Merge pull request #1919 from opentensor/thewhaleking/further-legacy-…

…torch-improvements

thewhaleking/further-legacy-torch-improvements

bittensor/commands/root.py

@@ -24,7 +24,6 @@
 from rich.prompt import Prompt
 from rich.table import Table
 from .utils import get_delegates_details, DelegatesDetails
-from bittensor.utils.registration import torch, use_torch
 
 from . import defaults
 
@@ -283,7 +282,6 @@ def run(cli: "bittensor.cli"):
     def _run(cli: "bittensor.cli", subtensor: "bittensor.subtensor"):
         r"""Set weights for root network."""
         wallet = bittensor.wallet(config=cli.config)
-        subnets: List[bittensor.SubnetInfo] = subtensor.get_all_subnets_info()
 
         root = subtensor.metagraph(0, lite=False)
         try:
@@ -301,11 +299,7 @@ def _run(cli: "bittensor.cli", subtensor: "bittensor.subtensor"):
             f"Boosting weight for netuid {cli.config.netuid} from {prev_weight} -> {new_weight}"
         )
         my_weights[cli.config.netuid] = new_weight
-        all_netuids = (
-            torch.tensor(list(range(len(my_weights))))
-            if use_torch()
-            else np.arange(len(my_weights))
-        )
+        all_netuids = np.arange(len(my_weights))
 
         bittensor.__console__.print("Setting root weights...")
         subtensor.root_set_weights(
@@ -405,7 +399,6 @@ def run(cli: "bittensor.cli"):
     @staticmethod
     def _run(cli: "bittensor.cli", subtensor: "bittensor.subtensor"):
         wallet = bittensor.wallet(config=cli.config)
-        subnets: List[bittensor.SubnetInfo] = subtensor.get_all_subnets_info()
 
         bittensor.__console__.print(
             "Slashing weight for subnet: {} by amount: {}".format(
@@ -423,11 +416,7 @@ def _run(cli: "bittensor.cli", subtensor: "bittensor.subtensor"):
         my_weights = root.weights[my_uid]
         my_weights[cli.config.netuid] -= cli.config.amount
         my_weights[my_weights < 0] = 0  # Ensure weights don't go negative
-        all_netuids = (
-            torch.tensor(list(range(len(my_weights))))
-            if use_torch()
-            else np.arange(len(my_weights))
-        )
+        all_netuids = np.arange(len(my_weights))
 
         subtensor.root_set_weights(
             wallet=wallet,
@@ -529,23 +518,12 @@ def _run(cli: "bittensor.cli", subtensor: "bittensor.subtensor"):
 
         # Parse from string
         matched_netuids = list(map(int, re.split(r"[ ,]+", cli.config.netuids)))
-        netuids = (
-            torch.tensor(matched_netuids, dtype=torch.long)
-            if use_torch()
-            else np.array(matched_netuids, dtype=np.int64)
-        )
+        netuids = np.array(matched_netuids, dtype=np.int64)
 
         matched_weights = [
             float(weight) for weight in re.split(r"[ ,]+", cli.config.weights)
         ]
-        weights = (
-            torch.tensor(matched_weights, dtype=torch.float32)
-            if use_torch()
-            else np.array(
-                matched_weights,
-                dtype=np.float32,
-            )
-        )
+        weights = np.array(matched_weights, dtype=np.float32)
 
         # Run the set weights operation.
         subtensor.root_set_weights(

bittensor/extrinsics/root.py

@@ -26,7 +26,7 @@
 from typing import Union, List
 import bittensor.utils.weight_utils as weight_utils
 from bittensor.btlogging.defines import BITTENSOR_LOGGER_NAME
-from bittensor.utils.registration import torch, use_torch
+from bittensor.utils.registration import torch, legacy_torch_api_compat
 
 logger = logging.getLogger(BITTENSOR_LOGGER_NAME)
 
@@ -100,6 +100,7 @@ def root_register_extrinsic(
                 )
 
 
+@legacy_torch_api_compat
 def set_root_weights_extrinsic(
     subtensor: "bittensor.subtensor",
     wallet: "bittensor.wallet",
@@ -133,36 +134,22 @@ def set_root_weights_extrinsic(
     """
     # First convert types.
     if isinstance(netuids, list):
-        netuids = (
-            torch.tensor(netuids, dtype=torch.int64)
-            if use_torch()
-            else np.array(netuids, dtype=np.int64)
-        )
+        netuids = np.array(netuids, dtype=np.int64)
     if isinstance(weights, list):
-        weights = (
-            torch.tensor(weights, dtype=torch.float32)
-            if use_torch()
-            else np.array(weights, dtype=np.float32)
-        )
+        weights = np.array(weights, dtype=np.float32)
 
