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

↓ 1 callersFunctionmain

(api_key: Optional[str], port: int, transport: str)

examples/mcp-rl/servers/python/mcp_balldontlie/server.py:51

↓ 1 callersFunctionmain

(api_key: Optional[str], port: int, transport: str)

examples/mcp-rl/servers/python/mcp_googlemaps/server.py:180

↓ 1 callersFunctionmain

(api_key: Optional[str], port: int, transport: str)

examples/mcp-rl/servers/python/mcp_alphavantage/server.py:62

↓ 1 callersFunctionmain

Main training entry point.

examples/mcp-rl/mcp_rl/train.py:172

↓ 1 callersFunctionmain

Main function with CLI argument parsing.

examples/mcp-rl/mcp_rl/scenario_generator.py:195

↓ 1 callersFunctionmain

Run test scenario.

examples/mcp-rl/mcp_rl/rollout.py:272

↓ 1 callersFunctionmake_group

(prompt: str)

tests/integration/test_tinker_native_backend.py:75

↓ 1 callersFunctionmake_group_a

(prompt: str)

tests/integration/test_tinker_native_backend.py:148

↓ 1 callersFunctionmake_group_b

(prompt: str)

tests/integration/test_tinker_native_backend.py:193

↓ 1 callersFunctionmake_group_rollout_fn

Create a RolloutFn from a SingleRolloutFn by running it N times in parallel.

src/art/pipeline_trainer/trainer.py:35

↓ 1 callersFunctionmake_patch_source

This is a patch to the art.TrainableModel class to change project name logging these runs as test runs.

tests/integration.py:46

↓ 1 callersFunctionmake_variable_override_source

Create source code to override variables in the notebook.

tests/integration.py:68

↓ 1 callersFunctionmigrate_s3_checkpoints_to_new_structure

Migrate existing checkpoints in S3 from the old structure to the new structure. Old: s3://bucket/prefix/project/models/model_name/0001/

src/art/utils/s3_checkpoint_utils.py:107

↓ 1 callersFunctionnot_none

(x: T | None)

src/art/tinker/cookbook_v/utils/misc_utils.py:94

↓ 1 callersMethodnote_group_enqueued

(self, _group: TrajectoryGroup)

src/art/pipeline_trainer/status.py:113

↓ 1 callersMethodnote_rollout_finished

(self, *, errored: bool)

src/art/pipeline_trainer/status.py:106

↓ 1 callersMethodnote_rollout_started

(self)

src/art/pipeline_trainer/status.py:102

↓ 1 callersMethodnote_stale

(self, count: int)

src/art/pipeline_trainer/status.py:122

↓ 1 callersMethodnote_training_batch

( self, batch: list[TrajectoryGroup], *, step: int, step_seconds: float )

src/art/pipeline_trainer/status.py:140

↓ 1 callersMethodnote_training_start

(self, group_count: int)

src/art/pipeline_trainer/status.py:132

↓ 1 callersMethodnote_val_finished

(self, step: int, reward: float | None)

src/art/pipeline_trainer/status.py:164

↓ 1 callersMethodnote_val_started

(self, step: int)

src/art/pipeline_trainer/status.py:160

↓ 1 callersMethodnote_zero_variance_discarded

(self, count: int)

src/art/pipeline_trainer/status.py:127

↓ 1 callersFunctionpacked_tensors_from_tokenized_results

( tokenized_results: list[TokenizedResult], seq_len: int, pad_token_id: int = -100, truncate_l

src/art/preprocessing/pack.py:34

↓ 1 callersFunctionpacked_tensors_to_dir

(tensors: PackedTensors, dir: str)

src/art/preprocessing/pack.py:225

↓ 1 callersFunctionparse_args

()

dev/math-vista/math-vista.py:116

↓ 1 callersFunctionparse_args

()

scripts/deploy-model.py:20

↓ 1 callersFunctionparse_completion_to_openai_message

( completion_tokens: list[int], renderer: Any, )

src/art/tinker_native/data.py:112

↓ 1 callersFunctionparse_content_blocks

Parse a string with <think>...</think> and <tool_call>...</tool_call> tags. Handles interleaved thinking, tool call, and text blocks. Conten

src/art/tinker/cookbook_v/renderers/base.py:513

↓ 1 callersFunctionparse_indexes

Parse index specification string into list of valid indexes. Supports: - Single indexes: "0", "2" - Comma-separated: "0,2,3" - R

tests/integration.py:231

↓ 1 callersMethodparse_response

(self, response: list[int])

src/art/tinker/cookbook_v/renderers/deepseek_v3.py:249

↓ 1 callersFunctionparse_sse_to_chat_completion

Parse SSE (Server-Sent Events) content and build a ChatCompletion. This handles the case where streaming responses have already been consumed

src/art/auto_trajectory.py:16

↓ 1 callersFunctionpatch_httpx

()

