MCPcopy Create free account

hub / github.com/ZZongzheng0918/TA-VLA / functions

Functions459 in github.com/ZZongzheng0918/TA-VLA

↓ 13 callersMethodinit
Convenience method for initializing all parameters, necessary due to the quirks of linen.
src/openpi/models/gemma.py:390
↓ 13 callersFunctiontransform
(k: str, v: T)
src/openpi/transforms.py:383
↓ 12 callersMethodupdate
Update the running statistics with a batch of vectors. Args: vectors (np.ndarray): A 2D array where each row is a new ve
src/openpi/shared/normalize.py:30
↓ 9 callersMethodcreate
(self, rng: at.KeyArrayLike)
src/openpi/models/pi0.py:88
↓ 8 callersFunction_name
(name, i)
src/openpi/models/gemma.py:419
↓ 7 callersMethodencode
(self, x)
src/openpi/models/gemma.py:140
↓ 7 callersMethodfake_obs
(self, batch_size: int = 1)
src/openpi/models/model.py:260
↓ 6 callersMethodget_freeze_filter
Returns the freeze filter based on the model config.
src/openpi/models/pi0.py:117
↓ 5 callersMethodfake_act
(self, batch_size: int = 1)
src/openpi/models/model.py:264
↓ 5 callersFunctionflatten_dict
Flatten a nested dictionary. Uses '/' as the separator.
src/openpi/transforms.py:313
↓ 5 callersMethodload
Loads the model weights. Args: params: Parameters of the model. This is a nested structure of array-like objects that
src/openpi/training/weight_loaders.py:18
↓ 5 callersMethodload
Create a model with the given parameters.
src/openpi/models/model.py:246
↓ 5 callersMethodsave
(self, directory: epath.Path, args: "CallbackSave")
src/openpi/training/checkpoints.py:136
↓ 5 callersMethodstep
(self, action)
examples/aloha_real/real_env.py:145
↓ 4 callersFunction_get_frozen_state
(config: _pi0.Pi0Config)
src/openpi/models/pi0_test.py:7
↓ 4 callersMethodcreate_base_config
(self, assets_dirs: pathlib.Path)
src/openpi/training/config.py:166
↓ 4 callersFunctionget_arm_gripper_positions
(bot)
examples/aloha_real/robot_utils.py:176
↓ 4 callersMethodimage_cb
(self, cam_name, data)
examples/aloha_real/robot_utils.py:48
↓ 4 callersMethodinfer
(self, obs: Dict)
packages/openpi-client/src/openpi_client/websocket_client_policy.py:38
↓ 4 callersMethodpush
Append transforms to the group and return a new group. Args: inputs: Appended to the *end* of the current input transforms.
src/openpi/transforms.py:49
↓ 3 callersMethod_dot
(self, x: at.Array, w: at.Array, lora_weights: tuple[at.Array, at.Array] | None)
src/openpi/models/lora.py:144
↓ 3 callersFunction_get_mtime
Get the mtime of a given date at midnight UTC.
src/openpi/shared/download.py:303
↓ 3 callersFunction_joint_flip_mask
Used to convert between aloha and pi joint angles.
src/openpi/policies/aloha_policy.py:110
↓ 3 callersFunction_normalize
(x, min_val, max_val)
src/openpi/policies/aloha_policy.py:115
↓ 3 callersMethod_process_effort_tokens
(self, obs: _model.Observation, mode: str)
src/openpi/models/pi0.py:202
↓ 3 callersMethod_project_effort
(self, effort: at.Float[at.Array, "b *d"])
src/openpi/models/pi0.py:197
↓ 3 callersFunction_set_folder_permission
Set folder permission to be read, write and searchable.
src/openpi/shared/download.py:260
↓ 3 callersFunction_set_permission
chmod requires executable permission to be set, so we skip if the permission is already match with the target.
src/openpi/shared/download.py:251
↓ 3 callersFunction_unnormalize
(x, min_val, max_val)
src/openpi/policies/aloha_policy.py:119
↓ 3 callersMethodcreate
( self, lr: optax.ScalarOrSchedule, weight_decay_mask: at.PyTree | None = None, )
src/openpi/training/optimizer.py:95
↓ 3 callersMethoddecode
(self, x)
src/openpi/models/gemma.py:145
↓ 3 callersFunctionget_arm_joint_positions
(bot)
examples/aloha_real/robot_utils.py:172
↓ 3 callersMethodget_observation
(self)
examples/aloha_real/real_env.py:123
↓ 3 callersMethodget_server_metadata
(self)
packages/openpi-client/src/openpi_client/websocket_client_policy.py:23
↓ 3 callersMethodget_statistics
Compute and return the statistics of the vectors processed so far. Returns: dict: A dictionary containing the computed s
src/openpi/shared/normalize.py:74
↓ 3 callersMethodinfer
(self, obs: dict)
src/openpi/policies/policy.py:41
↓ 3 callersFunctionmake_attn_mask
Adapted from big_vision. Tokens can attend to valid inputs tokens which have a cumulative mask_ar smaller or equal to theirs. This way `mask_
src/openpi/models/pi0.py:21
↓ 3 callersMethodreset
Reset the agent to its initial state.