     # Get weight restrictions.
     min_allowed_weights = subtensor.min_allowed_weights(netuid=0)
     max_weight_limit = subtensor.max_weight_limit(netuid=0)
 
     # Get non zero values.
-    non_zero_weight_idx = (
-        torch.argwhere(weights > 0).squeeze(dim=1)
-        if use_torch()
-        else np.argwhere(weights > 0).squeeze(axis=1)
-    )
+    non_zero_weight_idx = np.argwhere(weights > 0).squeeze(axis=1)
+    non_zero_weight_uids = netuids[non_zero_weight_idx]
     non_zero_weights = weights[non_zero_weight_idx]
-    non_zero_weights_size = (
-        non_zero_weights.numel() if use_torch() else non_zero_weights.size
-    )
-    if non_zero_weights_size < min_allowed_weights:
+    if non_zero_weights.size < min_allowed_weights:
         raise ValueError(
             "The minimum number of weights required to set weights is {}, got {}".format(
-                min_allowed_weights, non_zero_weights_size
+                min_allowed_weights, non_zero_weights.size
             )
         )
 

bittensor/subtensor.py

@@ -101,6 +101,7 @@
 from .utils.balance import Balance
 from .utils.registration import POWSolution
 from .utils.subtensor import get_subtensor_errors
+from .utils.registration import legacy_torch_api_compat
 
 
 KEY_NONCE: Dict[str, int] = {}
@@ -2347,6 +2348,7 @@ def make_substrate_call_with_retry():
 
         return make_substrate_call_with_retry()
 
+    @legacy_torch_api_compat
     def root_set_weights(
         self,
         wallet: "bittensor.wallet",

tests/unit_tests/extrinsics/test_root.py

@@ -212,3 +212,97 @@ def test_set_root_weights_extrinsic(
             mock_confirm.assert_called_once()
         else:
             mock_confirm.assert_not_called()
+
+
+@pytest.mark.parametrize(
+    "wait_for_inclusion, wait_for_finalization, netuids, weights, prompt, user_response, expected_success",
+    [
+        (True, False, [1, 2], [0.5, 0.5], True, True, True),  # Success - weights set
+        (
+            False,
+            False,
+            [1, 2],
+            [0.5, 0.5],
+            False,
+            None,
+            True,
+        ),  # Success - weights set no wait
+        (
+            True,
+            False,
+            [1, 2],
+            [2000, 20],
+            True,
+            True,
+            True,
+        ),  # Success - large value to be normalized
+        (
+            True,
+            False,
+            [1, 2],
+            [2000, 0],
+            True,
+            True,
+            True,
+        ),  # Success - single large value
+        (
+            True,
+            False,
+            [1, 2],
+            [0.5, 0.5],
+            True,
+            False,
+            False,
+        ),  # Failure - prompt declined
+        (
+            True,
+            False,
+            [1, 2],
+            [0.5, 0.5],
+            False,
+            None,
+            False,
+        ),  # Failure - setting weights failed
+        (
+            True,
+            False,
+            [],
+            [],
+            None,
+            False,
+            False,
+        ),  # Exception catched - ValueError 'min() arg is an empty sequence'
+    ],
+    ids=[
+        "success-weights-set",
+        "success-not-wait",
+        "success-large-value",
+        "success-single-value",
+        "failure-user-declines",
+        "failure-setting-weights",
+        "failure-value-error-exception",
+    ],
+)
+def test_set_root_weights_extrinsic_torch(
+    mock_subtensor,
+    mock_wallet,
+    wait_for_inclusion,
+    wait_for_finalization,
+    netuids,
+    weights,
+    prompt,
+    user_response,
+    expected_success,
+    force_legacy_torch_compat_api,
+):
+    test_set_root_weights_extrinsic(
+        mock_subtensor,
+        mock_wallet,
+        wait_for_inclusion,
+        wait_for_finalization,
+        netuids,
+        weights,
+        prompt,
+        user_response,
+        expected_success,
+    )

tests/unit_tests/utils/test_weight_utils.py

@@ -159,15 +159,15 @@ def test_normalize_with_max_weight__legacy_torch_api_compat(
     wn = weight_utils.normalize_max_weight(weights, limit=1)
     assert torch.isclose(wn, weights / weights.sum(), atol=1e-08, rtol=0).all()
 
-    # Check for eplison changes
-    eplison = 0.01
+    # Check for epsilon changes
+    epsilon = 0.01
     weights, _ = torch.sort(torch.rand(100))
     x = weights / weights.sum()
     limit = x[-10]
-    change = eplison * limit
+    change = epsilon * limit
     y = weight_utils.normalize_max_weight(x, limit=limit - change)
     z = weight_utils.normalize_max_weight(x, limit=limit + change)
-    assert (y - z).abs().sum() < eplison
+    assert (y - z).abs().sum() < epsilon
 
 
 @pytest.mark.parametrize(