Upload folder using huggingface_hub
Browse files- config.json +206 -0
- model.safetensors +3 -0
- policy_postprocessor.json +32 -0
- policy_postprocessor_step_0_unnormalizer_processor.safetensors +3 -0
- policy_preprocessor.json +86 -0
- policy_preprocessor_step_7_normalizer_processor.safetensors +3 -0
- train_config.json +346 -0
config.json
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"type": "xvla",
|
| 3 |
+
"n_obs_steps": 1,
|
| 4 |
+
"input_features": {
|
| 5 |
+
"observation.state": {
|
| 6 |
+
"type": "STATE",
|
| 7 |
+
"shape": [
|
| 8 |
+
7
|
| 9 |
+
]
|
| 10 |
+
},
|
| 11 |
+
"observation.images.front": {
|
| 12 |
+
"type": "VISUAL",
|
| 13 |
+
"shape": [
|
| 14 |
+
3,
|
| 15 |
+
480,
|
| 16 |
+
640
|
| 17 |
+
]
|
| 18 |
+
}
|
| 19 |
+
},
|
| 20 |
+
"output_features": {
|
| 21 |
+
"action": {
|
| 22 |
+
"type": "ACTION",
|
| 23 |
+
"shape": [
|
| 24 |
+
7
|
| 25 |
+
]
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"device": "cuda",
|
| 29 |
+
"use_amp": false,
|
| 30 |
+
"use_peft": false,
|
| 31 |
+
"push_to_hub": false,
|
| 32 |
+
"repo_id": "albus2024/xvla_long",
|
| 33 |
+
"private": null,
|
| 34 |
+
"tags": null,
|
| 35 |
+
"license": null,
|
| 36 |
+
"pretrained_path": "lerobot/xvla-base",
|
| 37 |
+
"chunk_size": 100,
|
| 38 |
+
"n_action_steps": 100,
|
| 39 |
+
"dtype": "bfloat16",
|
| 40 |
+
"normalization_mapping": {
|
| 41 |
+
"STATE": "IDENTITY",
|
| 42 |
+
"ACTION": "MEAN_STD",
|
| 43 |
+
"VISUAL": "IDENTITY"
|
| 44 |
+
},
|
| 45 |
+
"florence_config": {
|
| 46 |
+
"model_type": "florence2",
|
| 47 |
+
"bos_token_id": 0,
|
| 48 |
+
"eos_token_id": 2,
|
| 49 |
+
"ignore_index": -100,
|
| 50 |
+
"pad_token_id": 1,
|
| 51 |
+
"projection_dim": 1024,
|
| 52 |
+
"text_config": {
|
| 53 |
+
"vocab_size": 51289,
|
| 54 |
+
"activation_dropout": 0.1,
|
| 55 |
+
"activation_function": "gelu",
|
| 56 |
+
"add_bias_logits": false,
|
| 57 |
+
"add_final_layer_norm": false,
|
| 58 |
+
"attention_dropout": 0.1,
|
| 59 |
+
"bos_token_id": 0,
|
| 60 |
+
"classif_dropout": 0.1,
|
| 61 |
+
"classifier_dropout": 0.0,
|
| 62 |
+
"d_model": 1024,
|
| 63 |
+
"decoder_attention_heads": 16,
|
| 64 |
+
"decoder_ffn_dim": 4096,
|
| 65 |
+
"decoder_layerdrop": 0.0,
|
| 66 |
+
"decoder_layers": 12,
|
| 67 |
+
"decoder_start_token_id": 2,
|
| 68 |
+
"dropout": 0.1,
|
| 69 |
+
"early_stopping": true,
|
| 70 |
+
"encoder_attention_heads": 16,
|
| 71 |
+
"encoder_ffn_dim": 4096,
