Upload folder using huggingface_hub

InternVLPI-0318-eep-pretrain_T2_sft_touch_21000/checkpoints/steps_21000_pytorch_model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2751f3962975e4d642086d0dbee55c00e1d24a913697432eba99b43bd6412e64
+size 2966606180

InternVLPI-0318-eep-pretrain_T2_sft_touch_21000/config.yaml

@@ -0,0 +1,68 @@
+datasets:
+  vla_data:
+    data_mix: startouch
+    data_root_dir: /lumos-vePFS/shenzhen/data/data_warehouse/data_warehouse_touch
+    dataset_py: lerobot_datasets
+    image_size: 256
+    num_workers: 4
+    per_device_batch_size: 16
+    video_backend: pyav
+framework:
+  action_model:
+    action_dim: 20
+    add_pos_embed: true
+    diffusion_model_cfg:
+      dropout: 0.2
+      final_dropout: true
+      interleave_self_attention: true
+      norm_type: ada_norm
+      num_layers: 28
+      output_dim: 2560
+      positional_embeddings: null
+    future_action_window_size: 29
+    hidden_dim: 1024
+    max_seq_len: 1024
+    noise_beta_alpha: 1.5
+    noise_beta_beta: 1.0
+    noise_s: 0.999
+    num_inference_timesteps: 8
+    num_target_vision_tokens: 32
+    num_timestep_buckets: 1000
+    past_action_window_size: 0
+    smoothness_loss_weight: 1
+    state_dim: 20
+    use_kv_cache_dit: true
+    use_scaled_noise: true
+  name: internvlPI
+  qwenvl:
+    attn_implementation: eager
+    base_vlm: /lumos-vePFS/shenzhen/models/InternVL3_5-1B-Instruct
+    num_vl_layers: 28
+    vl_hidden_dim: 1024
+output_dir: /lumos-vePFS/shenzhen/models/lumos_pi/InternVLPI-0318-eep-pretrain_T2_sft_touch
+run_id: InternVLPI-0318-eep-pretrain_T2_sft_touch
+run_root_dir: /lumos-vePFS/shenzhen/models/lumos_pi
+seed: 42
+trainer:
+  eval_interval: 10000000
+  freeze_modules: null
+  gradient_accumulation_steps: 3
+  gradient_clipping: 1.0
+  learning_rate:
+    action_model: 0.0001
+    base: 1.0e-05
+    intern_vl_interface: 1.0e-05
+  logging_frequency: 10
+  lr_scheduler_type: cosine_with_min_lr
+  max_train_steps: 200000
+  num_warmup_steps: 5000
+  optimizer:
+    betas:
+    - 0.9
+    - 0.95
+    eps: 1.0e-08
+    weight_decay: 0
+  pretrained_checkpoint: /lumos-vePFS/shenzhen/models/lumos_pi/InternVLPI-0310-eep-pretrain_TStest_gripper_fix_no_alpha_stable/checkpoints/steps_81500_pytorch_model.pt
+  save_interval: 1000
+  scheduler_specific_kwargs:
+    min_lr: 1.0e-06

