Functions — github.com/OpenPipe/ART

CodeHub MCPcopy Create free account

hub / github.com/OpenPipe/ART / functions

Functions1,502 in github.com/OpenPipe/ART

⨍ Functions 1,502◇ Types & classes 261↳ Endpoints 14

↓ 1 callersFunctiontimed_request

Execute a single model request and measure elapsed time and token usage.

dev/new_models/benchmark_inference.py:21

↓ 1 callersFunctionto_cumulative_metric_key

src/art/metrics.py:36

↓ 1 callersMethodto_json

Convert the collection to JSON.

src/art/mcp/types.py:123

↓ 1 callersFunctiontokenize_sft_batch

Tokenize a single batch of trajectories for SFT. Args: trajectory_batch: List of trajectories in this batch learning_rate: Learni

src/art/preprocessing/tokenize.py:366

↓ 1 callersFunctiontokenize_trajectory

Tokenizes a trajectory and returns a TokenizedResult.

src/art/preprocessing/tokenize.py:154

↓ 1 callersFunctiontokenize_trajectory_groups

( tokenizer: "PreTrainedTokenizerBase", trajectory_groups: list[TrajectoryGroup], allow_training_w

src/art/preprocessing/tokenize.py:71

↓ 1 callersFunctiontotal_board_value

(game: TwentyFortyEightGame)

examples/2048/utils.py:179

↓ 1 callersMethodtrack_duration

(self, metric_name: str)

src/art/trajectories.py:59

↓ 1 callersFunctiontrain

(model: art.TrainableModel[JustTheFactsConfig])

examples/just-the-facts/just_the_facts/train.py:20

↓ 1 callersFunctiontrain

examples/2048/train.py:32

↓ 1 callersFunctiontrain_mcp_agent

Example training function that creates AlphaMcpServer and passes it in scenarios.

examples/mcp-rl/mcp_rl/train.py:33

↓ 1 callersFunctiontrain_model

(model: art.TrainableModel)

examples/benchmarking_comparison_models.py:116

↓ 1 callersMethodtrain_sft

Train using SFT on pre-computed batches. Args: batches: List of SFTBatch objects to train on. verbose: Whether to pri

src/art/unsloth/service.py:763

↓ 1 callersMethodtrain_sft

Train using SFT on pre-computed batches. Args: batches: List of SFTBatch objects to train on. verbose: Whether to pri

src/art/local/service.py:33

↓ 1 callersFunctiontrain_sft_from_file

Train a model using supervised fine-tuning from a JSONL file. Streams data without loading all into memory. Suitable for large files (10GB+)

src/art/utils/sft.py:332

↓ 1 callersFunctiontrajectory_group_to_dict

(trajectory_group: TrajectoryGroup)

src/art/utils/trajectory_migration.py:37

↓ 1 callersFunctiontrajectory_groups_to_datums

( trajectory_groups: Iterable[TrajectoryGroup], renderer: Any, tokenizer: Any, normalize_advan

src/art/tinker_native/data.py:131

↓ 1 callersFunctiontrajectory_to_dict

(trajectory: Trajectory)

src/art/utils/trajectory_migration.py:58

↓ 1 callersFunctionunwrap_move

examples/tic_tac_toe_self_play/game_utils.py:45

↓ 1 callersFunctionupdate_version

Update version in pyproject.toml.

scripts/bump_version.py:41

↓ 1 callersFunctionwith_quotes

dev/yes-no-maybe-megatron.py:34

↓ 1 callersFunctionwith_quotes

dev/yes-no-maybe-metrics.py:44

↓ 1 callersFunctionwith_quotes

dev/yes-no-maybe.py:36

↓ 1 callersFunctionwith_quotes

dev/yes-no-maybe-kl-advantage.py:47

↓ 1 callersMethodwithout_prompt

src/art/preprocessing/tokenize.py:36

↓ 1 callersFunctionwrap_trajectories_awaitable

( awaitable: Awaitable[Trajectory] | Awaitable[Iterable[Trajectory]], )

src/art/gather.py:159

( exceptions: list[BaseException], metadata: dict[str, MetadataValue] | None,

src/art/trajectories.py:268

Combine two migration results.

src/art/utils/trajectory_migration.py:166

Method__await__

src/art/trajectories.py:305

Support for copy.copy()

src/art/trajectories.py:181

Method__deepcopy__

Support for copy.deepcopy()

src/art/trajectories.py:197

Method__deepcopy__

tests/unit/test_strip_logprobs.py:157

Method__deepcopy__

tests/unit/test_strip_logprobs.py:183

Method__enter__

src/art/auto_trajectory.py:74

Method__enter__

src/art/yield_trajectory.py:24

Method__enter__

src/art/local/backend.py:159

(self, exc_type: Any, exc_value: Any, traceback: Any)

src/art/auto_trajectory.py:78

(self, exc_type: Any, exc_value: Any, traceback: Any)

src/art/yield_trajectory.py:27

( self, exc_type: type[BaseException] | None, exc: BaseException | None, tb: T

src/art/local/backend.py:162

Method__getattr__

(self, name: str)

src/mp_actors/move.py:179

Method__getitem__

src/art/mcp/types.py:133

Function__init__

(self, *args: Any, **kwargs: Any)

src/art/cli.py:316

( self, obj: object, log_file: str | None = None, process_name: str | None = None )

src/mp_actors/move.py:105

(self, coro, num_trajectories)

src/art/trajectories.py:301

(self, message: str)

src/art/errors.py:26

(self, message: str)

src/art/errors.py:37

(self, message: str)

src/art/errors.py:48

(self, message: str)