|
| 72 |
+
"encoder_layerdrop": 0.0,
|
| 73 |
+
"encoder_layers": 12,
|
| 74 |
+
"eos_token_id": 2,
|
| 75 |
+
"forced_eos_token_id": 2,
|
| 76 |
+
"forced_bos_token_id": 0,
|
| 77 |
+
"gradient_checkpointing": false,
|
| 78 |
+
"init_std": 0.02,
|
| 79 |
+
"is_encoder_decoder": true,
|
| 80 |
+
"label2id": {
|
| 81 |
+
"LABEL_0": 0,
|
| 82 |
+
"LABEL_1": 1,
|
| 83 |
+
"LABEL_2": 2
|
| 84 |
+
},
|
| 85 |
+
"max_position_embeddings": 4096,
|
| 86 |
+
"no_repeat_ngram_size": 3,
|
| 87 |
+
"normalize_before": false,
|
| 88 |
+
"num_hidden_layers": 12,
|
| 89 |
+
"pad_token_id": 1,
|
| 90 |
+
"scale_embedding": false,
|
| 91 |
+
"num_beams": 3
|
| 92 |
+
},
|
| 93 |
+
"vision_config": {
|
| 94 |
+
"model_type": "davit",
|
| 95 |
+
"drop_path_rate": 0.1,
|
| 96 |
+
"patch_size": [
|
| 97 |
+
7,
|
| 98 |
+
3,
|
| 99 |
+
3,
|
| 100 |
+
3
|
| 101 |
+
],
|
| 102 |
+
"patch_stride": [
|
| 103 |
+
4,
|
| 104 |
+
2,
|
| 105 |
+
2,
|
| 106 |
+
2
|
| 107 |
+
],
|
| 108 |
+
"patch_padding": [
|
| 109 |
+
3,
|
| 110 |
+
1,
|
| 111 |
+
1,
|
| 112 |
+
1
|
| 113 |
+
],
|
| 114 |
+
"patch_prenorm": [
|
| 115 |
+
false,
|
| 116 |
+
true,
|
| 117 |
+
true,
|
| 118 |
+
true
|
| 119 |
+
],
|
| 120 |
+
"enable_checkpoint": false,
|
| 121 |
+
"dim_embed": [
|
| 122 |
+
256,
|
| 123 |
+
512,
|
| 124 |
+
1024,
|
| 125 |
+
2048
|
| 126 |
+
],
|
| 127 |
+
"num_heads": [
|
| 128 |
+
8,
|
| 129 |
+
16,
|
| 130 |
+
32,
|
| 131 |
+
64
|
| 132 |
+
],
|
| 133 |
+
"num_groups": [
|
| 134 |
+
8,
|
| 135 |
+
16,
|
| 136 |
+
32,
|
| 137 |
+
64
|
| 138 |
+
],
|
| 139 |
+
"depths": [
|
| 140 |
+
1,
|
| 141 |
+
1,
|
| 142 |
+
9,
|
| 143 |
+
1
|
| 144 |
+
],
|
| 145 |
+
"window_size": 12,
|
| 146 |
+
"projection_dim": 1024,
|
| 147 |
+
"visual_temporal_embedding": {
|
| 148 |
+
"type": "COSINE",
|
| 149 |
+
"max_temporal_embeddings": 100
|
| 150 |
+
},
|
| 151 |
+
"image_pos_embed": {
|
| 152 |
+
"type": "learned_abs_2d",
|
| 153 |
+
"max_pos_embeddings": 50
|
| 154 |
+
},
|
| 155 |
+
"image_feature_source": [
|
| 156 |
+
"spatial_avg_pool",
|
| 157 |
+
"temporal_avg_pool"
|
| 158 |
+
]
|
| 159 |
+
},
|
| 160 |
+
"vocab_size": 51289,
|
| 161 |
+
"torch_dtype": "float32",
|
| 162 |
+
"is_encoder_decoder": true
|
| 163 |
+
},
|
| 164 |
+
"tokenizer_name": "facebook/bart-large",
|
| 165 |
+
"tokenizer_max_length": 1024,
|
| 166 |
+
"tokenizer_padding_side": "right",
|
| 167 |
+
"pad_language_to": "max_length",
