Hugging Face's logo

Spaces:
joeyaintjoking
/
ml_dart_model
Runtime error

App Files Community
ml_dart_model
File size: 3,686 Bytes 
4aea73a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
 
model:
  names:
  - ft_transformer
  - fusion_mlp
  - hf_text
  hf_text:
    checkpoint_name: local://hf_text
    gradient_checkpointing: false
    pooling_mode: cls
    data_types:
    - text
    tokenizer_name: hf_auto
    use_fast: true
    max_text_len: 512
    insert_sep: true
    low_cpu_mem_usage: false
    text_segment_num: 2
    stochastic_chunk: false
    text_aug_detect_length: 10
    text_trivial_aug_maxscale: 0.1
    text_train_augment_types: null
  fusion_mlp:
    aux_loss_weight: null
    adapt_in_features: max
    hidden_sizes:
    - 128
    activation: leaky_relu
    dropout: 0.1
    normalization: layer_norm
    data_types: null
  ft_transformer:
    data_types:
    - numerical
    embedding_arch:
    - linear
    token_dim: 192
    hidden_size: 192
    num_blocks: 3
    attention_num_heads: 8
    attention_dropout: 0.2
    residual_dropout: 0.0
    ffn_dropout: 0.1
    ffn_hidden_size: 192
    ffn_activation: geglu
    head_activation: relu
    normalization: layer_norm
    merge: concat
    requires_all_dtypes: false
    additive_attention: false
    share_qv_weights: false
    pooling_mode: cls
    checkpoint_name: null
data:
  image:
    missing_value_strategy: zero
  text:
    normalize_text: false
  categorical:
    minimum_cat_count: 100
    maximum_num_cat: 20
    convert_to_text: false
    convert_to_text_template: latex
  numerical:
    convert_to_text: false
    scaler_with_mean: true
    scaler_with_std: true
  document:
    missing_value_strategy: zero
  label:
    numerical_preprocessing: standardscaler
  pos_label: null
  column_features_pooling_mode: concat
  mixup:
    turn_on: false
    mixup_alpha: 0.8
    cutmix_alpha: 1.0
    cutmix_minmax: null
    prob: 1.0
    switch_prob: 0.5
    mode: batch
    turn_off_epoch: 5
    label_smoothing: 0.1
  modality_dropout: 0
  templates:
    turn_on: false
    num_templates: 30
    template_length: 2048
    preset_templates:
    - super_glue
    - rte
    custom_templates: null
optim:
  optim_type: adamw
  lr: 0.0001
  weight_decay: 0.001
  lr_choice: layerwise_decay
  lr_decay: 0.9
  lr_schedule: cosine_decay
  max_epochs: 20
  max_steps: -1
  warmup_steps: 0.1
  end_lr: 0
  lr_mult: 1
  patience: 10
  val_check_interval: 0.5
  check_val_every_n_epoch: 1
  skip_final_val: false
  gradient_clip_val: 1
  gradient_clip_algorithm: norm
  track_grad_norm: -1
  log_every_n_steps: 10
  label_smoothing: 0
  top_k: 3
  top_k_average_method: greedy_soup
  peft: null
  lora:
    module_filter: null
    filter:
    - query
    - value
    - ^q$
    - ^v$
    - ^k$
    - ^o$
    r: 8
    alpha: 8
    conv_lora_expert_num: 8
  loss_func: auto
  focal_loss:
    alpha: null
    gamma: 2.0
    reduction: mean
  mask2former_loss:
    loss_cross_entropy_weight: 10.0
    loss_mask_weight: 5.0
    loss_dice_weight: 5.0
  extra_trainable_params: []
  cross_modal_align: null
  cross_modal_align_weight: 0
  automatic_optimization: true
  lemda:
    turn_on: false
    arch_type: mlp_vae
    z_dim: 8
    num_layers: 6
    kld_weight: 0.1
    mse_weight: 0.1
    adv_weight: 0.0001
    consist_weight: 0.01
    consist_threshold: 0.5
    lr: 0.0001
    optim_type: adamw
    weight_decay: 1.0e-05
env:
  num_gpus: 2
  num_nodes: 1
  batch_size: 128
  per_gpu_batch_size: 8
  inference_batch_size_ratio: 4
  precision: 16-mixed
  num_workers: 2
  num_workers_inference: 2
  accelerator: auto
  fast_dev_run: false
  deterministic: false
  auto_select_gpus: true
  strategy: ddp_fork_find_unused_parameters_true
  deepspeed_allgather_size: 1000000000.0
  deepspeed_allreduce_size: 1000000000.0
  compile:
    turn_on: false
    mode: default
    dynamic: true
    backend: inductor