Method device_name

accelerator/hpu_accelerator.py:56–58 · view source on GitHub ↗

(self, device_index=None)

Source from the content-addressed store, hash-verified

54	return True
55
56	def device_name(self, device_index=None):
57	# ignoring device_index.
58	return 'hpu'
59
60	def device(self, device_index=None):
61	return torch.device(self.device_name(device_index))

deviceMethod · 0.95

set_compile_backendMethod · 0.95

setup.pyFile · 0.45

set_accelerator_visibleFunction · 0.45

_launch_daemonic_procsMethod · 0.45

_launch_procsMethod · 0.45

get_gpt2_modelFunction · 0.45

skip_on_archFunction · 0.45

skip_on_cudaFunction · 0.45

bf16_required_version_checkFunction · 0.45

train_cifarFunction · 0.45

test_cpu_adagrad_gpu_errorMethod · 0.45

no outgoing calls

test_cpu_adagrad_gpu_errorMethod · 0.36

init_softmax_inputsFunction · 0.36

init_matmul_inputsFunction · 0.36

test_fused_lion_equalMethod · 0.36

test_cpu_lion_gpu_errorMethod · 0.36

test_DS4Sci_EvoformerAttentionFunction · 0.36

get_q_propsFunction · 0.36

get_scale_zero_pointFunction · 0.36

test_float_quantizeFunction · 0.36

test_fake_quant_dequantFunction · 0.36

create_modelsFunction · 0.36