|
| 168 |
+
"hidden_size": 1024,
|
| 169 |
+
"depth": 24,
|
| 170 |
+
"num_heads": 16,
|
| 171 |
+
"mlp_ratio": 4.0,
|
| 172 |
+
"num_domains": 30,
|
| 173 |
+
"len_soft_prompts": 32,
|
| 174 |
+
"dim_time": 32,
|
| 175 |
+
"max_len_seq": 512,
|
| 176 |
+
"use_hetero_proj": false,
|
| 177 |
+
"action_mode": "auto",
|
| 178 |
+
"num_denoising_steps": 10,
|
| 179 |
+
"use_proprio": true,
|
| 180 |
+
"max_state_dim": 20,
|
| 181 |
+
"max_action_dim": 20,
|
| 182 |
+
"domain_feature_key": null,
|
| 183 |
+
"resize_imgs_with_padding": [
|
| 184 |
+
224,
|
| 185 |
+
224
|
| 186 |
+
],
|
| 187 |
+
"num_image_views": 3,
|
| 188 |
+
"empty_cameras": 0,
|
| 189 |
+
"freeze_vision_encoder": false,
|
| 190 |
+
"freeze_language_encoder": false,
|
| 191 |
+
"train_policy_transformer": true,
|
| 192 |
+
"train_soft_prompts": true,
|
| 193 |
+
"optimizer_lr": 0.0001,
|
| 194 |
+
"optimizer_betas": [
|
| 195 |
+
0.9,
|
| 196 |
+
0.95
|
| 197 |
+
],
|
| 198 |
+
"optimizer_eps": 1e-08,
|
| 199 |
+
"optimizer_weight_decay": 0.0001,
|
| 200 |
+
"optimizer_grad_clip_norm": 10.0,
|
| 201 |
+
"optimizer_soft_prompt_lr_scale": 1.0,
|
| 202 |
+
"optimizer_soft_prompt_warmup_lr_scale": null,
|
| 203 |
+
"scheduler_warmup_steps": 1000,
|
| 204 |
+
"scheduler_decay_steps": 30000,
|
| 205 |
+
"scheduler_decay_lr": 2.5e-06
|
| 206 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aa5fd20d2c27012e6d3e2652e4cf80289bb9c916a7620d4d4cfda235a28e56dd
|
| 3 |
+
size 1759596986
|
policy_postprocessor.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"name": "policy_postprocessor",
|
| 3 |
+
"steps": [
|
| 4 |
+
{
|
| 5 |
+
"registry_name": "unnormalizer_processor",
|
| 6 |
+
"config": {
|
| 7 |
+
"eps": 1e-08,
|
| 8 |
+
"features": {
|
| 9 |
+
"action": {
|
| 10 |
+
"type": "ACTION",
|
| 11 |
+
"shape": [
|
| 12 |
+
7
|
| 13 |
+
]
|
| 14 |
+
}
|
| 15 |
+
},
|
| 16 |
+
"norm_map": {
|
| 17 |
+
"STATE": "IDENTITY",
|
| 18 |
+
"ACTION": "MEAN_STD",
|
| 19 |
+
"VISUAL": "IDENTITY"
|
| 20 |
+
}
|
| 21 |
+
},
|
| 22 |
+
"state_file": "policy_postprocessor_step_0_unnormalizer_processor.safetensors"
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"registry_name": "device_processor",
|
| 26 |
+
"config": {
|
| 27 |
+
"device": "cpu",
|
| 28 |
+
"float_dtype": null
|
| 29 |
+
}
|
| 30 |
+
}
|
| 31 |
+
]
|
| 32 |
+
}
|
policy_postprocessor_step_0_unnormalizer_processor.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:06007d2e6c8080bbfb924e38a3a458f67143d2b0f6d80b31b151632ac561f6f7
