llama-stack-mirror/llama_toolchain/agentic_system/event_logger.py
Ashwin Bharambe 7bc7785b0d
API Updates: fleshing out RAG APIs, introduce "llama stack" CLI command (#51)
* add tools to chat completion request

* use templates for generating system prompts

* Moved ToolPromptFormat and jinja templates to llama_models.llama3.api

* <WIP> memory changes

- inlined AgenticSystemInstanceConfig so API feels more ergonomic
- renamed it to AgentConfig, AgentInstance -> Agent
- added a MemoryConfig and `memory` parameter
- added `attachments` to input and `output_attachments` to the response

- some naming changes

* InterleavedTextAttachment -> InterleavedTextMedia, introduce memory tool

* flesh out memory banks API

* agentic loop has a RAG implementation

* faiss provider implementation

* memory client works

* re-work tool definitions, fix FastAPI issues, fix tool regressions

* fix agentic_system utils

* basic RAG seems to work

* small bug fixes for inline attachments

* Refactor custom tool execution utilities

* Bug fix, show memory retrieval steps in EventLogger

* No need for api_key for Remote providers

* add special unicode character ↵ to showcase newlines in model prompt templates

* remove api.endpoints imports

* combine datatypes.py and endpoints.py into api.py

* Attachment / add TTL api

* split batch_inference from inference

* minor import fixes

* use a single impl for ChatFormat.decode_assistant_mesage

* use interleaved_text_media_as_str() utilityt

* Fix api.datatypes imports

* Add blobfile for tiktoken

* Add ToolPromptFormat to ChatFormat.encode_message so that tools are encoded properly

* templates take optional --format={json,function_tag}

* Rag Updates

* Add `api build` subcommand -- WIP

* fix

* build + run image seems to work

* <WIP> adapters

* bunch more work to make adapters work

* api build works for conda now

* ollama remote adapter works

* Several smaller fixes to make adapters work

Also, reorganized the pattern of __init__ inside providers so
configuration can stay lightweight

* llama distribution -> llama stack + containers (WIP)

* All the new CLI for api + stack work

* Make Fireworks and Together into the Adapter format

* Some quick fixes to the CLI behavior to make it consistent

* Updated README phew

* Update cli_reference.md

* llama_toolchain/distribution -> llama_toolchain/core

* Add termcolor

* update paths

* Add a log just for consistency

* chmod +x scripts

* Fix api dependencies not getting added to configuration

* missing import lol

* Delete utils.py; move to agentic system

* Support downloading of URLs for attachments for code interpreter

* Simplify and generalize `llama api build` yay

* Update `llama stack configure` to be very simple also

* Fix stack start

* Allow building an "adhoc" distribution

* Remote `llama api []` subcommands

* Fixes to llama stack commands and update docs

* Update documentation again and add error messages to llama stack start

* llama stack start -> llama stack run

* Change name of build for less confusion

* Add pyopenapi fork to the repository, update RFC assets

* Remove conflicting annotation

* Added a "--raw" option for model template printing

---------

Co-authored-by: Hardik Shah <hjshah@fb.com>
Co-authored-by: Ashwin Bharambe <ashwin@meta.com>
Co-authored-by: Dalton Flanagan <6599399+dltn@users.noreply.github.com>
2024-09-03 22:39:39 -07:00

187 lines
7.1 KiB
Python

# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from typing import Optional
from llama_models.llama3.api.datatypes import * # noqa: F403
from llama_models.llama3.api.tool_utils import ToolUtils
from termcolor import cprint
from llama_toolchain.agentic_system.api import (
AgenticSystemTurnResponseEventType,
StepType,
)
class LogEvent:
def __init__(
self,
role: Optional[str] = None,
content: str = "",
end: str = "\n",
color="white",
):
self.role = role
self.content = content
self.color = color
self.end = "\n" if end is None else end
def __str__(self):
if self.role is not None:
return f"{self.role}> {self.content}"
else:
return f"{self.content}"
def print(self, flush=True):
cprint(f"{str(self)}", color=self.color, end=self.end, flush=flush)
EventType = AgenticSystemTurnResponseEventType
class EventLogger:
async def log(
self,
event_generator,
stream=True,
tool_prompt_format: ToolPromptFormat = ToolPromptFormat.json,
):
previous_event_type = None
previous_step_type = None
async for chunk in event_generator:
if not hasattr(chunk, "event"):
# Need to check for custom tool first
# since it does not produce event but instead
# a Message
if isinstance(chunk, ToolResponseMessage):
yield chunk, LogEvent(
role="CustomTool", content=chunk.content, color="grey"
)
continue
event = chunk.event
event_type = event.payload.event_type
if event_type in {
EventType.turn_start.value,
EventType.turn_complete.value,
}:
# Currently not logging any turn realted info
yield event, None
continue
step_type = event.payload.step_type
# handle safety
if (
step_type == StepType.shield_call
and event_type == EventType.step_complete.value
):
response = event.payload.step_details.response
if not response.is_violation:
yield event, LogEvent(
role=step_type, content="No Violation", color="magenta"
)
else:
yield event, LogEvent(
role=step_type,
content=f"{response.violation_type} {response.violation_return_message}",
color="red",
)
# handle inference
if step_type == StepType.inference:
if stream:
if event_type == EventType.step_start.value:
# TODO: Currently this event is never received
yield event, LogEvent(
role=step_type, content="", end="", color="yellow"
)
elif event_type == EventType.step_progress.value:
# HACK: if previous was not step/event was not inference's step_progress
# this is the first time we are getting model inference response
# aka equivalent to step_start for inference. Hence,
# start with "Model>".
if (
previous_event_type != EventType.step_progress.value
and previous_step_type != StepType.inference
):
yield event, LogEvent(
role=step_type, content="", end="", color="yellow"
)
if event.payload.tool_call_delta:
if isinstance(event.payload.tool_call_delta.content, str):
yield event, LogEvent(
role=None,
content=event.payload.tool_call_delta.content,
end="",
color="cyan",
)
else:
yield event, LogEvent(
role=None,
content=event.payload.model_response_text_delta,
end="",
color="yellow",
)
else:
# step_complete
yield event, LogEvent(role=None, content="")
else:
# Not streaming
if event_type == EventType.step_complete.value:
response = event.payload.step_details.model_response
if response.tool_calls:
content = ToolUtils.encode_tool_call(
response.tool_calls[0], tool_prompt_format
)
else:
content = response.content
yield event, LogEvent(
role=step_type,
content=content,
color="yellow",
)
# handle tool_execution
if (
step_type == StepType.tool_execution
and
# Only print tool calls and responses at the step_complete event
event_type == EventType.step_complete.value
):
details = event.payload.step_details
for t in details.tool_calls:
yield event, LogEvent(
role=step_type,
content=f"Tool:{t.tool_name} Args:{t.arguments}",
color="green",
)
for r in details.tool_responses:
yield event, LogEvent(
role=step_type,
content=f"Tool:{r.tool_name} Response:{r.content}",
color="green",
)
if (
step_type == StepType.memory_retrieval
and event_type == EventType.step_complete.value
):
details = event.payload.step_details
content = interleaved_text_media_as_str(details.inserted_context)
content = content[:200] + "..." if len(content) > 200 else content
yield event, LogEvent(
role=step_type,
content=f"Retrieved context from banks: {details.memory_bank_ids}.\n====\n{content}\n>",
color="cyan",
)
preivous_event_type = event_type
previous_step_type = step_type