src/art/errors.py:58

src/art/auto_trajectory.py:71

(self, **kwargs)

src/art/__init__.py:21

( self, cost_context: str, *, _shared_state: _SharedMetricsState | None = None

src/art/metrics.py:80

( self, *, name: str, project: str, entity: str | None = None,

src/art/model.py:896

src/art/yield_trajectory.py:21

( self, name: str, display_name: str | None = None, split: str | None = None )

src/art/utils/benchmarking/types.py:6

(self, model: str, split: str, step_indices: list[int] | None = None)

src/art/utils/old_benchmarking/types.py:9

(self, index: int, metrics: dict[str, float] | None = None)

src/art/utils/old_benchmarking/types.py:29

( self, model_key: BenchmarkedModelKey, steps: list[BenchmarkedModelStep] | None = Non

src/art/utils/old_benchmarking/types.py:41

( self, *, in_process: bool = False, path: str | None = None, )

src/art/megatron/backend.py:10

( self, config: TransformerConfig, layer_number: int, attn_mask_type: AttnMask

src/art/megatron/flex_attention.py:113

( self, adapter_model_prefix: str, linear_proj: TERowParallelLinear, rank: int

src/art/megatron/lora.py:166

( self, adapter_model_prefix: str, linear_qkv: TELayerNormColumnParallelLinear,

src/art/megatron/lora.py:206

( self, adapter_model_prefix: str, linear_fc1: TEColumnParallelGroupedLinear,

src/art/megatron/lora.py:294

( self, adapter_model_prefix: str, linear_fc2: TERowParallelGroupedLinear, ran

src/art/megatron/lora.py:338

(self, filepath)

src/art/langgraph/logging.py:6

(self, llm, logger, structured_output=None, tools=None)

src/art/langgraph/llm_wrapper.py:131

( self, *, api_key: str | None = None, base_url: str | None = None )

src/art/serverless/client.py:259

( self, *, api_key: str | None = None, base_url: str | None = None )

src/art/serverless/backend.py:79

( self, *, tinker_api_key: str | None = None, path: str | None = None, )

src/art/tinker_native/backend.py:115

(self, *args: object, **kwargs: object)

src/art/vllm/patches.py:13

(self, scenarios: List[GeneratedScenario])

src/art/mcp/types.py:100

( self, *, tinker_api_key: str | None = None, in_process: bool = False,

src/art/tinker/backend.py:16

(self, api_key: str, max_concurrent_sampling_clients: int)

src/art/tinker/server.py:309

(self, label: tuple[int, ...], child: _RadixNode)

src/art/tinker/prefix_cache.py:32

( self, parent: _RadixNode | None = None, parent_token: int | None = None )

src/art/tinker/prefix_cache.py:40

(self, max_entries: int = 16_384)

src/art/tinker/prefix_cache.py:62

( self, tokenizer: Tokenizer, use_system_prompt: bool = False, reasoning_effor

src/art/tinker/cookbook_v/renderers/gpt_oss.py:192

( self, tokenizer: Tokenizer, image_processor: ImageProcessor | None = None, s

src/art/tinker/cookbook_v/renderers/kimi_k25.py:40

(self, tokenizer: Tokenizer)

src/art/tinker/cookbook_v/renderers/base.py:739

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:22

(self, type: str, extra_props: dict[str, Any] | None = None)

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:88

( self, json_schema_object: dict[str, Any], registry: _SchemaRegistry | None = None,

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:112

( self, json_schema_object: dict[str, Any], registry: _SchemaRegistry | None = None,

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:195

(self, json_schema_object: dict[str, Any])

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:229

( self, json_schema_object: dict[str, Any], registry: _SchemaRegistry | None = None,

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:267

(self, json_schema_object: dict[str, Any])

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:284

(self, json_schema_object: dict[str, Any], registry: _SchemaRegistry)

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:306

( self, tokenizer: Tokenizer, system_role_as_user: bool = False, strip_thinkin

src/art/tinker/cookbook_v/renderers/deepseek_v3.py:44

Args: tokenizer: The tokenizer to use for encoding. strip_thinking_from_history: When True (default), strips <think>.

src/art/tinker/cookbook_v/renderers/qwen3.py:86

( self, tokenizer: Tokenizer, image_processor: ImageProcessor | None = None, s

src/art/tinker/cookbook_v/renderers/qwen3.py:424

(self, tokenizer: Tokenizer, strip_thinking_from_history: bool = True)

src/art/tinker/cookbook_v/renderers/kimi_k2.py:372

( self, model: art.TrainableModel, backend: art.Backend, rollout_fn: RolloutFn

src/art/pipeline_trainer/trainer.py:60

( self, *, get_scenario_offset: Callable[[], int], log_interval_seconds: float

src/art/pipeline_trainer/status.py:15

Initializes a local, directory-based Backend interface at the given path. Note: The local Backend uses Weights & Biases

src/art/local/backend.py:74

( self, model_name: str, base_model: str, config: dev.InternalModelConfig,

src/art/local/service.py:10

(self, notebook_configs: list[dict])

tests/integration.py:168

tests/unit/test_unsloth_metrics.py:8

( self, prompt_tokens: int, completion_tokens: int, *, cached_tokens:

tests/unit/test_track_api_cost.py:15

( self, prompt_tokens: int, completion_tokens: int, *, cached_tokens:

tests/unit/test_track_api_cost.py:36

( self, input_tokens: int, output_tokens: int, *, cache_creation_input

tests/unit/test_track_api_cost.py:57

( self, input_tokens: int, output_tokens: int, *, cache_creation_input

tests/unit/test_track_api_cost.py:72

← previous next →801–900 of 1,502, ranked by callers