|
| 3 |
+
size 6632
|
policy_preprocessor.json
ADDED
|
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"name": "policy_preprocessor",
|
| 3 |
+
"steps": [
|
| 4 |
+
{
|
| 5 |
+
"registry_name": "rename_observations_processor",
|
| 6 |
+
"config": {
|
| 7 |
+
"rename_map": {}
|
| 8 |
+
}
|
| 9 |
+
},
|
| 10 |
+
{
|
| 11 |
+
"registry_name": "to_batch_processor",
|
| 12 |
+
"config": {}
|
| 13 |
+
},
|
| 14 |
+
{
|
| 15 |
+
"registry_name": "tokenizer_processor",
|
| 16 |
+
"config": {
|
| 17 |
+
"max_length": 50,
|
| 18 |
+
"task_key": "task",
|
| 19 |
+
"padding_side": "right",
|
| 20 |
+
"padding": "max_length",
|
| 21 |
+
"truncation": true,
|
| 22 |
+
"tokenizer_name": "facebook/bart-large"
|
| 23 |
+
}
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"registry_name": "xvla_add_domain_id",
|
| 27 |
+
"config": {
|
| 28 |
+
"domain_id": 0
|
| 29 |
+
}
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"registry_name": "xvla_image_to_float",
|
| 33 |
+
"config": {
|
| 34 |
+
"image_keys": null,
|
| 35 |
+
"validate_range": true
|
| 36 |
+
}
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"registry_name": "xvla_imagenet_normalize",
|
| 40 |
+
"config": {
|
| 41 |
+
"image_keys": null
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"registry_name": "device_processor",
|
| 46 |
+
"config": {
|
| 47 |
+
"device": "cuda",
|
| 48 |
+
"float_dtype": null
|
| 49 |
+
}
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"registry_name": "normalizer_processor",
|
| 53 |
+
"config": {
|
| 54 |
+
"eps": 1e-08,
|
| 55 |
+
"features": {
|
| 56 |
+
"observation.state": {
|
| 57 |
+
"type": "STATE",
|
| 58 |
+
"shape": [
|
| 59 |
+
7
|
| 60 |
+
]
|
| 61 |
+
},
|
| 62 |
+
"observation.images.front": {
|
| 63 |
+
"type": "VISUAL",
|
| 64 |
+
"shape": [
|
| 65 |
+
3,
|
| 66 |
+
480,
|
| 67 |
+
640
|
| 68 |
+
]
|
| 69 |
+
},
|
| 70 |
+
"action": {
|
| 71 |
+
"type": "ACTION",
|
| 72 |
+
"shape": [
|
| 73 |
+
7
|
| 74 |
+
]
|
| 75 |
+
}
|
| 76 |
+
},
|
| 77 |
+
"norm_map": {
|
| 78 |
+
"STATE": "IDENTITY",
|
| 79 |
+
"ACTION": "MEAN_STD",
|
| 80 |
+
"VISUAL": "IDENTITY"
|
| 81 |
+
}
|
| 82 |
+
},
|
| 83 |
+
"state_file": "policy_preprocessor_step_7_normalizer_processor.safetensors"
|
| 84 |
+
}
|
| 85 |
+
]
|
| 86 |
+
}
|
