Merge branch 'staging' into patch-4

Author: thewhaleking

.github/workflows/e2e-subtensor-tests.yaml

@@ -39,6 +39,8 @@ jobs:
         id: get-tests
         run: |
           test_files=$(find tests/e2e_tests -name "test*.py" | jq -R -s -c 'split("\n") | map(select(. != ""))')
+          # keep it here for future debug
+          # test_files=$(find tests/e2e_tests -type f -name "test*.py" | grep -E 'test_(incentive|commit_weights|set_weights)\.py$' | jq -R -s -c 'split("\n") | map(select(. != ""))')
           echo "::set-output name=test-files::$test_files"
         shell: bash
 
@@ -61,7 +63,7 @@ jobs:
           path: subtensor-localnet.tar
 
   # Job to run tests in parallel
-  run:
+  run-e2e-test:
     name: ${{ matrix.test-file }} / Python ${{ matrix.python-version }}
     needs:
       - find-tests
@@ -70,7 +72,7 @@ jobs:
     timeout-minutes: 45
     strategy:
       fail-fast: false  # Allow other matrix jobs to run even if this job fails
-      max-parallel: 32  # Set the maximum number of parallel jobs (same as we have cores in SubtensorCI runner)
+      max-parallel: 32  # Set the maximum number of parallel jobs (same as we have cores in ubuntu-latest runner)
       matrix:
         os:
           - ubuntu-latest

README.md

@@ -11,7 +11,7 @@
 
 ## Internet-scale Neural Networks <!-- omit in toc -->
 
-[Discord](https://discord.gg/qasY3HA9F9) • [Network](https://taostats.io/) • [Research](https://bittensor.com/whitepaper)
+[Discord](https://discord.gg/qasY3HA9F9) • [Network](https://taostats.io/) • [Research](https://bittensor.com/whitepaper) • [Documentation](https://docs.bittensor.com)
 
 </div>
 

bittensor/core/async_subtensor.py

@@ -1687,6 +1687,38 @@ async def get_neuron_for_pubkey_and_subnet(
                 reuse_block=reuse_block,
             )
 
+    async def get_next_epoch_start_block(
+        self,
+        netuid: int,
+        block: Optional[int] = None,
+        block_hash: Optional[str] = None,
+        reuse_block: bool = False,
+    ) -> Optional[int]:
+        """
+        Calculates the first block number of the next epoch for the given subnet.
+
+        If `block` is not provided, the current chain block will be used. Epochs are
+        determined based on the subnet's tempo (i.e., blocks per epoch). The result
+        is the block number at which the next epoch will begin.
+
+        Args:
+            netuid (int): The unique identifier of the subnet.
+            block (Optional[int], optional): The reference block to calculate from.
+                If None, uses the current chain block height.
+            block_hash (Optional[int]): The blockchain block number at which to perform the query.
+            reuse_block (bool): Whether to reuse the last-used blockchain block hash.
+
+
+        Returns:
+            int: The block number at which the next epoch will start.
+        """
+        block_hash = await self.determine_block_hash(block, block_hash, reuse_block)
+        if not block_hash and reuse_block:
+            block_hash = self.substrate.last_block_hash
+        block = await self.substrate.get_block_number(block_hash=block_hash)
+        tempo = await self.tempo(netuid=netuid, block_hash=block_hash)
+        return (((block // tempo) + 1) * tempo) + 1 if tempo else None
+
     async def get_owned_hotkeys(
         self,
         coldkey_ss58: str,
@@ -3813,6 +3845,7 @@ async def set_weights(
         wait_for_finalization: bool = False,
         max_retries: int = 5,
         block_time: float = 12.0,
+        period: int = 5,
     ):
         """
         Sets the inter-neuronal weights for the specified neuron. This process involves specifying the influence or
@@ -3833,6 +3866,7 @@ async def set_weights(
                 ``False``.
             max_retries (int): The number of maximum attempts to set weights. Default is ``5``.
             block_time (float): The amount of seconds for block duration. Default is 12.0 seconds.
+            period (int, optional): The period in seconds to wait for extrinsic inclusion or finalization. Defaults to 5.
 
