File size: 3,497 Bytes
4d423a9 f43498c 4d423a9 f43498c 4d423a9 10c3b6b 049539c 10c3b6b 4d423a9 449a109 4d423a9 449a109 f6dcb7a 4d423a9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 |
import gradio as gr
def get_process_config():
return {
"process.numactl": gr.Checkbox(
value=True,
label="process.numactl",
info="Runs the model with numactl",
),
"process.numactl_kwargs": gr.Textbox(
label="process.numactl_kwargs",
value="{'cpunodebind': 0, 'membind': 0}",
info="Additional python dict of kwargs to pass to numactl",
),
}
def get_inference_config():
return {
"inference.warmup_runs": gr.Slider(
step=1,
value=10,
minimum=0,
maximum=10,
label="inference.warmup_runs",
info="Number of warmup runs",
),
"inference.duration": gr.Slider(
step=1,
value=10,
minimum=0,
maximum=10,
label="inference.duration",
info="Minimum duration of the benchmark in seconds",
),
"inference.iterations": gr.Slider(
step=1,
value=10,
minimum=0,
maximum=10,
label="inference.iterations",
info="Minimum number of iterations of the benchmark",
),
"inference.latency": gr.Checkbox(
value=True,
label="inference.latency",
info="Measures the latency of the model",
),
"inference.memory": gr.Checkbox(
value=True,
label="inference.memory",
info="Measures the peak memory consumption",
),
"inference.input_shapes": gr.Textbox(
label="inference.input_shapes",
value="{'batch_size': 2, 'sequence_length': 16}",
info="Input shapes to use for the benchmark",
),
"inference.generate_kwargs": gr.Textbox(
label="inference.generate_kwargs",
value="{'max_new_tokens': 32, 'min_new_tokens': 32}",
info="Additional python dict of kwargs to pass to the generate function",
),
}
def get_pytorch_config():
return {
"pytorch.torch_dtype": gr.Dropdown(
value="float32",
label="pytorch.torch_dtype",
choices=["bfloat16", "float16", "float32", "auto"],
info="The dtype to use for the model",
),
}
def get_openvino_config():
return {
"openvino.half": gr.Checkbox(
value=False,
label="openvino.half",
info="Converts model to half precision",
),
"openvino.reshape": gr.Checkbox(
value=False,
label="openvino.reshape",
info="Reshapes the model to the input shape",
),
"openvino.reshape_kwargs": gr.Textbox(
label="openvino.reshape_kwargs",
value="{'batch_size': 2, 'sequence_length': 16}",
info="Additional python dict of kwargs to pass to the reshape function",
),
"openvino.compile": gr.Checkbox(
value=False,
label="openvino.compile",
info="Compiles model for the current device",
),
"openvino.load_in_8bit": gr.Checkbox(
value=False,
label="openvino.load_in_8bit",
info="Loads model in 8 bits precision",
),
"openvino.load_in_4bit": gr.Checkbox(
value=False,
label="openvino.load_in_4bit",
info="Loads model in 4 bits precision",
),
}
|