policy_preprocessor_step_7_normalizer_processor.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:06007d2e6c8080bbfb924e38a3a458f67143d2b0f6d80b31b151632ac561f6f7
|
| 3 |
+
size 6632
|
train_config.json
ADDED
|
@@ -0,0 +1,346 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"dataset": {
|
| 3 |
+
"repo_id": "/home/lyanam/project/lerobot_real_robot/datasets/so101_long",
|
| 4 |
+
"root": null,
|
| 5 |
+
"episodes": null,
|
| 6 |
+
"image_transforms": {
|
| 7 |
+
"enable": false,
|
| 8 |
+
"max_num_transforms": 3,
|
| 9 |
+
"random_order": false,
|
| 10 |
+
"tfs": {
|
| 11 |
+
"brightness": {
|
| 12 |
+
"weight": 1.0,
|
| 13 |
+
"type": "ColorJitter",
|
| 14 |
+
"kwargs": {
|
| 15 |
+
"brightness": [
|
| 16 |
+
0.8,
|
| 17 |
+
1.2
|
| 18 |
+
]
|
| 19 |
+
}
|
| 20 |
+
},
|
| 21 |
+
"contrast": {
|
| 22 |
+
"weight": 1.0,
|
| 23 |
+
"type": "ColorJitter",
|
| 24 |
+
"kwargs": {
|
| 25 |
+
"contrast": [
|
| 26 |
+
0.8,
|
| 27 |
+
1.2
|
| 28 |
+
]
|
| 29 |
+
}
|
| 30 |
+
},
|
| 31 |
+
"saturation": {
|
| 32 |
+
"weight": 1.0,
|
| 33 |
+
"type": "ColorJitter",
|
| 34 |
+
"kwargs": {
|
| 35 |
+
"saturation": [
|
| 36 |
+
0.5,
|
| 37 |
+
1.5
|
| 38 |
+
]
|
| 39 |
+
}
|
| 40 |
+
},
|
| 41 |
+
"hue": {
|
| 42 |
+
"weight": 1.0,
|
| 43 |
+
"type": "ColorJitter",
|
| 44 |
+
"kwargs": {
|
| 45 |
+
"hue": [
|
| 46 |
+
-0.05,
|
| 47 |
+
0.05
|
| 48 |
+
]
|
| 49 |
+
}
|
| 50 |
+
},
|
| 51 |
+
"sharpness": {
|
| 52 |
+
"weight": 1.0,
|
| 53 |
+
"type": "SharpnessJitter",
|
| 54 |
+
"kwargs": {
|
| 55 |
+
"sharpness": [
|
| 56 |
+
0.5,
|
| 57 |
+
1.5
|
| 58 |
+
]
|
| 59 |
+
}
|
| 60 |
+
},
|
| 61 |
+
"affine": {
|
| 62 |
+
"weight": 1.0,
|
| 63 |
+
"type": "RandomAffine",
|
| 64 |
+
"kwargs": {
|
| 65 |
+
"degrees": [
|
| 66 |
+
-5.0,
|
| 67 |
+
5.0
|
| 68 |
+
],
|
| 69 |
+
"translate": [
|
| 70 |
+
0.05,
|
| 71 |
+
0.05
|
| 72 |
+
]
|
| 73 |
+
}
|
| 74 |
+
}
|
| 75 |
+
}
|
| 76 |
+
},
|
| 77 |
+
"revision": null,
|
| 78 |
+
"use_imagenet_stats": true,
|
| 79 |
+
"video_backend": "torchcodec",
|
| 80 |
+
"return_uint8": false,
|
| 81 |
+
"streaming": false
|
| 82 |
+
},
|
| 83 |
+
"env": null,
|
| 84 |
+
"policy": {
|
| 85 |
+
"type": "xvla",
|
| 86 |
+
"n_obs_steps": 1,
|
| 87 |
+
"input_features": {
|
| 88 |
+
"observation.state": {
|
| 89 |
+
"type": "STATE",
|
| 90 |
+
"shape": [
|
| 91 |
+
7
|
| 92 |
+
]
|
| 93 |
+
},
|
| 94 |
+
"observation.images.front": {
|
| 95 |
+
"type": "VISUAL",
|
| 96 |
+
"shape": [
|
| 97 |
+
3,
|
| 98 |
+
480,
|
| 99 |
+
640
|
