Functions — github.com/OpenPipe/ART

hub / github.com/OpenPipe/ART / functions

Functions1,502 in github.com/OpenPipe/ART

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

Functiontest_ruler_skips_cost_when_pricing_is_unavailable

(monkeypatch)

tests/unit/test_ruler_metrics.py:107

Methodtest_served_model_name_uses_base_model_when_lora_enabled

With LoRA enabled, served model name should remain the base model.

tests/unit/test_multi_checkpoint_inference.py:232

Functiontest_server

Start a test server for the module.

tests/unit/test_auto_trajectory.py:189

Functiontest_server

Start a test server for the module.

tests/unit/test_yield_trajectory.py:111

Functiontest_serverless_backend

Test multi-checkpoint inference with ServerlessBackend.

tests/integration/test_multi_checkpoint_training.py:175

Functiontest_shallow_copy

Test that shallow copy works correctly.

tests/unit/test_trajectory_copy.py:62

Functiontest_shared_mode_empty_config

()

tests/unit/test_dedicated_config.py:11

Functiontest_shared_mode_with_other_keys

()

tests/unit/test_dedicated_config.py:16

Methodtest_should_log_wandb_logic_default

Test the should_log_wandb logic with default report_metrics.

tests/unit/test_frontend_logging.py:830

Methodtest_should_log_wandb_logic_empty_list

Test the should_log_wandb logic with empty report_metrics list.

tests/unit/test_frontend_logging.py:883

Methodtest_should_log_wandb_logic_explicit_wandb

Test the should_log_wandb logic with explicit wandb in report_metrics.

tests/unit/test_frontend_logging.py:868

Methodtest_should_log_wandb_logic_with_key

Test the should_log_wandb logic with WANDB_API_KEY present.

tests/unit/test_frontend_logging.py:852

Methodtest_simple_trajectory

Basic round-trip test with minimal trajectory.

tests/unit/test_trajectory_parquet.py:87

Methodtest_sparse_steps_omit_rollup_for_missing_costs

(self)

tests/unit/test_metrics_builder.py:142

Methodtest_standard_metrics_present

Verify standard metrics (reward, exception_rate, reward_std_dev) are computed.

tests/unit/test_frontend_logging.py:359

Methodtest_state_dict_round_trip_preserves_cumulative_state

(self)

tests/unit/test_metrics_builder.py:157

Methodtest_step_numbering_format

Verify step numbers are zero-padded to 4 digits.

tests/unit/test_frontend_logging.py:272

Methodtest_strip_dict_with_logprobs

Test stripping logprobs from dictionary.

tests/unit/test_strip_logprobs.py:15

Methodtest_strip_empty_structures

Test stripping logprobs from empty structures.

tests/unit/test_strip_logprobs.py:109

Methodtest_strip_list_with_logprobs

Test stripping logprobs from lists.

tests/unit/test_strip_logprobs.py:46

Methodtest_strip_mixed_nested_structure

Test stripping logprobs from mixed nested structures.

tests/unit/test_strip_logprobs.py:89

Methodtest_strip_nested_dict

Test stripping logprobs from deeply nested dictionaries.

tests/unit/test_strip_logprobs.py:29

Methodtest_strip_none_and_primitives

Test stripping logprobs from None and primitive values.

tests/unit/test_strip_logprobs.py:115

Methodtest_strip_object_with_logprobs

Test stripping logprobs from objects with __dict__.

tests/unit/test_strip_logprobs.py:73

Methodtest_strip_tuple_with_logprobs

Test stripping logprobs from tuples.

tests/unit/test_strip_logprobs.py:59

Functiontest_subclass_chat_completion_request_forces_logprobs

()

tests/unit/test_vllm_patches_contract.py:13

Functiontest_summarize_trajectory_groups_only_counts_explicit_scenario_id

()

tests/unit/test_metrics_taxonomy.py:56

Methodtest_throughput_metrics_derive_from_time_and_token_cumulatives

( self, )

tests/unit/test_metrics_builder.py:82

Functiontest_tinker_backend

Test multi-checkpoint inference with TinkerBackend.

tests/integration/test_multi_checkpoint_training.py:116

Functiontest_tinker_native_backend

()

tests/integration/test_tinker_native_backend.py:58

Functiontest_tinker_native_fork_checkpoint

Train model A for 1 step with save_checkpoint, fork to model B, train model B.

