fix perf_sigmoid_fitter.py (official-stockfish#346)

Disservin · web-flow · commit dc89b24aeb58 · 2025-09-21T20:02:38.000+02:00
`SparseBatchDataset` expects a list of strings but a string was given
diff --git a/data_loader/dataset.py b/data_loader/dataset.py
@@ -7,6 +7,8 @@
 from . import stream
 from .config import DataloaderSkipConfig
 
+from typing import List
+
 
 class FenBatchProvider:
     def __init__(
@@ -62,7 +64,7 @@ def __init__(
         destroy_stream,
         fetch_next,
         destroy_part,
-        filenames,
+        filenames: List[str],
         cyclic,
         num_workers,
         batch_size=None,
@@ -114,7 +116,7 @@ class SparseBatchProvider(TrainingDataProvider):
     def __init__(
         self,
         feature_set: str,
-        filenames,
+        filenames: List[str],
         batch_size,
         cyclic=True,
         num_workers=1,
@@ -138,7 +140,7 @@ class SparseBatchDataset(torch.utils.data.IterableDataset):
     def __init__(
         self,
         feature_set: str,
-        filenames,
+        filenames: List[str],
         batch_size,
         cyclic=True,
         num_workers=1,
diff --git a/data_loader/stream.py b/data_loader/stream.py
@@ -3,6 +3,7 @@
 from ._native import c_lib
 from .config import CDataloaderSkipConfig, DataloaderSkipConfig
 from features.feature_set import FeatureSet
+from typing import List
 
 
 def _to_c_str_array(str_list):
@@ -13,7 +14,7 @@ def _to_c_str_array(str_list):
 
 def create_fen_batch_stream(
     concurrency,
-    filenames,
+    filenames: List[str],
     batch_size,
     cyclic,
     config: DataloaderSkipConfig,
@@ -43,7 +44,7 @@ def destroy_fen_batch(fen_batch):
 def create_sparse_batch_stream(
     feature_set: str,
     concurrency,
-    filenames,
+    filenames: List[str],
     batch_size,
     cyclic,
     config: DataloaderSkipConfig,
diff --git a/perf_sigmoid_fitter.py b/perf_sigmoid_fitter.py
@@ -92,7 +92,7 @@ def gather_statistics_from_batches(batches, bucket_size):
     return data
 
 
-def gather_statistics_from_data(filename, count, bucket_size):
+def gather_statistics_from_data(filename: str, count: int, bucket_size: int):
     """
     Takes a .bin or .binpack file and produces perf% statistics
     The result is a dictionary of the form { eval : (perf%, count) }
@@ -105,7 +105,7 @@ def gather_statistics_from_data(filename, count, bucket_size):
     # this is just the easiest way to do it
     dataset = data_loader.SparseBatchDataset(
         "HalfKP",
-        filename,
+        [filename],
         batch_size,
         cyclic,
         1,