| 100 |
+
]
|
| 101 |
+
}
|
| 102 |
+
},
|
| 103 |
+
"output_features": {
|
| 104 |
+
"action": {
|
| 105 |
+
"type": "ACTION",
|
| 106 |
+
"shape": [
|
| 107 |
+
7
|
| 108 |
+
]
|
| 109 |
+
}
|
| 110 |
+
},
|
| 111 |
+
"device": "cuda",
|
| 112 |
+
"use_amp": false,
|
| 113 |
+
"use_peft": false,
|
| 114 |
+
"push_to_hub": false,
|
| 115 |
+
"repo_id": "albus2024/xvla_long",
|
| 116 |
+
"private": null,
|
| 117 |
+
"tags": null,
|
| 118 |
+
"license": null,
|
| 119 |
+
"pretrained_path": "lerobot/xvla-base",
|
| 120 |
+
"chunk_size": 100,
|
| 121 |
+
"n_action_steps": 100,
|
| 122 |
+
"dtype": "bfloat16",
|
| 123 |
+
"normalization_mapping": {
|
| 124 |
+
"STATE": "IDENTITY",
|
| 125 |
+
"ACTION": "MEAN_STD",
|
| 126 |
+
"VISUAL": "IDENTITY"
|
| 127 |
+
},
|
| 128 |
+
"florence_config": {
|
| 129 |
+
"model_type": "florence2",
|
| 130 |
+
"bos_token_id": 0,
|
| 131 |
+
"eos_token_id": 2,
|
| 132 |
+
"ignore_index": -100,
|
| 133 |
+
"pad_token_id": 1,
|
| 134 |
+
"projection_dim": 1024,
|
| 135 |
+
"text_config": {
|
| 136 |
+
"vocab_size": 51289,
|
| 137 |
+
"activation_dropout": 0.1,
|
| 138 |
+
"activation_function": "gelu",
|
| 139 |
+
"add_bias_logits": false,
|
| 140 |
+
"add_final_layer_norm": false,
|
| 141 |
+
"attention_dropout": 0.1,
|
| 142 |
+
"bos_token_id": 0,
|
| 143 |
+
"classif_dropout": 0.1,
|
| 144 |
+
"classifier_dropout": 0.0,
|
| 145 |
+
"d_model": 1024,
|
| 146 |
+
"decoder_attention_heads": 16,
|
| 147 |
+
"decoder_ffn_dim": 4096,
|
| 148 |
+
"decoder_layerdrop": 0.0,
|
| 149 |
+
"decoder_layers": 12,
|
| 150 |
+
"decoder_start_token_id": 2,
|
| 151 |
+
"dropout": 0.1,
|
| 152 |
+
"early_stopping": true,
|
| 153 |
+
"encoder_attention_heads": 16,
|
| 154 |
+
"encoder_ffn_dim": 4096,
|
| 155 |
+
"encoder_layerdrop": 0.0,
|
| 156 |
+
"encoder_layers": 12,
|
| 157 |
+
"eos_token_id": 2,
|
| 158 |
+
"forced_eos_token_id": 2,
|
| 159 |
+
"forced_bos_token_id": 0,
|
| 160 |
+
"gradient_checkpointing": false,
|
| 161 |
+
"init_std": 0.02,
|
| 162 |
+
"is_encoder_decoder": true,
|
| 163 |
+
"label2id": {
|
| 164 |
+
"LABEL_0": 0,
|
| 165 |
+
"LABEL_1": 1,
|
| 166 |
+
"LABEL_2": 2
|
| 167 |
+
},
|
| 168 |
+
"max_position_embeddings": 4096,
|
| 169 |
+
"no_repeat_ngram_size": 3,
|
| 170 |
+
"normalize_before": false,
|
| 171 |
+
"num_hidden_layers": 12,
|
| 172 |
+
"pad_token_id": 1,
|
| 173 |
+
"scale_embedding": false,
|
| 174 |
+
"num_beams": 3
|
| 175 |