         Returns:
             tuple[bool, str]: ``True`` if the setting of weights is successful, False otherwise. And `msg`, a string
@@ -3907,6 +3941,7 @@ async def _blocks_weight_limit() -> bool:
                         version_key=version_key,
                         wait_for_inclusion=wait_for_inclusion,
                         wait_for_finalization=wait_for_finalization,
+                        period=period,
                     )
                 except Exception as e:
                     logging.error(f"Error setting weights: {e}")

bittensor/core/extrinsics/asyncex/weights.py

@@ -287,6 +287,7 @@ async def set_weights_extrinsic(
     version_key: int = 0,
     wait_for_inclusion: bool = False,
     wait_for_finalization: bool = False,
+    period: int = 5,
 ) -> tuple[bool, str]:
     """Sets the given weights and values on chain for wallet hotkey account.
 
@@ -302,6 +303,7 @@ async def set_weights_extrinsic(
             returns ``False`` if the extrinsic fails to enter the block within the timeout.
         wait_for_finalization (bool): If set, waits for the extrinsic to be finalized on the chain before returning
             ``True``, or returns ``False`` if the extrinsic fails to be finalized within the timeout.
+        period (int, optional): The period in seconds to wait for extrinsic inclusion or finalization. Defaults to 5.
 
     Returns:
         success (bool): Flag is ``True`` if extrinsic was finalized or included in the block. If we did not wait for
@@ -331,6 +333,7 @@ async def set_weights_extrinsic(
             version_key=version_key,
             wait_for_finalization=wait_for_finalization,
             wait_for_inclusion=wait_for_inclusion,
+            period=period,
         )
 
         if not wait_for_finalization and not wait_for_inclusion:

bittensor/core/extrinsics/set_weights.py

@@ -91,6 +91,7 @@ def set_weights_extrinsic(
     version_key: int = 0,
     wait_for_inclusion: bool = False,
     wait_for_finalization: bool = False,
+    period: int = 5,
 ) -> tuple[bool, str]:
     """Sets the given weights and values on chain for wallet hotkey account.
 
@@ -106,6 +107,7 @@ def set_weights_extrinsic(
             returns ``False`` if the extrinsic fails to enter the block within the timeout.
         wait_for_finalization (bool): If set, waits for the extrinsic to be finalized on the chain before returning
             ``True``, or returns ``False`` if the extrinsic fails to be finalized within the timeout.
+        period (int, optional): The period in seconds to wait for extrinsic inclusion or finalization. Defaults to 5.
 
     Returns:
         success (bool): Flag is ``True`` if extrinsic was finalized or included in the block. If we did not wait for
@@ -135,6 +137,7 @@ def set_weights_extrinsic(
             version_key=version_key,
             wait_for_finalization=wait_for_finalization,
             wait_for_inclusion=wait_for_inclusion,
+            period=period,
         )
 
         if not wait_for_finalization and not wait_for_inclusion:

bittensor/core/metagraph.py

@@ -1,4 +1,5 @@
 import asyncio
+import contextlib
 import copy
 import os
 import pickle
@@ -11,6 +12,7 @@
 import numpy as np
 from async_substrate_interface.errors import SubstrateRequestException
 from numpy.typing import NDArray
+from packaging import version
 
 from bittensor.core import settings
 from bittensor.core.chain_data import (
@@ -143,6 +145,27 @@ def latest_block_path(dir_path: str) -> str:
         return latest_file_full_path
 
