Rechain reloc

.github/workflows/run-tests.yaml

@@ -14,8 +14,8 @@ jobs:
           - { model: 'bigscience/bloom-560m', os: 'ubuntu', python-version: '3.11' }
           - { model: 'Maykeye/TinyLLama-v0', os: 'ubuntu', python-version: '3.8' }
           - { model: 'Maykeye/TinyLLama-v0', os: 'ubuntu', python-version: '3.11' }
-          - { model: 'Maykeye/TinyLLama-v0', os: 'macos-14', python-version: '3.10' }
-          - { model: 'Maykeye/TinyLLama-v0', os: 'macos-14', python-version: '3.11' }
+          - { model: 'Maykeye/TinyLLama-v0', os: 'macos', python-version: '3.10' }
+          - { model: 'Maykeye/TinyLLama-v0', os: 'macos', python-version: '3.11' }
           - { model: 'artek0chumak/TestMixtral', os: 'ubuntu', python-version: '3.8' }
           - { model: 'artek0chumak/TestMixtral', os: 'ubuntu', python-version: '3.11' }
       fail-fast: false

src/petals/server/throughput.py

@@ -206,7 +206,7 @@ def measure_compute_rps(
         block = block.to(dtype)
         block = convert_block(block, 0, config, tensor_parallel_devices, device, quant_type=quant_type, freeze=True)
 
-        cache = (DUMMY_KEY_PAST.to(dtype).to(device), DUMMY_KEY_PAST.to(dtype).to(device))
+        cache = (DUMMY_KEY_PAST.to(dtype=dtype, device=device), DUMMY_KEY_PAST.to(dtype=dtype, device=device))
         elapsed = 0
         dummy_input = torch.randn(1, n_tokens, config.hidden_size, device=device, dtype=dtype)