+
},
|
| 176 |
+
"vision_config": {
|
| 177 |
+
"model_type": "davit",
|
| 178 |
+
"drop_path_rate": 0.1,
|
| 179 |
+
"patch_size": [
|
| 180 |
+
7,
|
| 181 |
+
3,
|
| 182 |
+
3,
|
| 183 |
+
3
|
| 184 |
+
],
|
| 185 |
+
"patch_stride": [
|
| 186 |
+
4,
|
| 187 |
+
2,
|
| 188 |
+
2,
|
| 189 |
+
2
|
| 190 |
+
],
|
| 191 |
+
"patch_padding": [
|
| 192 |
+
3,
|
| 193 |
+
1,
|
| 194 |
+
1,
|
| 195 |
+
1
|
| 196 |
+
],
|
| 197 |
+
"patch_prenorm": [
|
| 198 |
+
false,
|
| 199 |
+
true,
|
| 200 |
+
true,
|
| 201 |
+
true
|
| 202 |
+
],
|
| 203 |
+
"enable_checkpoint": false,
|
| 204 |
+
"dim_embed": [
|
| 205 |
+
256,
|
| 206 |
+
512,
|
| 207 |
+
1024,
|
| 208 |
+
2048
|
| 209 |
+
],
|
| 210 |
+
"num_heads": [
|
| 211 |
+
8,
|
| 212 |
+
16,
|
| 213 |
+
32,
|
| 214 |
+
64
|
| 215 |
+
],
|
| 216 |
+
"num_groups": [
|
| 217 |
+
8,
|
| 218 |
+
16,
|
| 219 |
+
32,
|
| 220 |
+
64
|
| 221 |
+
],
|
| 222 |
+
"depths": [
|
| 223 |
+
1,
|
| 224 |
+
1,
|
| 225 |
+
9,
|
| 226 |
+
1
|
| 227 |
+
],
|
| 228 |
+
"window_size": 12,
|
| 229 |
+
"projection_dim": 1024,
|
| 230 |
+
"visual_temporal_embedding": {
|
| 231 |
+
"type": "COSINE",
|
| 232 |
+
"max_temporal_embeddings": 100
|
| 233 |
+
},
|
| 234 |
+
"image_pos_embed": {
|
| 235 |
+
"type": "learned_abs_2d",
|
| 236 |
+
"max_pos_embeddings": 50
|
| 237 |
+
},
|
| 238 |
+
"image_feature_source": [
|
| 239 |
+
"spatial_avg_pool",
|
| 240 |
+
"temporal_avg_pool"
|
| 241 |
+
]
|
| 242 |
+
},
|
| 243 |
+
"vocab_size": 51289,
|
| 244 |
+
"torch_dtype": "float32",
|
| 245 |
+
"is_encoder_decoder": true
|
| 246 |
+
},
|
| 247 |
+
"tokenizer_name": "facebook/bart-large",
|
| 248 |
+
"tokenizer_max_length": 1024,
|
| 249 |
+
"tokenizer_padding_side": "right",
|
| 250 |
+
"pad_language_to": "max_length",
|
| 251 |
+
"hidden_size": 1024,
|
| 252 |
+
"depth": 24,
|
| 253 |
+
"num_heads": 16,
|
| 254 |
+
"mlp_ratio": 4.0,
|
| 255 |
+
"num_domains": 30,
|
| 256 |
+
"len_soft_prompts": 32,
|
| 257 |
+
"dim_time": 32,
|
| 258 |
+
"max_len_seq": 512,
|
| 259 |
+
"use_hetero_proj": false,
|
| 260 |
+
"action_mode": "auto",
|
| 261 |
+
"num_denoising_steps": 10,
|
| 262 |
+
"use_proprio": true,
|
| 263 |
+
"max_state_dim": 20,
|
| 264 |
+
"max_action_dim": 20,
|
| 265 |
+
"domain_feature_key": null,
|
| 266 |
+
"resize_imgs_with_padding": [
|
| 267 |
+
224,
|
| 268 |
+
224
|
| 269 |
+
],
|
| 270 |
+
"num_image_views": 3,
|
| 271 |
+
"empty_cameras": 0,
|
| 272 |