packages/openpi-client/src/openpi_client/runtime/agent.py:16
↓ 3 callersMethodtokenize
( self, prompt: str, state: np.ndarray, actions: np.ndarray | None )
src/openpi/models/tokenizer.py:52
↓ 3 callersFunctiontorque_on
(bot)
examples/aloha_real/robot_utils.py:242
↓ 2 callersMethod_act_tokens_to_paligemma_tokens
(self, tokens: np.ndarray | list[int])
src/openpi/models/tokenizer.py:124
↓ 2 callersFunction_apply_rope
Applies RoPE positions [B, L] to x [B, L, H, D].
src/openpi/models/gemma.py:400
↓ 2 callersFunction_apply_rope
Applies RoPE positions [B, L] to x [B, L, H, D].
src/openpi/models/gemma_fast.py:425
↓ 2 callersFunction_assert_quantile_stats
(norm_stats: at.PyTree[NormStats])
src/openpi/transforms.py:428
↓ 2 callersMethod_convert_observation
(self, gym_obs: dict)
examples/aloha_sim/env.py:47
↓ 2 callersFunction_download_boto3
Download a file from the OpenPI S3 bucket using boto3. This is a more performant version of download but can only handle s3 urls. In openpi repo,
src/openpi/shared/download.py:143
↓ 2 callersFunction_ensure_permissions
Since we are sharing cache directory with containerized runtime as well as training script, we need to ensure that the cache directory has the cor
src/openpi/shared/download.py:265
↓ 2 callersFunction_merge_params
Merges the loaded parameters with the reference parameters. Args: loaded_params: The parameters to merge. params: The reference p
src/openpi/training/weight_loaders.py:77
↓ 2 callersFunction_parse_image
(image)
src/openpi/policies/droid_policy.py:21
↓ 2 callersFunction_parse_image
(image)
src/openpi/policies/libero_policy.py:20
↓ 2 callersFunction_parse_image
(image)
src/openpi/policies/tavla_policy.py:10
↓ 2 callersFunction_split_params
(state: training_utils.TrainState)
src/openpi/training/checkpoints.py:157
↓ 2 callersFunctionapply_tree
( tree: at.PyTree[T], selector: at.PyTree[S], fn: Callable[[T, S], T], *, strict: bool = False )
src/openpi/transforms.py:377
↓ 2 callersMethodclose
(self)
src/openpi/training/checkpoints.py:133
↓ 2 callersMethodcompute_loss
( self, rng: at.KeyArrayLike, observation: _model.Observation, actions: _model.Actions, *, train: bool
src/openpi/models/pi0.py:328
↓ 2 callersMethodcreate
(self, rng: at.KeyArrayLike)
src/openpi/models/pi0_fast.py:91
↓ 2 callersMethoddata_config
Get the data config for this data loader.
src/openpi/training/data_loader.py:34
↓ 2 callersMethoddecode
(self, x)
src/openpi/models/gemma_fast.py:120
↓ 2 callersMethodembed_inputs
( self, obs: _model.Observation )
src/openpi/models/pi0_fast.py:154
↓ 2 callersMethodembed_prefix
( self, obs: _model.Observation )
src/openpi/models/pi0.py:239
↓ 2 callersMethodembed_suffix
( self, obs: _model.Observation, noisy_actions: _model.Actions, timestep: at.Float[at.Array, " b"]
src/openpi/models/pi0.py:280
↓ 2 callersMethodextract_actions
(self, tokens: np.ndarray, action_horizon: int, action_dim: int)
src/openpi/models/tokenizer.py:107
↓ 2 callersMethodfrom_dict
This method defines the mapping between unstructured data (i.e., nested dict) to the structured Observation format.
src/openpi/models/model.py:110
↓ 2 callersFunctionget_cache_dir
()
src/openpi/shared/download.py:29
↓ 2 callersMethodget_reward
(self)
examples/aloha_real/real_env.py:131
↓ 2 callersMethodinputs_spec
Returns the input specification for the model. Values are jax.ShapeDtypeStruct.
src/openpi/models/model.py:257
↓ 2 callersFunctionmake_attn_mask
Adapted from big_vision. Tokens can attend to valid inputs tokens which have a cumulative mask_ar smaller or equal to theirs. This way `mask_
src/openpi/models/pi0_fast.py:22
↓ 2 callersMethodreset
(self, *, fake=False)
examples/aloha_real/real_env.py:134
↓ 2 callersMethodrestore
(self, *args, **kwargs)
src/openpi/training/checkpoints.py:143
↓ 2 callersMethodrun
Runs the runtime loop continuously until stop() is called or the environment is done.
