Functions — github.com/SqueezeAILab/LLMCompiler

hub / github.com/SqueezeAILab/LLMCompiler / functions

Functions245 in github.com/SqueezeAILab/LLMCompiler

⨍ Functions 245◇ Types & classes 38

↓ 14 callersMethodjoin

( self, input_query: str, agent_scratchpad: str, is_final: bool )

src/llm_compiler/llm_compiler.py:201

↓ 12 callersFunctionlog

Print the given string only if logging is enabled.

src/utils/logger_utils.py:59

↓ 7 callersMethodparse

Parse text into agent action/finish.

src/agents/agent.py:36

↓ 6 callersMethodtool_run_logging_kwargs

(self)

src/agents/agent.py:284

↓ 5 callersMethodapredict

Format prompt with kwargs and pass to LLM. Args: callbacks: Callbacks to pass to LLMChain **kwargs: Keys to pass to p

src/chains/llm_chain.py:307

↓ 5 callersFunctionget_model

( model_type, model_name, vllm_port, stream, temperature=0, )

src/utils/model_utils.py:6

↓ 5 callersMethodpredict

Format prompt with kwargs and pass to LLM. Args: callbacks: Callbacks to pass to LLMChain **kwargs: Keys to pass to p

src/chains/llm_chain.py:290

↓ 5 callersMethodrun

Convenience method for executing chain. The main difference between this method and `Chain.__call__` is that this method expects inpu

src/chains/chain.py:444

↓ 5 callersMethodsearch

Try to search for wiki page. If page exists, return the page summary, and a PageWithLookups object. If page does not exist, return si

src/docstore/wikipedia.py:144

↓ 4 callersMethod_areturn

( self, output: AgentFinish, intermediate_steps: list, run_manager: Optional[A

src/executors/agent_executor.py:224

↓ 4 callersMethodacall

Asynchronously execute the chain. Args: inputs: Dictionary of inputs, or single input if chain expects only one p

src/chains/chain.py:308

↓ 4 callersFunctionclean_str

(p)

src/docstore/wikipedia.py:14

↓ 4 callersMethodcreate_outputs

Create outputs from response.

src/chains/llm_chain.py:268

↓ 4 callersFunctionnormalize_answer

(s)

src/utils/evaluation_utils.py:8

↓ 3 callersFunction_make_with_name

(tool_name: str)

src/tools/base.py:413

↓ 3 callersMethod_return

( self, output: AgentFinish, intermediate_steps: list, run_manager: Optional[C

src/executors/agent_executor.py:211

↓ 3 callersMethodarun

Convenience method for executing chain. The main difference between this method and `Chain.__call__` is that this method expects inpu

src/chains/chain.py:514

↓ 3 callersMethodfrom_llm

( cls, llm: BaseLanguageModel, prompt: BasePromptTemplate = PROMPT, **kwargs:

src/chains/llm_math_chain.py:289

↓ 3 callersMethodreturn_stopped_response

Return response when agent has been stopped due to max iterations.

src/agents/agent.py:242

↓ 2 callersMethod_acall

Asynchronously execute the chain. This is a private method that is not user-facing. It is only called within `Chain.acall`, which

src/chains/chain.py:216

↓ 2 callersMethod_all_tasks_done

(self)

src/llm_compiler/task_fetching_unit.py:97

↓ 2 callersMethod_call

Execute the chain. This is a private method that is not user-facing. It is only called within `Chain.__call__`, which is the user

src/chains/chain.py:193

↓ 2 callersMethod_evaluate_expression

(self, expression: str)

src/chains/llm_math_chain.py:186

↓ 2 callersMethod_get_all_executable_tasks

(self)

src/llm_compiler/task_fetching_unit.py:100

↓ 2 callersMethod_get_alternative

(self, result: str)

src/docstore/wikipedia.py:67

↓ 2 callersMethod_get_page_obs

(page)

src/docstore/wikipedia.py:55

↓ 2 callersMethod_get_tool_return

Check if the tool is a returning tool.

src/executors/agent_executor.py:548

↓ 2 callersMethod_match_buffer_and_generate_task

Runs every time "\n" is encountered in the input stream or at the end of the stream. Matches the buffer against the regex patterns and generat

src/llm_compiler/planner.py:96

↓ 2 callersMethod_parse_generation

( self, generation: List[Dict[str, str]] )

src/chains/llm_chain.py:363

↓ 2 callersMethod_prepare_intermediate_steps

( self, intermediate_steps: List[Tuple[AgentAction, str]] )

src/executors/agent_executor.py:563

↓ 2 callersMethod_run_task

(self, task: Task)

src/llm_compiler/task_fetching_unit.py:117

↓ 2 callersMethod_should_continue

(self, iterations: int, time_elapsed: float)

src/executors/agent_executor.py:200

↓ 2 callersMethodagenerate

Generate LLM result from inputs.