 
+def safe_globals():
+    """
+    Context manager to load torch files for version 2.6+
+    """
+    if version.parse(torch.__version__).release < version.parse("2.6").release:
+        return contextlib.nullcontext()
+
+    np_core = (
+        np._core if version.parse(np.__version__) >= version.parse("2.0.0") else np.core
+    )
+    allow_list = [
+        np_core.multiarray._reconstruct,
+        np.ndarray,
+        np.dtype,
+        type(np.dtype(np.uint32)),
+        np.dtypes.Float32DType,
+        bytes,
+    ]
+    return torch.serialization.safe_globals(allow_list)
+
+
 class MetagraphMixin(ABC):
     """
     The metagraph class is a core component of the Bittensor network, representing the neural graph that forms the
@@ -1124,7 +1147,8 @@ def load_from_path(self, dir_path: str) -> "MetagraphMixin":
         """
 
         graph_file = latest_block_path(dir_path)
-        state_dict = torch.load(graph_file)
+        with safe_globals():
+            state_dict = torch.load(graph_file)
         self.n = torch.nn.Parameter(state_dict["n"], requires_grad=False)
         self.block = torch.nn.Parameter(state_dict["block"], requires_grad=False)
         self.uids = torch.nn.Parameter(state_dict["uids"], requires_grad=False)
@@ -1256,7 +1280,8 @@ def load_from_path(self, dir_path: str) -> "MetagraphMixin":
             try:
                 import torch as real_torch
 
-                state_dict = real_torch.load(graph_filename)
+                with safe_globals():
+                    state_dict = real_torch.load(graph_filename)
                 for key in METAGRAPH_STATE_DICT_NDARRAY_KEYS:
                     state_dict[key] = state_dict[key].detach().numpy()
                 del real_torch

bittensor/core/subtensor.py

@@ -1300,6 +1300,28 @@ def get_neuron_for_pubkey_and_subnet(
 
         return NeuronInfo.from_dict(result)
 
+    def get_next_epoch_start_block(
+        self, netuid: int, block: Optional[int] = None
+    ) -> Optional[int]:
+        """
+        Calculates the first block number of the next epoch for the given subnet.
+
+        If `block` is not provided, the current chain block will be used. Epochs are
+        determined based on the subnet's tempo (i.e., blocks per epoch). The result
+        is the block number at which the next epoch will begin.
+
+        Args:
+            netuid (int): The unique identifier of the subnet.
+            block (Optional[int], optional): The reference block to calculate from.
+                If None, uses the current chain block height.
+
+        Returns:
+            int: The block number at which the next epoch will start.
+        """
+        block = block or self.block
+        tempo = self.tempo(netuid=netuid, block=block)
+        return (((block // tempo) + 1) * tempo) + 1 if tempo else None
+
     def get_owned_hotkeys(
         self,
         coldkey_ss58: str,
@@ -3100,6 +3122,7 @@ def set_weights(
         wait_for_finalization: bool = False,
         max_retries: int = 5,
         block_time: float = 12.0,
+        period: int = 5,
     ) -> tuple[bool, str]:
         """
         Sets the inter-neuronal weights for the specified neuron. This process involves specifying the influence or
@@ -3120,6 +3143,7 @@ def set_weights(
                 ``False``.
             max_retries (int): The number of maximum attempts to set weights. Default is ``5``.
             block_time (float): The amount of seconds for block duration. Default is 12.0 seconds.
+            period (int, optional): The period in seconds to wait for extrinsic inclusion or finalization. Defaults to 5.
 
         Returns:
             tuple[bool, str]: ``True`` if the setting of weights is successful, False otherwise. And `msg`, a string
@@ -3183,6 +3207,7 @@ def _blocks_weight_limit() -> bool:
                         version_key=version_key,
                         wait_for_inclusion=wait_for_inclusion,
                         wait_for_finalization=wait_for_finalization,
+                        period=period,
                     )
                 except Exception as e:
                     logging.error(f"Error setting weights: {e}")

bittensor/utils/__init__.py

@@ -250,6 +250,16 @@ def format_error_message(error_message: Union[dict, Exception]) -> str:
             err_docs = error_message.get("docs", [err_description])
             err_description = err_docs[0] if err_docs else err_description
 
+        elif error_message.get("code") and error_message.get("message"):
+            err_type = error_message.get("code", err_name)
+            err_name = "Custom type"
+            err_description = error_message.get("message", err_description)
+
+        else:
+            logging.error(
+                f"String representation of real error_message: {str(error_message)}"
+            )
+
     return f"Subtensor returned `{err_name}({err_type})` error. This means: `{err_description}`."
 