InternVLPI-0318-eep-pretrain_T2_sft_touch_21000/dataset_statistics.json

@@ -0,0 +1,218 @@
+{
+  "new_embodiment": {
+    "action": {
+      "mean": [
+        0.3061756193637848,
+        -0.09595837444067001,
+        0.18800029158592224,
+        -1.2036051750183105,
+        0.8906517624855042,
+        -0.07171571999788284,
+        0.045723315328359604,
+        0.31182044744491577,
+        0.10779045522212982,
+        0.18970179557800293,
+        0.33844712376594543,
+        0.9169597625732422,
+        -1.4967726469039917,
+        0.0458984449505806
+      ],
+      "std": [
+        0.053043268620967865,
+        0.10675930976867676,
+        0.03928906098008156,
+        2.502779245376587,
+        0.4669472575187683,
+        2.704206705093384,
+        0.008583752438426018,
+        0.05179132521152496,
+        0.1147320419549942,
+        0.04499182850122452,
+        2.7472758293151855,
+        0.40838322043418884,
+        2.2489378452301025,
+        0.009146665222942829
+      ],
+      "max": [
+        0.48259955644607544,
+        0.06450944393873215,
+        0.29384109377861023,
+        3.1415865421295166,
+        1.5705276727676392,
+        3.1415348052978516,
+        0.049341894686222076,
+        0.5161705017089844,
+        0.4711654782295227,
+        0.28638094663619995,
+        3.1415677070617676,
+        1.5686907768249512,
+        3.14151930809021,
+        0.04957195371389389
+      ],
+      "min": [
+        0.17984026670455933,
+        -0.3791758418083191,
+        0.07392188161611557,
+        -3.1415746212005615,
+        -0.23330658674240112,
+        -3.141547203063965,
+        0.02287493832409382,
+        0.19148361682891846,
+        -0.035632796585559845,
+        0.06067722290754318,
+        -3.1415812969207764,
+        -0.017380600795149803,
+        -3.141544818878174,
+        0.0159837007522583
+      ],
+      "q01": [
+        0.22693637013435364,
+        -0.3524001133441925,
+        0.08320557430386544,
+        -3.1340175437927247,
+        0.01365514937788248,
+        -3.1312057113647462,
+        0.02368013560771942,
+        0.23438963130116464,
+        -0.01688349975273013,
+        0.07071292966604233,
+        -3.1332657647132875,
+        0.11979441478848457,
+        -3.128193700313568,
+        0.021442314609885216
+      ],
+      "q99": [
+        0.44192082911729813,
+        0.037914081923663606,
+        0.2743035826086998,
+        3.1349051451683043,
+        1.5705275535583496,
+        3.1305520725250244,
+        0.049341894686222076,
+        0.4372120979428291,
+        0.4155946430563926,
+        0.25710681200027463,
+        3.13244158744812,
+        1.5648791790008545,
+        3.1239338278770448,
+        0.04956149682402611
+      ],
+      "mask": [
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        false,
+        true,
+        true,
+        true,
+        true,
+        true,
+        true,
+        false
+      ]
+    },
+    "state": {
+      "mean": [
+        0.3061756193637848,
+        -0.09595837444067001,
+        0.18800029158592224,
+        -1.2036051750183105,
+        0.8906517624855042,
+        -0.07171571999788284,
+        0.045723315328359604,
+        0.31182044744491577,
+        0.10779045522212982,
+        0.18970179557800293,
+        0.33844712376594543,
+        0.9169597625732422,
+        -1.4967726469039917,
+        0.0458984449505806
+      ],
+      "std": [
+        0.053043268620967865,
+        0.10675930976867676,
+        0.03928906098008156,
+        2.502779245376587,
+        0.4669472575187683,
+        2.704206705093384,
+        0.008583752438426018,
+        0.05179132521152496,
+        0.1147320419549942,
+        0.04499182850122452,
+        2.7472758293151855,
+        0.40838322043418884,
+        2.2489378452301025,
+        0.009146665222942829
+      ],
+      "max": [
+        0.48259955644607544,
+        0.06450944393873215,
+        0.29384109377861023,
+        3.1415865421295166,
+        1.5705276727676392,
+        3.1415348052978516,
+        0.049341894686222076,
+        0.5161705017089844,
+        0.4711654782295227,
+        0.28638094663619995,
+        3.1415677070617676,
+        1.5686907768249512,
+        3.14151930809021,
+        0.04957195371389389
+      ],
+      "min": [
+        0.17984026670455933,
+        -0.3791758418083191,
+        0.07392188161611557,
+        -3.1415746212005615,
+        -0.23330658674240112,
+        -3.141547203063965,
+        0.02287493832409382,
+        0.19148361682891846,
+        -0.035632796585559845,
+        0.06067722290754318,
+        -3.1415812969207764,
+        -0.017380600795149803,
+        -3.141544818878174,
+        0.0159837007522583
+      ],
+      "q01": [
+        0.22693637013435364,
+        -0.3524001133441925,
+        0.08320557430386544,
+        -3.1340175437927247,
+        0.01365514937788248,
+        -3.1312057113647462,
+        0.02368013560771942,
+        0.23438963130116464,
+        -0.01688349975273013,
+        0.07071292966604233,
+        -3.1332657647132875,
+        0.11979441478848457,
+        -3.128193700313568,
+        0.021442314609885216
+      ],
+      "q99": [
+        0.44192082911729813,
+        0.037914081923663606,
+        0.2743035826086998,
+        3.1349051451683043,
+        1.5705275535583496,
+        3.1305520725250244,
+        0.049341894686222076,
+        0.4372120979428291,
+        0.4155946430563926,
+        0.25710681200027463,
+        3.13244158744812,
+        1.5648791790008545,
+        3.1239338278770448,
+        0.04956149682402611
+      ]
+    },
+    "num_transitions": 3897,
+    "num_trajectories": 115
+  }
+}