+
"freeze_vision_encoder": false,
|
| 273 |
+
"freeze_language_encoder": false,
|
| 274 |
+
"train_policy_transformer": true,
|
| 275 |
+
"train_soft_prompts": true,
|
| 276 |
+
"optimizer_lr": 0.0001,
|
| 277 |
+
"optimizer_betas": [
|
| 278 |
+
0.9,
|
| 279 |
+
0.95
|
| 280 |
+
],
|
| 281 |
+
"optimizer_eps": 1e-08,
|
| 282 |
+
"optimizer_weight_decay": 0.0001,
|
| 283 |
+
"optimizer_grad_clip_norm": 10.0,
|
| 284 |
+
"optimizer_soft_prompt_lr_scale": 1.0,
|
| 285 |
+
"optimizer_soft_prompt_warmup_lr_scale": null,
|
| 286 |
+
"scheduler_warmup_steps": 1000,
|
| 287 |
+
"scheduler_decay_steps": 30000,
|
| 288 |
+
"scheduler_decay_lr": 2.5e-06
|
| 289 |
+
},
|
| 290 |
+
"reward_model": null,
|
| 291 |
+
"output_dir": "outputs/xvla_training_long",
|
| 292 |
+
"job_name": "xvla_training",
|
| 293 |
+
"resume": false,
|
| 294 |
+
"seed": 1000,
|
| 295 |
+
"cudnn_deterministic": false,
|
| 296 |
+
"num_workers": 4,
|
| 297 |
+
"batch_size": 32,
|
| 298 |
+
"prefetch_factor": 4,
|
| 299 |
+
"persistent_workers": true,
|
| 300 |
+
"steps": 30000,
|
| 301 |
+
"eval_freq": 20000,
|
| 302 |
+
"log_freq": 200,
|
| 303 |
+
"tolerance_s": 0.0001,
|
| 304 |
+
"save_checkpoint": true,
|
| 305 |
+
"save_freq": 10000,
|
| 306 |
+
"use_policy_training_preset": true,
|
| 307 |
+
"optimizer": {
|
| 308 |
+
"type": "xvla-adamw",
|
| 309 |
+
"lr": 0.0001,
|
| 310 |
+
"weight_decay": 0.0001,
|
| 311 |
+
"grad_clip_norm": 10.0,
|
| 312 |
+
"betas": [
|
| 313 |
+
0.9,
|
| 314 |
+
0.95
|
| 315 |
+
],
|
| 316 |
+
"eps": 1e-08,
|
| 317 |
+
"soft_prompt_lr_scale": 1.0,
|
| 318 |
+
"soft_prompt_warmup_lr_scale": null
|
| 319 |
+
},
|
| 320 |
+
"scheduler": {
|
| 321 |
+
"type": "cosine_decay_with_warmup",
|
| 322 |
+
"num_warmup_steps": 1000,
|
| 323 |
+
"num_decay_steps": 30000,
|
| 324 |
+
"peak_lr": 0.0001,
|
| 325 |
+
"decay_lr": 2.5e-06
|
| 326 |
+
},
|
| 327 |
+
"eval": {
|
| 328 |
+
"n_episodes": 50,
|
| 329 |
+
"batch_size": 50,
|
| 330 |
+
"use_async_envs": true
|
| 331 |
+
},
|
| 332 |
+
"wandb": {
|
| 333 |
+
"enable": false,
|
| 334 |
+
"disable_artifact": false,
|
| 335 |
+
"project": "lerobot",
|
| 336 |
+
"entity": null,
|
| 337 |
+
"notes": null,
|
| 338 |
+
"run_id": null,
|
| 339 |
+
"mode": null,
|
| 340 |
+
"add_tags": true
|
| 341 |
+
},
|
| 342 |
+
"peft": null,
|
| 343 |
+
"sample_weighting": null,
|
| 344 |
+
"rename_map": {},
|
| 345 |
+
"checkpoint_path": null
|
| 346 |
+
}
|