torsh-cli 0.1.2

Command-line tools for the ToRSh deep learning framework
Documentation 
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

# Benchmarking configuration for multi-device comparison
# Usage: torsh benchmark --config examples/configs/benchmark_multi_device.yaml

model_path: ./models/resnet50.torsh

# Test configurations
input_shapes:
  - [3, 224, 224]    # Standard ImageNet input
  - [3, 299, 299]    # Inception-style input
  - [3, 512, 512]    # High-resolution input

batch_sizes:
  - 1      # Latency-optimized
  - 8      # Small batch
  - 16     # Medium batch
  - 32     # Large batch
  - 64     # Very large batch

# Devices to test
devices:
  - cpu
  - cuda:0
  - cuda:1  # If available
  - metal:0

# Benchmark settings
warmup_iterations: 10
benchmark_iterations: 100

# Profiling options
profile_memory: true
profile_compute: true

# Output settings
output_format: html  # json, csv, or html
output_path: ./benchmarks/multi_device_report.html

# Optional: Performance thresholds
thresholds:
  min_throughput: 100  # samples/sec
  max_latency_p99: 50  # milliseconds