 

contrib/CONTRIBUTING.md

@@ -252,7 +252,7 @@ Explain the problem and include additional details to help maintainers reproduce
 * **Describe the behavior you observed after following the steps** and point out what exactly is the problem with that behavior.
 * **Explain which behavior you expected to see instead and why.**
 * **Include screenshots and animated GIFs** which show you following the described steps and clearly demonstrate the problem. You can use [this tool](https://www.cockos.com/licecap/) to record GIFs on macOS and Windows, and [this tool](https://github.com/colinkeenan/silentcast) or [this tool](https://github.com/GNOME/byzanz) on Linux.
-* **If you're reporting that Bittensor crashed**, include a crash report with a stack trace from the operating system. On macOS, the crash report will be available in `Console.app` under "Diagnostic and usage information" > "User diagnostic reports". Include the crash report in the issue in a [code block](https://help.github.com/articles/markdown-basics/#multiple-lines), a [file attachment](https://help.github.com/articles/file-attachments-on-issues-and-pull-requests/), or put it in a [gist](https://gist.github.com/) and provide link to that gist.
+* **If you're reporting that Bittensor crashed**, include a crash report with a stack trace from the operating system. On macOS, the crash report will be available in `Console.app` under "Diagnostic and usage information" > "User diagnostic reports". Include the crash report in the issue in a [code block](https://help.github.com/articles/markdown-basics/#multiple-lines), a [file attachment](https://docs.github.com/articles/file-attachments-on-issues-and-pull-requests/), or put it in a [gist](https://gist.github.com/) and provide link to that gist.
 * **If the problem is related to performance or memory**, include a CPU profile capture with your report, if you're using a GPU then include a GPU profile capture as well. Look into the [PyTorch Profiler](https://pytorch.org/tutorials/recipes/recipes/profiler_recipe.html) to look at memory usage of your model.
 * **If the problem wasn't triggered by a specific action**, describe what you were doing before the problem happened and share more information using the guidelines below.
 

pyproject.toml

@@ -1,5 +1,5 @@
 [build-system]
-requires = ["setuptools~=70.0.0", "wheel"]
+requires = ["setuptools>=70.0.0", "wheel"]
 build-backend = "setuptools.build_meta"
 
 [project]
@@ -40,11 +40,11 @@ dependencies = [
 
 [project.optional-dependencies]
 dev = [
-    "pytest==7.2.0",
-    "pytest-asyncio==0.23.7",
-    "pytest-mock==3.12.0",
-    "pytest-split==0.8.0",
-    "pytest-xdist==3.0.2",
+    "pytest==8.3.5",
+    "pytest-asyncio==0.26.0",
+    "pytest-mock==3.14.0",
+    "pytest-split==0.10.0",
+    "pytest-xdist==3.6.1",
     "pytest-rerunfailures==10.2",
     "coveralls==3.3.1",
     "pytest-cov==4.0.0",
@@ -59,10 +59,10 @@ dev = [
     "aioresponses==0.7.6",
     "factory-boy==3.3.0",
     "types-requests",
-    "torch>=1.13.1,<2.6.0"
+    "torch>=1.13.1,<3.0"
 ]
 torch = [
-    "torch>=1.13.1,<2.6.0"
+    "torch>=1.13.1,<3.0"
 ]
 cli = [
     "bittensor-cli>=9.0.2"