Add files using upload-large-folder tool
Browse files- .gitattributes +2 -0
- MGRKD/mgrkd_distance_angle_flatten_action_encoder/default/phase2/model-00001-of-00002.safetensors +3 -0
- MGRKD/mgrkd_distance_angle_flatten_action_encoder/default/phase2/model-00002-of-00002.safetensors +3 -0
- MGRKD/mgrkd_distance_angle_flatten_action_encoder/default/phase2/training_args.bin +3 -0
- MGRKD/mgrkd_distance_angle_flatten_raw_action/default/phase2/model-00001-of-00002.safetensors +3 -0
- MGRKD/mgrkd_distance_angle_flatten_raw_action/default/phase2/model-00002-of-00002.safetensors +3 -0
- MGRKD/mgrkd_distance_angle_flatten_raw_action/default/phase2/trainer_state.json +0 -0
- MGRKD/mgrkd_distance_angle_flatten_raw_action/default/phase2/training_args.bin +3 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/config.json +77 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/model-00001-of-00002.safetensors +3 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/model-00002-of-00002.safetensors +3 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/optimizer.pt +3 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/rng_state.pth +3 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/scheduler.pt +3 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/trainer_state.json +3 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/model-00001-of-00002.safetensors +3 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/model-00002-of-00002.safetensors +3 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/trainer_state.json +3 -0
- rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/training_args.bin +3 -0
- rkd_v2_2/launch_logs/action_encoder_flatten_gpu4_5_bs32_ngpu2.pid +1 -0
.gitattributes
CHANGED
|
@@ -37,3 +37,5 @@ RKD_Processing_line_Only/train.log filter=lfs diff=lfs merge=lfs -text
|
|
| 37 |
VLM_Only/RKD_Raw_VLM/train.log filter=lfs diff=lfs merge=lfs -text
|
| 38 |
VLM_Only/MGD_Raw_VLM/train.log filter=lfs diff=lfs merge=lfs -text
|
| 39 |
experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_gpu01_lane.log filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
| 37 |
VLM_Only/RKD_Raw_VLM/train.log filter=lfs diff=lfs merge=lfs -text
|
| 38 |
VLM_Only/MGD_Raw_VLM/train.log filter=lfs diff=lfs merge=lfs -text
|
| 39 |
experiment_cfg/processing_line_only_v2/MGD_v2/mgd_v2_mask_ratio_gpu01_lane.log filter=lfs diff=lfs merge=lfs -text
|
| 40 |
+
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/trainer_state.json filter=lfs diff=lfs merge=lfs -text
|
| 41 |
+
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/trainer_state.json filter=lfs diff=lfs merge=lfs -text
|
MGRKD/mgrkd_distance_angle_flatten_action_encoder/default/phase2/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98cf98087479b3e570d7fc66a0d397f42423009d07149e2cac936dbf1290a117
|
| 3 |
+
size 4999367032
|
MGRKD/mgrkd_distance_angle_flatten_action_encoder/default/phase2/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a0f28fe9c881a8cffb1ee8fe2c79dd0794c8acd2049ceda44b73eb00160b1453
|
| 3 |
+
size 2643339852
|
MGRKD/mgrkd_distance_angle_flatten_action_encoder/default/phase2/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:14c635c36e20ae1998985a3b966636f01ee76cd6352938403953a9dacf8c9284
|
| 3 |
+
size 5969
|
MGRKD/mgrkd_distance_angle_flatten_raw_action/default/phase2/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98cf98087479b3e570d7fc66a0d397f42423009d07149e2cac936dbf1290a117
|
| 3 |
+
size 4999367032
|
MGRKD/mgrkd_distance_angle_flatten_raw_action/default/phase2/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b96955bfa63dff0f5736220ca44621fe301400ca2814f26ada7edb4fc4926a0c
|
| 3 |
+
size 2660191348
|
MGRKD/mgrkd_distance_angle_flatten_raw_action/default/phase2/trainer_state.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
MGRKD/mgrkd_distance_angle_flatten_raw_action/default/phase2/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b677c8c36ac6dd461856c22cc9a39214ff88ad0315480673d8e1d4e81c50fe65
|
| 3 |
+
size 5969
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/config.json
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"action_dim": 32,
|
| 3 |
+
"action_head_cfg": {
|
| 4 |
+
"action_dim": 32,
|
| 5 |
+
"action_horizon": 16,
|
| 6 |
+
"add_pos_embed": true,
|
| 7 |
+