packages/openpi-client/src/openpi_client/runtime/runtime.py:32
↓ 2 callersMethodrun
(self)
src/openpi/serving/websocket_policy_server.py:33
↓ 2 callersMethodserve_forever
(self)
src/openpi/serving/websocket_policy_server.py:30
↓ 2 callersMethodtokenize
(self, prompt: str)
src/openpi/models/tokenizer.py:18
↓ 2 callersFunctionunflatten_dict
Unflatten a flattened dictionary. Assumes that '/' was used as a separator.
src/openpi/transforms.py:318
↓ 1 callersMethod_adjust_histograms
Adjust histograms when min or max changes.
src/openpi/shared/normalize.py:89
↓ 1 callersMethod_compute_quantiles
Compute quantiles based on histograms.
src/openpi/shared/normalize.py:107
↓ 1 callersFunction_decode_aloha
(data: dict, *, adapt_to_pi: bool = False)
src/openpi/policies/aloha_policy.py:163
↓ 1 callersFunction_decode_state
(state: np.ndarray, *, adapt_to_pi: bool = False)
src/openpi/policies/aloha_policy.py:185
↓ 1 callersFunction_download_fsspec
Download a file from a remote filesystem to the local cache, and return the local path.
src/openpi/shared/download.py:125
↓ 1 callersFunction_encode_actions
(actions: np.ndarray, *, adapt_to_pi: bool = False)
src/openpi/policies/aloha_policy.py:194
↓ 1 callersFunction_encode_actions_inv
(actions: np.ndarray, *, adapt_to_pi: bool = False)
src/openpi/policies/aloha_policy.py:202
↓ 1 callersFunction_extract_observation
(args: Args, obs_dict, *, save_to_disk=False)
examples/droid/main.py:198
↓ 1 callersFunction_get_libero_env
Initializes and returns the LIBERO environment, along with the task description.
examples/libero/main.py:189
↓ 1 callersFunction_get_s3_transfer_manager
( session: boto3.Session, workers: int, botocore_config: botocore.config.Config | None = None )
src/openpi/shared/download.py:236
↓ 1 callersFunction_gripper_from_angular
(value)
src/openpi/policies/aloha_policy.py:145
↓ 1 callersFunction_gripper_from_angular_inv
(value)
src/openpi/policies/aloha_policy.py:157
↓ 1 callersFunction_gripper_to_angular
(value)
src/openpi/policies/aloha_policy.py:123
↓ 1 callersMethod_init_cache
Initialize KV cache
src/openpi/models/gemma_fast.py:165
↓ 1 callersFunction_is_openpi_url
Check if the url is an OpenPI S3 bucket url.
src/openpi/shared/download.py:298
↓ 1 callersMethod_load_norm_stats
(self, assets_dir: epath.Path, asset_id: str | None)
src/openpi/training/config.py:176
↓ 1 callersFunction_load_weights_and_validate
Loads and validates the weights. Returns a loaded subset of the weights.
scripts/train.py:72
↓ 1 callersMethod_make_lora_eqns
(self, eqn: str)
src/openpi/models/lora.py:67
↓ 1 callersFunction_merge_params
(train_state: training_utils.TrainState, params: dict[str, at.Params])
src/openpi/training/checkpoints.py:167
↓ 1 callersFunction_quat2axisangle
Copied from robosuite: https://github.com/ARISE-Initiative/robosuite/blob/eafb81f54ffc104f905ee48a16bb15f059176ad3/robosuite/utils/transform_util
examples/libero/main.py:199
↓ 1 callersMethod_reset_gripper
Set to position mode and do position resets: first open then close. Then change back to PWM mode
examples/aloha_real/real_env.py:114
↓ 1 callersMethod_reset_joints
(self)
examples/aloha_real/real_env.py:109
↓ 1 callersFunction_resize_with_pad_pil
Replicates tf.image.resize_with_pad for one image using PIL. Resizes an image to a target height and width without distortion by padding with zero
packages/openpi-client/src/openpi_client/image_tools.py:38
↓ 1 callersMethod_run_episode
Runs a single episode.
packages/openpi-client/src/openpi_client/runtime/runtime.py:50
↓ 1 callersFunction_set_file_permission
Set all files to be read & writable, if it is a script, keep it as a script.
src/openpi/shared/download.py:278
↓ 1 callersFunction_setup_folder_permission_between_cache_dir_and_path
(path: pathlib.Path)
src/openpi/shared/download.py:270
next →1–100 of 459, ranked by callers