src/art/auto_trajectory.py:143

↓ 1 callersFunctionpatch_preprocess_mask_arguments

()

src/art/transformers/patches.py:36

↓ 1 callersMethodplace_autocomplete

Get place autocomplete predictions

examples/mcp-rl/servers/python/mcp_googlemaps/server.py:147

↓ 1 callersMethodplace_details

Get details for a specific place

examples/mcp-rl/servers/python/mcp_googlemaps/server.py:124

↓ 1 callersMethodplaces_nearby_search

Search for nearby places

examples/mcp-rl/servers/python/mcp_googlemaps/server.py:72

↓ 1 callersMethodplaces_text_search

Search for places by text query

examples/mcp-rl/servers/python/mcp_googlemaps/server.py:101

↓ 1 callersFunctionplot_packed_tensors

( packed_tensors: PackedTensors, output_dir: str | None = None )

src/art/preprocessing/pack.py:262

↓ 1 callersFunctionprecalculate_new_logprobs

Precalculate logprobs for all offsets and return as a tensor.

src/art/unsloth/service.py:58

↓ 1 callersMethodprint_difficulty_distribution

Print a visual representation of the difficulty distribution.

src/art/mcp/types.py:184

↓ 1 callersFunctionprint_history_summary

(model: art.TrainableModel)

dev/yes-no-maybe-metrics.py:151

↓ 1 callersFunctionprint_history_summary

(model: art.TrainableModel, tail: int = 5)

src/art/pipeline_trainer/binary_prefix_tool_pipeline.py:154

↓ 1 callersMethodprompt_tokens

( self, base_model: str, messages: list[ChatCompletionMessageParam], tools: li

src/art/tinker/server.py:386

↓ 1 callersFunctionpush_model_to_s3

Push a model to S3. Args: model_name: The name of the model to push. project: The project name. s3_bucket: The S3 bucket

src/art/utils/s3.py:223

↓ 1 callersMethodraise_error

(self, message: str)

tests/unit/test_mp_actors.py:38

↓ 1 callersFunctionrecord_first_move_metrics

(trajectory: art.Trajectory, square: str)

examples/tic_tac_toe_self_play/rollout.py:95

↓ 1 callersMethodregister

Registers a model with the local Backend for logging and/or training. Args: model: An art.Model instance.

src/art/local/backend.py:183

↓ 1 callersMethodregister_cost_extractor

(self, provider: str, extractor: CostExtractor)

src/art/metrics.py:262

↓ 1 callersMethodregister_definitions

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:26

↓ 1 callersMethodregister_lora_for_step

src/art/tinker/service.py:147

↓ 1 callersFunctionreload_to_gpu

Reload model params and optimizer state to GPU.

src/art/megatron/offload.py:88

↓ 1 callersFunctionremove_thinking

Filter out ThinkingPart elements from a content part list.

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

↓ 1 callersFunctionrender_board

(game: TicTacToeGame)

examples/tic_tac_toe_self_play/game_utils.py:29

↓ 1 callersFunctionrender_board

(game: TicTacToeGame)

examples/tic_tac_toe/game_utils.py:23

↓ 1 callersMethodrender_message

(self, message: Message, ctx: RenderContext)

src/art/tinker/cookbook_v/renderers/qwen3.py:130

↓ 1 callersMethodrequest_stop

Request a clean shutdown of the pipeline stages.

src/art/pipeline_trainer/trainer.py:255

↓ 1 callersMethodresolve_ref

Resolve a reference to its schema definition

src/art/tinker/cookbook_v/renderers/kimi_k2_5_tool_declaration_ts.py:33

↓ 1 callersFunctionresponse_total_tokens

( response: openai.types.chat.chat_completion.ChatCompletion, )

dev/yes-no-maybe-metrics.py:75

↓ 1 callersMethodreverse_geocode

Reverse geocode coordinates to address

examples/mcp-rl/servers/python/mcp_googlemaps/server.py:48

↓ 1 callersFunctionreward

(config: MyConfig, scenario: MyTask, response: str)

examples/benchmarking_comparison_models.py:49

↓ 1 callersFunctionrollout

( client: openai.AsyncOpenAI, model_name: str, prompt: str )

dev/yes-no-maybe-megatron.py:13

↓ 1 callersFunctionrollout

( client: openai.AsyncOpenAI, model: art.TrainableModel, prompt: str )

dev/yes-no-maybe-kl-advantage.py:21

↓ 1 callersFunctionrollout

(model: art.TrainableModel, prompt: str)

dev/new_models/gemma3.py:12

↓ 1 callersFunctionrollout

(model: art.TrainableModel, prompt: str)

dev/new_models/qwen3_try.py:12

↓ 1 callersFunctionrollout

(model: art.TrainableModel, env_client: EchoEnv)

examples/openenv_echo.py:25

↓ 1 callersMethodrollout_fn

( _model: TrainableModel, _scenario: dict, _config: dict, )

