johndpope
commented
3 months ago https://talkshow.is.tue.mpg.de/login.php
I downloaded a 16gb file 16G May 14 17:48 SHOW_dataset_v1.0.zip
but it doesn't include this file...
what's the specific download I need?
JeremyCJM
(diffsheg) ➜ DiffSHEG git:(main) ✗ ./inference_custom_audio_show.sh ------------ Options ------------- ExprID_off: False ExprID_off_uncond: False PE: pe_sinu ablation: None addBlend: True addEmoCond: False addHubert: True addTextCond: False addWav2Vec2: False add_vel_loss: True audio_dim: 128 audio_feat: mel audio_latent_dim: 256 axis_angle: True batch_size: 32 beat_cache_name: beat_4english_15_141 checkpoints_dir: ./checkpoints ckpt: ckpt_e2599.tar classifier_free: True cond_projection: mlp_includeX cond_residual: True cond_scale: 1.15 data_mode: original data_parallel: False data_root: ./Freeform/processed_data_200 data_sel: upperbody data_type: pos dataset_name: talkshow ddim: True debug: False decomp_name: Decomp_SP001_SM001_H512 diffusion_steps: 1000 dim_att_vec: 512 dim_dec_hidden: 1024 dim_movement_dec_hidden: 512 dim_movement_enc_hidden: 512 dim_movement_latent: 512 dim_pos_hidden: 1024 dim_pri_hidden: 1024 dim_text_hidden: 512 dim_z: 128 dist_backend: nccl dist_url: tcp://224.66.41.62:23456 distributed: False embed_net_path: feature_extractor/gesture_autoencoder_checkpoint_best.bin encode_hubert: True encode_wav2vec2: False eval_every_e: 5 expAddHubert: False expCondition_gesture_only: None expr_weight: 1 expression_only: False feat_bias: 5 fix_head_var: False fix_very_first: False gesCondition_expression_only: False gesture_only: False gpu_id: None is_train: True jump_length: 3 jump_n_sample: 5 latent_dim: 512 log_every: 50 lr: 0.0002 max_eval_samples: -1 mode: test_custom_audio model_base: transformer_encoder model_mean_type: epsilon multiprocessing_distributed: False n_layers_dec: 1 n_layers_pos: 1 n_layers_pri: 1 n_poses: 88 name: talkshow_GesExpr_unify_addHubert_encodeHubert_mdlpIncludeX_condRes_LN_ClsFree no_clip: False no_eff: False no_fgd: False no_repaint: False no_resample: False no_style: False null_cond_prob: 0.2 num_epochs: 5000 num_layers: 8 output_gt: False overlap_len: 10 rank: 0 rebuttal: None remove_audio: False remove_hand: False remove_style: False rename: None reset_lr: False resume: False same_overlap_noisy: False save_every_e: 5 save_latest: 500 separate: None separate_pure: False style_dim: 4 test_audio_path: audios/Forrest_tts.wav test_on_trainset: False test_on_val: False textExpEmoCondition_gesture_only: False times: 1 timestep_respacing: ddim25 unidiffuser: True usePredExpr: None use_aud_feat: None use_joints: False use_single_style: False vel_interval: 10 vel_loss_start: -1 visualize_unify_x0_step: None workers: 0 world_size: 1 -------------- End ---------------- => creating model 'transformer_encoder' init 'HalfEmbeddingNet' success Traceback (most recent call last): File "/media/oem/12TB/DiffSHEG/runner.py", line 327, in
main()
File "/media/oem/12TB/DiffSHEG/runner.py", line 86, in main
main_worker(opt.gpu_id, ngpus_per_node, opt)
File "/media/oem/12TB/DiffSHEG/runner.py", line 312, in main_worker
test_dataset = ShowDataset(opt, 'data/SHOW/cached_data/talkshow_test_cache')
File "/media/oem/12TB/DiffSHEG/datasets/show.py", line 43, in init
mean_std_dict = np.load("data/SHOW/talkshow_mean_std.npy", allow_pickle=True)[()]
File "/home/oem/miniconda3/envs/diffsheg/lib/python3.9/site-packages/numpy/lib/npyio.py", line 427, in load
fid = stack.enter_context(open(os_fspath(file), "rb"))
FileNotFoundError: [Errno 2] No such file or directory: 'data/SHOW/talkshow_mean_std.npy'