↓ 4 callersMethod__init__(self,
in_channels=[512, 1024, 1024],
out_channels=[256, 512, 1024])
modeling/vl_model/layers.py:260
↓ 3 callersFunctionidentify_correct_shelf(commands, metadata, instr, predicted_object, unique_object_type_dict, bot_position, best_object_match_index,
arena_wrapper/util/object_class_decoder.py:506
↓ 2 callersFunctioncollapse_classes collapse_dict = { "Table": ["Table", "Desk", "Counter"], "Apple": ["Apple", "Apple Slice"] } means collapse ["Table", "De
modeling/vision_model/prepare_clean_data.py:193
↓ 2 callersFunctionpredict_action_and_object_e2e_vl(vl_model, images_list, instruction, eos, ht_1, ct_1, id2act, id2obj, device)
modeling/inference/models/vl_model.py:53