src/chains/llm_chain.py:138

↓ 2 callersMethodapost_process

( self, response_text: str, entity: str, skip_retry_when_postprocess: bool = False )

src/docstore/wikipedia.py:111

↓ 2 callersMethodasearch

Try to search for wiki page. If page exists, return the page summary, and a PageWithLookups object. If page does not exist, return si

src/docstore/wikipedia.py:182

↓ 2 callersFunctioncompare_answer

Compare the answer (from Agent) and label (GT). Label can be either a string or a number. If label is a number, we allow 10% margin. Other

src/utils/evaluation_utils.py:63

↓ 2 callersFunctionenable_logging

Toggle logging on or off based on the given argument.

src/utils/logger_utils.py:53

↓ 2 callersMethodgenerate

Generate LLM result from inputs.

src/chains/llm_chain.py:103

↓ 2 callersFunctiongenerate_llm_compiler_prompt

( tools: Sequence[Union[Tool, StructuredTool]], example_prompt=str, is_replan: bool = False, )

src/llm_compiler/planner.py:36

↓ 2 callersMethodget_full_inputs

Create the full inputs for the LLMChain from intermediate steps.

src/agents/agent.py:161

↓ 2 callersMethodget_stats

(self)

src/docstore/wikipedia.py:49

↓ 2 callersMethodget_though_action_observation

( self, include_action=True, include_thought=True, include_action_idx=False )

src/llm_compiler/task_fetching_unit.py:59

↓ 2 callersFunctioninstantiate_task

( tools: Sequence[Union[Tool, StructuredTool]], idx: int, tool_name: str, args: str, thoug

src/llm_compiler/output_parser.py:111

↓ 2 callersMethodpost_process

( self, response_text: str, entity: str, skip_retry_when_postprocess: bool = False )

src/docstore/wikipedia.py:78

↓ 2 callersMethodprep_inputs

Validate and prepare chain inputs, including adding inputs from memory. Args: inputs: Dictionary of raw inputs, or single input i

src/chains/chain.py:403

↓ 2 callersMethodprep_outputs

Validate and prepare chain outputs, and save info about this run to memory. Args: inputs: Dictionary of chain inputs, including a

src/chains/chain.py:377

↓ 2 callersMethodreset

(self)

src/docstore/wikipedia.py:46

↓ 2 callersMethodrun_llm

Run the LLM.

src/llm_compiler/planner.py:216

↓ 2 callersMethodset_tasks

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

src/llm_compiler/task_fetching_unit.py:92

↓ 1 callersMethod__init__

(self)

src/callbacks/callbacks.py:11

↓ 1 callersMethod_aprocess_llm_result

( self, llm_output: str, run_manager: AsyncCallbackManagerForChainRun, )

src/chains/llm_math_chain.py:234

↓ 1 callersMethod_atake_next_step

Take a single step in the thought-action-observation loop. Override this to take control of how the agent makes and acts on choices.

src/executors/agent_executor.py:335

↓ 1 callersMethod_construct_scratchpad

Construct the scratchpad that lets the agent continue its thought process.

src/agents/agent.py:89

↓ 1 callersFunction_create_subset_model

Create a pydantic model with only a subset of model's fields.

src/tools/base.py:30

↓ 1 callersFunction_default_stringify_rule_for_arguments

(args)

src/llm_compiler/task_fetching_unit.py:12

↓ 1 callersFunction_find_tool

Find a tool by name. Args: tool_name: Name of the tool to find. Returns: Tool or StructuredTool.

src/llm_compiler/output_parser.py:80

↓ 1 callersMethod_format_contexts

contexts is a list of context each context is formatted as the description of _generate_context_for_replanner

src/llm_compiler/llm_compiler.py:191

↓ 1 callersMethod_generate_context_for_replanner

Formatted like this: ``` 1. action 1 Observation: xxx 2. action 2 Observation: yyy ... Thought

src/llm_compiler/llm_compiler.py:165

↓ 1 callersMethod_get_accuracy

(self, key: str)

src/utils/logger_utils.py:26

↓ 1 callersMethod_get_default_output_parser

Get default output parser for this class.

src/agents/agent.py:213

↓ 1 callersMethod_get_default_output_parser

( cls, llm: Optional[BaseLanguageModel] = None, **kwargs: Any )

src/agents/structured_chat_agent.py:64

↓ 1 callersFunction_get_dependencies_from_graph

Get dependencies from a graph.

src/llm_compiler/output_parser.py:97

↓ 1 callersFunction_get_filtered_args

Get the arguments from a function's signature.

