reference_data.yaml

tinyllama_backend_FP32:
  metric_value: 1.0
  num_int4: 0
  num_int8: 0
  num_sparse_activations: 0
tinyllama_ffn_sparse20_backend_CUDA_TORCH:
  metric_value: 0.7818
  atol: 0.025
  num_int4: 0
  num_int8: 0
  num_sparse_activations: 44
tinyllama_ffn_sparse20_backend_TORCH:
  metric_value: 0.7879
  atol: 0.025
  num_int4: 0
  num_int8: 0
  num_sparse_activations: 44
tinyllama_int8_asym_data_free_ffn_sparse20_backend_CUDA_TORCH:
  metric_value: 0.8044
  atol: 0.025
  num_int4: 0
  num_int8: 312
  num_sparse_activations: 44
tinyllama_int8_asym_data_free_ffn_sparse20_backend_TORCH:
  metric_value: 0.7846
  atol: 0.030
  num_int4: 0
  num_int8: 312
  num_sparse_activations: 44
timm/deit3_small_patch16_224_backend_FP32:
  metric_value: 0.8135
  atol: 0.001
  num_int4: 0
  num_int8: 0
  num_sparse_activations: 0
timm/deit3_small_patch16_224_qkv_sparse20_fc1_sparse20_fc2_sparse30_backend_CUDA_TORCH:
  metric_value: 0.8102
  atol: 0.001
  num_int4: 0
  num_int8: 0
  num_sparse_activations: 36
timm/deit3_small_patch16_224_qkv_sparse20_fc1_sparse20_fc2_sparse30_backend_TORCH:
  metric_value: 0.8102
  atol: 0.001
  num_int4: 0
  num_int8: 0
  num_sparse_activations: 36