev/tcp/31338/p2p/QmQGTqmM7NKjV6ggU1ZCap8zWiyKR89RViDXiqehSiCpY5",
-    # Default DNS addresses
+    # IPv6 DNS addresses
-    "/dns/bootstrap1.petals.ml/tcp/31337/p2p/QmedTaZXmULqwspJXz44SsPZyTNKxhnnFvYRajfH7MGhCY",
+    "/dns6/bootstrap1.petals.dev/tcp/31337/p2p/QmedTaZXmULqwspJXz44SsPZyTNKxhnnFvYRajfH7MGhCY",
-    "/dns6/bootstrap1.petals.ml/tcp/31337/p2p/QmedTaZXmULqwspJXz44SsPZyTNKxhnnFvYRajfH7MGhCY",
+    "/dns6/bootstrap2.petals.dev/tcp/31338/p2p/QmQGTqmM7NKjV6ggU1ZCap8zWiyKR89RViDXiqehSiCpY5",
-    "/dns/bootstrap2.petals.ml/tcp/31338/p2p/QmQGTqmM7NKjV6ggU1ZCap8zWiyKR89RViDXiqehSiCpY5",
+    # Reserved IPs
-    "/dns6/bootstrap2.petals.ml/tcp/31338/p2p/QmQGTqmM7NKjV6ggU1ZCap8zWiyKR89RViDXiqehSiCpY5",
+    "/ip4/159.89.214.152/tcp/31337/p2p/QmedTaZXmULqwspJXz44SsPZyTNKxhnnFvYRajfH7MGhCY",
+    "/ip4/159.203.156.48/tcp/31338/p2p/QmQGTqmM7NKjV6ggU1ZCap8zWiyKR89RViDXiqehSiCpY5",
 ]
 
 # The reachability API is currently used only when connecting to the public swarm
-REACHABILITY_API_URL = "http://health.petals.ml"
+REACHABILITY_API_URL = "https://health.petals.dev"
 
 DTYPE_MAP = dict(bfloat16=torch.bfloat16, float16=torch.float16, float32=torch.float32, auto="auto")

src/petals/server/reachability.py

@@ -28,7 +28,7 @@ def validate_reachability(peer_id, wait_time: float = 7 * 60, retry_delay: float
             response = r.json()
 
             if response["success"]:
-                logger.info("Server is reachable from the Internet. It will appear at http://health.petals.ml soon")
+                logger.info("Server is reachable from the Internet. It will appear at https://health.petals.dev soon")
                 return
 
             if attempt_no == 0:
@@ -37,7 +37,7 @@ def validate_reachability(peer_id, wait_time: float = 7 * 60, retry_delay: float
                 logger.info("Detected a NAT or a firewall, connecting to libp2p relays. This takes a few minutes")
             time.sleep(retry_delay)
         except Exception as e:
-            logger.warning(f"Skipping reachability check because health.petals.ml is down: {repr(e)}")
+            logger.warning(f"Skipping reachability check because health.petals.dev is down: {repr(e)}")
             return
 
     raise RuntimeError(