src/tools/base.py:41

↓ 1 callersMethod_get_mean_latency

(self, key: str)

src/utils/logger_utils.py:22

↓ 1 callersFunction_get_verbosity

()

src/chains/chain.py:34

↓ 1 callersMethod_parse_joinner_output

We expect the joinner output format to be: ``` Thought: xxx Action: Finish/Replan(yyy) ``` Returns:

src/llm_compiler/llm_compiler.py:144

↓ 1 callersFunction_parse_llm_compiler_action_args

Parse arguments from a string.

src/llm_compiler/output_parser.py:64

↓ 1 callersMethod_preprocess_args

Replace dependency placeholders, i.e. ${1}, in task.args with the actual observation.

src/llm_compiler/task_fetching_unit.py:109

↓ 1 callersMethod_process_llm_result

( self, llm_output: str, run_manager: CallbackManagerForChainRun )

src/chains/llm_math_chain.py:214

↓ 1 callersFunction_replace_arg_mask_with_real_value

( args, dependencies: List[int], tasks: Dict[str, Task] )

src/llm_compiler/task_fetching_unit.py:19

↓ 1 callersMethod_take_next_step

Take a single step in the thought-action-observation loop. Override this to take control of how the agent makes and acts on choices.

src/executors/agent_executor.py:239

↓ 1 callersMethod_validate_inputs

Check that all inputs are present.

src/chains/chain.py:181

↓ 1 callersMethod_validate_outputs

(self, outputs: Dict[str, Any])

src/chains/chain.py:187

↓ 1 callersMethod_validate_tools

Validate that appropriate tools are passed in.

src/agents/agent.py:207

↓ 1 callersMethod_validate_tools

(cls, tools: Sequence[BaseTool])

src/agents/structured_chat_agent.py:60

↓ 1 callersMethodaapply

Utilize the LLM generate method for speed gains.

src/chains/llm_chain.py:244

↓ 1 callersMethodainvoke

( self, input: Union[str, Dict], config: Optional[RunnableConfig] = None, **kw

src/tools/base.py:107

↓ 1 callersMethodaplan

Given input, asynchronously decide what to do.

src/llm_compiler/planner.py:265

↓ 1 callersMethodaplan

Given input, decided what to do. Args: intermediate_steps: Steps the LLM has taken to date, along with observatio

src/agents/agent.py:127

↓ 1 callersMethodapply

Utilize the LLM generate method for speed gains.

src/chains/llm_chain.py:224

↓ 1 callersMethodaprep_prompts

Prepare prompts from inputs.

src/chains/llm_chain.py:198

↓ 1 callersMethodarun

(self, prompt: str, callbacks=None)

src/llm_compiler/llm_compiler.py:28

↓ 1 callersFunctionarun_and_time

helper function to run and time a function. Since function can error, we catch the error and return "ERROR" as the result

src/utils/evaluation_utils.py:40

↓ 1 callersMethodaschedule

Asynchronously listen to task_queue and schedule tasks as they arrive.

src/llm_compiler/task_fetching_unit.py:137

↓ 1 callersMethodcreate_prompt

( cls, tools: Sequence[BaseTool], prefix: str = PREFIX, suffix: str = SUFFIX,

src/agents/structured_chat_agent.py:74

↓ 1 callersFunctioncreate_schema_from_function

Create a pydantic schema from a function's signature. Args: model_name: Name to assign to the generated pydandic schema func: Func

src/tools/base.py:58

↓ 1 callersFunctiondefault_dependency_rule

(idx, args: str)

src/llm_compiler/output_parser.py:19

↓ 1 callersMethoddict

Dictionary representation of chain. Expects `Chain._chain_type` property to be implemented and for memory to be null. Ar

src/chains/chain.py:585

↓ 1 callersMethoddict

Return dictionary representation of agent.

src/agents/agent.py:60

↓ 1 callersMethodfinalize

(self)

src/llm_compiler/planner.py:141

↓ 1 callersFunctionflush_results

(save_path, results)

src/utils/logger_utils.py:69

↓ 1 callersMethodfrom_agent_and_tools

Create from agent and tools.

src/executors/agent_executor.py:104

↓ 1 callersMethodfrom_function

Initialize tool from a function.

src/tools/base.py:199

↓ 1 callersMethodfrom_llm_and_tools

Construct an agent from an LLM and tools.

src/agents/agent.py:217

↓ 1 callersMethodget_all_stats

(self)

src/llm_compiler/llm_compiler.py:117

↓ 1 callersMethodget_allowed_tools

(self)

src/agents/agent.py:71

↓ 1 callersFunctionget_configs

(args)

run_llm_compiler.py:108

↓ 1 callersFunctionget_dataset

(args)

run_llm_compiler.py:76

↓ 1 callersMethodget_stats

(self)

src/callbacks/callbacks.py:35

↓ 1 callersFunctionget_tools

(model_name, args)

run_llm_compiler.py:87

next →1–100 of 245, ranked by callers