tests/unit/test_track_api_cost.py:597

↓ 1 callersFunctionruler

Core RULER implementation that scores a list of message trajectories. This is the low-level API that works with raw message lists. For integratio

src/art/rewards/ruler.py:87

↓ 1 callersFunctionrun_benchmarks

()

examples/just-the-facts/just_the_facts/generate_benchmarks.py:78

↓ 1 callersFunctionrun_benchmarks

()

examples/2048/generate_benchmarks.py:63

↓ 1 callersFunctionrun_benchmarks

(server: str = "mcp_alphavantage")

examples/mcp-rl/mcp_rl/benchmarks/generate_benchmarks.py:124

↓ 1 callersMethodsampling_client

(self)

src/art/tinker/server.py:294

↓ 1 callersFunctionsave_prompt_images

Render each prompt as centered text on a white background and save as PNG. Args: prompts: Sequence of prompt strings. output

dev/yes-no-maybe-vision/generate_images.py:132

↓ 1 callersFunctionsay_hi

A method that says hi to an assistant and returns the response.

tests/unit/test_yield_trajectory.py:132

↓ 1 callersFunctionscenario_iter

()

src/art/pipeline_trainer/binary_prefix_tool_pipeline.py:315

↓ 1 callersFunctionscore_group

(group_idx: int, trajectory_idx: int)

examples/mcp-rl/mcp_rl/benchmarks/generate_benchmarks.py:64

↓ 1 callersFunctionscore_title_async

(row_data: dict)

examples/hn_title_generator/reference_grpo_trainer.py:107

↓ 1 callersMethodset_cost_calculator

(self, calculator: CostCalculator | None)

src/art/model.py:931

↓ 1 callersFunctionset_vllm_log_file

Sets the vLLM log file to the given path.

src/art/vllm/server.py:184

↓ 1 callersMethodsharded_lora_state_dict

(self)

src/art/megatron/lora.py:119

↓ 1 callersFunctionshared_prefix_len

(guess: str, secret: str)

src/art/pipeline_trainer/binary_prefix_tool_pipeline.py:82

↓ 1 callersFunctionsimple_rollout

A simple rollout function for testing.

tests/test_backend_train_api.py:65

↓ 1 callersFunctionsimple_rollout

( client: openai.AsyncOpenAI, model_name: str, prompt: str )

tests/integration/test_vllm_contract.py:63

↓ 1 callersFunctionsimple_rollout

A simple rollout function that rewards 'yes' responses.

tests/integration/test_push_and_fork.py:31

↓ 1 callersFunctionsimple_rollout

A simple rollout function for testing.

tests/integration/test_multi_checkpoint_training.py:38

↓ 1 callersMethodslow

(self, delay: float = 5.0)

tests/unit/test_mp_actors.py:34

↓ 1 callersMethodstart_openai_server

( self, config: dev.OpenAIServerConfig | None )

src/art/local/service.py:19

↓ 1 callersMethodstop

Stop the MCP server connection.

examples/mcp-rl/mcp_rl/mcp_server.py:26

↓ 1 callersFunctionstream

()

src/art/cli.py:359

↓ 1 callersFunctionsuppress_litellm_serialization_warnings

Suppress litellm's internal Pydantic serialization warnings. These warnings occur due to a known litellm issue (#11759) where response t

src/art/utils/suppress_litellm_serialization_warnings.py:4

↓ 1 callersFunctiontest_basic_scenario_generation

Test basic scenario generation with tools only.

examples/mcp-rl/test_scenario_generation.py:109

↓ 1 callersFunctiontest_client

()

src/art/vllm/server.py:104

↓ 1 callersFunctiontest_dict_input_compatibility

Test backward compatibility with dictionary inputs.

examples/mcp-rl/test_scenario_generation.py:182

↓ 1 callersFunctiontest_error_handling

Test error handling scenarios.

examples/mcp-rl/test_scenario_generation.py:239

↓ 1 callersFunctiontest_rollout

()

examples/mcp-rl/mcp_rl/rollout.py:254

↓ 1 callersFunctiontest_scenario_generation_with_resources

Test scenario generation with both tools and resources.

examples/mcp-rl/test_scenario_generation.py:150

↓ 1 callersFunctiontest_scraper

Test the scrape_article function with example URLs

examples/just-the-facts/test_scraper.py:9

↓ 1 callersFunctiontest_step_skipping

Test that step counting works correctly when training is skipped.

src/art/test/test_step_skipping.py:39

↓ 1 callersFunctiontest_tool_resource_classes

Test Tool and Resource class functionality.

examples/mcp-rl/test_scenario_generation.py:275

↓ 1 callersMethodticker

(self, n: int = 1_000)

tests/unit/test_mp_actors.py:29

← previous next →701–800 of 1,502, ranked by callers