tests/integration/test_tinker_native_backend.py:122

Methodtest_tool_calls

Test trajectories with tool calls.

tests/unit/test_trajectory_parquet.py:122

Methodtest_train_logs_add_default_data_metrics_from_trajectory_groups

( self, tmp_path: Path )

tests/unit/test_frontend_logging.py:569

Methodtest_train_logs_automatic_wall_time_and_gpu_cost

( self, tmp_path: Path )

tests/unit/test_frontend_logging.py:902

Methodtest_train_sft_aggregates_metrics

Verify train_sft aggregates metrics from multiple batches into one log entry.

tests/unit/test_frontend_logging.py:1017

Methodtest_train_sft_no_metrics_when_empty

Verify train_sft handles empty training gracefully.

tests/unit/test_frontend_logging.py:1125

Methodtest_train_sft_single_step_increment

Verify train_sft results in single step increment regardless of batch count.

tests/unit/test_frontend_logging.py:1087

Methodtest_train_trajectory_metrics_default_to_train_prefix

( self, tmp_path: Path )

tests/unit/test_frontend_logging.py:531

Methodtest_trainable_model_get_inference_name_with_step

TrainableModel should also support step parameter.

tests/unit/test_multi_checkpoint_inference.py:69

Functiontest_trainer_not_contiguous

()

tests/unit/test_dedicated_config.py:93

Functiontest_trainer_not_starting_at_zero

()

tests/unit/test_dedicated_config.py:86

Methodtest_unicode_content

Test trajectories with unicode and special characters.

tests/unit/test_trajectory_parquet.py:267

Methodtest_unique_scenario_count_tracks_exact_ids

(self)

tests/unit/test_metrics_builder.py:211

Methodtest_unknown_local_gpu_skips_cost_but_keeps_wall_time

( self, tmp_path: Path )

tests/unit/test_frontend_logging.py:952

Methodtest_update_wandb_config_rejects_conflicting_values

( self, tmp_path: Path )

tests/unit/test_metric_routing.py:198

Methodtest_update_wandb_config_seeds_wandb_init

(self, tmp_path: Path)

tests/unit/test_metric_routing.py:128

Methodtest_update_wandb_config_updates_active_run

(self, tmp_path: Path)

tests/unit/test_metric_routing.py:159

Functiontest_valid_dedicated_four_gpus

()

tests/unit/test_dedicated_config.py:42

Functiontest_valid_dedicated_three_gpus

()

tests/unit/test_dedicated_config.py:36

Functiontest_valid_dedicated_two_gpus

()

tests/unit/test_dedicated_config.py:30

Functiontest_valid_shared_mode

()

tests/unit/test_dedicated_config.py:26

Methodtest_wandb_not_called_without_api_key

Verify _get_wandb_run returns None without WANDB_API_KEY.

tests/unit/test_frontend_logging.py:809

Functiontest_yield_trajectory

(test_server: None)

tests/unit/test_yield_trajectory.py:131

Functiontimed

(key: str, metrics: dict[str, Any])

src/art/tinker/cookbook_v/utils/misc_utils.py:19

Methodtinker_native_backend_class

Import TinkerNativeBackend, skipping if dependency unavailable.

tests/unit/test_multi_checkpoint_inference.py:258

Methodto_dict

Convert the resource to a dictionary.

src/art/mcp/types.py:62

Methodto_dict

Convert the scenario to a dictionary.

src/art/mcp/types.py:86

Methodto_openai_message

Convert a Message to OpenAI API format with reasoning_content for thinking. GptOss uses the analysis channel for thinking, which maps to reas

src/art/tinker/cookbook_v/renderers/gpt_oss.py:464

Methodto_openai_message

Convert a Message to OpenAI API format with reasoning_content for thinking. DeepSeek's API uses reasoning_content for thinking models, simila

src/art/tinker/cookbook_v/renderers/deepseek_v3.py:285

Methodto_openai_message

Convert a Message to OpenAI API format with reasoning_content for thinking. Qwen3's HF template accepts either: - message['reasoning_

src/art/tinker/cookbook_v/renderers/qwen3.py:232

Methodto_openai_message

Convert a Message to OpenAI API format with reasoning_content for thinking. Kimi K2's HF template explicitly expects reasoning_content as a s

src/art/tinker/cookbook_v/renderers/kimi_k2.py:785

Methodto_tool_schema

Convert the tool to a tool schema.