"backbone_embedding_dim": 2048,
|
| 8 |
+
"diffusion_model_cfg": {
|
| 9 |
+
"attention_head_dim": 48,
|
| 10 |
+
"cross_attention_dim": 2048,
|
| 11 |
+
"dropout": 0.2,
|
| 12 |
+
"final_dropout": true,
|
| 13 |
+
"interleave_self_attention": true,
|
| 14 |
+
"norm_type": "ada_norm",
|
| 15 |
+
"num_attention_heads": 32,
|
| 16 |
+
"num_layers": 16,
|
| 17 |
+
"output_dim": 1024,
|
| 18 |
+
"positional_embeddings": null
|
| 19 |
+
},
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"input_embedding_dim": 1536,
|
| 22 |
+
"max_action_dim": 32,
|
| 23 |
+
"max_state_dim": 64,
|
| 24 |
+
"model_dtype": "float32",
|
| 25 |
+
"noise_beta_alpha": 1.5,
|
| 26 |
+
"noise_beta_beta": 1.0,
|
| 27 |
+
"noise_s": 0.999,
|
| 28 |
+
"num_inference_timesteps": 4,
|
| 29 |
+
"num_target_vision_tokens": 32,
|
| 30 |
+
"num_timestep_buckets": 1000,
|
| 31 |
+
"tune_diffusion_model": true,
|
| 32 |
+
"tune_projector": true,
|
| 33 |
+
"use_vlln": true,
|
| 34 |
+
"vl_self_attention_cfg": {
|
| 35 |
+
"attention_head_dim": 64,
|
| 36 |
+
"dropout": 0.2,
|
| 37 |
+
"final_dropout": true,
|
| 38 |
+
"num_attention_heads": 32,
|
| 39 |
+
"num_layers": 4,
|
| 40 |
+
"positional_embeddings": null
|
| 41 |
+
}
|
| 42 |
+
},
|
| 43 |
+
"action_horizon": 16,
|
| 44 |
+
"architectures": [
|
| 45 |
+
"GR00T_N1_5_RKD"
|
| 46 |
+
],
|
| 47 |
+
"attn_implementation": null,
|
| 48 |
+
"backbone_cfg": {
|
| 49 |
+
"eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops",
|
| 50 |
+
"load_bf16": false,
|
| 51 |
+
"project_to_dim": null,
|
| 52 |
+
"reproject_vision": false,
|
| 53 |
+
"select_layer": 12,
|
| 54 |
+
"tune_llm": false,
|
| 55 |
+
"tune_visual": true,
|
| 56 |
+
"use_flash_attention": true
|
| 57 |
+
},
|
| 58 |
+
"compute_dtype": "bfloat16",
|
| 59 |
+
"hidden_size": 2048,
|
| 60 |
+
"model_dtype": "float32",
|
| 61 |
+
"model_type": "gr00t_n1_5",
|
| 62 |
+
"rkd_action_temp": 0.1,
|
| 63 |
+
"rkd_angle_loss_weight": 2.0,
|
| 64 |
+
"rkd_distance_loss_weight": 1.0,
|
| 65 |
+
"rkd_enabled": true,
|
| 66 |
+
"rkd_exclude_diagonal": true,
|
| 67 |
+
"rkd_fm_loss_weight": 1.0,
|
| 68 |
+
"rkd_loss_type": "distance_angle",
|
| 69 |
+
"rkd_loss_weight": 0.0,
|
| 70 |
+
"rkd_loss_weight_end": 0.0,
|
| 71 |
+
"rkd_loss_weight_schedule": "cosine_decay",
|
| 72 |
+
"rkd_loss_weight_start": 1.0,
|
| 73 |
+
"rkd_relation_mode": "token_pair_mean",
|
| 74 |
+
"rkd_vlm_temp": 0.07,
|
| 75 |
+
"torch_dtype": "bfloat16",
|
| 76 |
+
"transformers_version": "4.51.3"
|
| 77 |
+
}
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f7df31ab3578b7e17bf5ed86eebefb092136fe2d9ef6f4bf6ded14bd737eec26
|
| 3 |
+
size 4999367032
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd000e03c20de46e02b50c40bbd73528adbd9b3e1c97f50482c6a14f5f93a324
|
| 3 |
+
size 2586705312
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5a3b708858365d1e3ac22717a5f056d76b9d74797ba9d5e73b4403deac1e1b49
|
| 3 |
+
size 8550719959
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b8231c0964284a4afc7bd970e811e4b6900c7976cedef2246d0a7f73c61c3353
|
| 3 |
+
size 14645
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:48f412abc82c519159c43890e65e38c5467c83b8c0467a1cc8ef436b8360d71f
|
| 3 |
+
size 1465
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/checkpoint-120000/trainer_state.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:17c0e27d37f9998049fd597fb231f4f9fd418cb9863c81fade26d61866e8c792
|
| 3 |
+
size 11163983
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f7df31ab3578b7e17bf5ed86eebefb092136fe2d9ef6f4bf6ded14bd737eec26
|
| 3 |
+
size 4999367032
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd000e03c20de46e02b50c40bbd73528adbd9b3e1c97f50482c6a14f5f93a324
|
| 3 |
+
size 2586705312
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/trainer_state.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:211893ac0260dc95ebf66bdac349b4aa459faa040d04960792b918a02cbf882a
|
| 3 |
+
size 11164235
|
rkd_v2_1/rkd_v2_1_distance_angle_full_120k_cosine_decay/default/full_fm_rkd_da_cosine_decay_120k/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:738964e12da6d59086f611cfa744c4de5ef327f3a3ca83637ff71053261861ff
|
| 3 |
+
size 6033
|
rkd_v2_2/launch_logs/action_encoder_flatten_gpu4_5_bs32_ngpu2.pid
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
2402009
|