src/art/mcp/types.py:35

Methodto_typescript_style

(self, indent: str = "")

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:101

Methodto_typescript_style

(self, indent: str = "")

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:152

Methodto_typescript_style

(self, indent: str = "")

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:208

Methodto_typescript_style

(self, indent: str = "")

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:258

Methodto_typescript_style

(self, indent: str = "")

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:277

Methodto_typescript_style

(self, indent: str = "")

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:298

Methodto_typescript_style

(self, indent: str = "")

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:318

Methodto_typescript_style

(self, indent: str = "")

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:357

Methodtokens

(self)

src/art/preprocessing/tokenize.py:33

Functiontokens_weights_from_strings_weights

( strings_weights: list[tuple[str, float]], tokenizer: Tokenizer, )

src/art/tinker/cookbook_v/renderers/base.py:1165

Functiontrack_api_cost

( *, source: str, provider: str, model_name: str, response_getter: ResponseGetter | None =

src/art/api_costs.py:483

Methodtrain

( self, disk_packed_tensors: DiskPackedTensors, config: types.TrainConfig, _co

src/art/megatron/service.py:215

Methodtrain

( self, disk_packed_tensors: DiskPackedTensors, config: types.TrainConfig, _co

src/art/tinker/service.py:58

Methodtrain

( self, disk_packed_tensors: DiskPackedTensors, config: types.TrainConfig, _co

src/art/local/service.py:25

Functiontrain_rl

Interactive wizard to create an RL training script.

src/art/cli.py:128

Functiontrain_sft

Interactive wizard to create an SFT training script.

src/art/cli.py:122

Methodtrain_sft

( self, batches: list[Any], verbose: bool = False, )

src/art/megatron/service.py:299

Methodtrain_sft

( self, batches: list[Any], verbose: bool = False, )

src/art/tinker/service.py:164

Methodtraining_jobs

(self)

src/art/serverless/client.py:297

Functiontraining_progress_chart

Plot and save a line chart of *metric_name* over training *step* for every model. Parameters ---------- df : pl.DataFrame Table p

src/art/utils/benchmarking/charts/training_progress_chart.py:15

Methodtrajectories

()

tests/unit/test_frontend_logging.py:509

Functiontrajectory_group_batches

( groups: Iterable[Awaitable[TrajectoryGroup]], *, batch_size: int, max_batch_exceptions: int

src/art/batches.py:10

Functionunregister_renderer

Unregister a custom renderer factory. Args: name: The renderer name to unregister. Returns: True if the renderer was unregis

src/art/tinker/cookbook_v/renderers/__init__.py:80

Functionunregister_tokenizer

Unregister a custom tokenizer factory. Args: name: The tokenizer name to unregister. Returns: True if the tokenizer was unre

src/art/tinker/cookbook_v/tokenizer_utils.py:56

Methodunsloth_service_class

Import UnslothService, skipping if dependencies unavailable.

tests/unit/test_multi_checkpoint_inference.py:300

Methodvllm_engine_is_sleeping

(self)

src/art/megatron/service.py:212

Methodvllm_engine_is_sleeping

(self)

src/art/unsloth/service.py:505

Methodvllm_engine_is_sleeping

(self)

src/art/tinker/service.py:55

Methodwith_config

( self, config=None, **kwargs: Any, )

src/art/langgraph/llm_wrapper.py:212

Methodwith_retry

( self, *, retry_if_exception_type=(Exception,), wait_exponential_jitter=True,

src/art/langgraph/llm_wrapper.py:202

Functionwith_serialized_stories

(df: pl.DataFrame)

examples/hn_title_generator/utils.py:39

Methodwith_structured_output

(self, tools)

src/art/langgraph/llm_wrapper.py:191

Methodworker

()

tests/unit/test_metrics_builder.py:239

Functionwrap_rollout

(model, fn)

src/art/langgraph/llm_wrapper.py:95

Functionwrapper

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

src/mp_actors/traceback.py:15

Functionwrapper

(*args, **kwargs)

src/art/utils/limit_concurrency.py:11

Functionwrapper

(*args, **kwargs)

src/art/utils/log_http_errors.py:8

Functionwrapper

(*args, **kwargs)

src/art/langgraph/llm_wrapper.py:96

← previous next →1,401–1,500 of 1,502, ranked by callers