phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-06-28 02:53:30 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Use inference APIs for running llama guard

Browse source 
Test Plan:

First, start a TGI container with `meta-llama/Llama-Guard-3-8B` model
serving on port 5099. See https://github.com/meta-llama/llama-stack/pull/53 and its
description for how.

Then run llama-stack with the following run config:

```
image_name: safety
docker_image: null
conda_env: safety
apis_to_serve:
- models
- inference
- shields
- safety
api_providers:
  inference:
    providers:
    - remote::tgi
  safety:
    providers:
    - meta-reference
  telemetry:
    provider_id: meta-reference
    config: {}
routing_table:
  inference:
  - provider_id: remote::tgi
    config:
      url: http://localhost:5099
      api_token: null
      hf_endpoint_name: null
    routing_key: Llama-Guard-3-8B
  safety:
  - provider_id: meta-reference
    config:
      llama_guard_shield:
        model: Llama-Guard-3-8B
        excluded_categories: []
        disable_input_check: false
        disable_output_check: false
      prompt_guard_shield: null
    routing_key: llama_guard
```

Now simply run `python -m llama_stack.apis.safety.client localhost
<port>` and check that the llama_guard shield calls run correctly. (The
injection_shield calls fail as expected since we have not set up a
router for them.)
This commit is contained in:
Ashwin Bharambe 2024-09-24 17:02:57 -07:00
parent c4534217c8
commit 0d2eb3bd25
9 changed files with 56 additions and 81 deletions
Download patch file Download diff file Expand all files Collapse all files

28
llama_stack/providers/impls/meta_reference/safety/safety.py
View file

@ -7,8 +7,10 @@
from llama_models.sku_list import resolve_model
from llama_stack.distribution.utils.model_utils import model_local_dir
from llama_stack.apis.inference import * # noqa: F403
from llama_stack.apis.safety import * # noqa: F403
from llama_models.llama3.api.datatypes import * # noqa: F403
from llama_stack.distribution.datatypes import Api
from llama_stack.providers.impls.meta_reference.safety.shields.base import (
OnViolationAction,
@ -34,20 +36,11 @@ def resolve_and_get_path(model_name: str) -> str:
class MetaReferenceSafetyImpl(Safety):
def __init__(self, config: SafetyConfig) -> None:
def __init__(self, config: SafetyConfig, deps) -> None:
self.config = config
self.inference_api = deps[Api.inference]
async def initialize(self) -> None:
shield_cfg = self.config.llama_guard_shield
if shield_cfg is not None:
model_dir = resolve_and_get_path(shield_cfg.model)
_ = LlamaGuardShield.instance(
model_dir=model_dir,
excluded_categories=shield_cfg.excluded_categories,
disable_input_check=shield_cfg.disable_input_check,
disable_output_check=shield_cfg.disable_output_check,
)
shield_cfg = self.config.prompt_guard_shield
if shield_cfg is not None:
model_dir = resolve_and_get_path(shield_cfg.model)
@ -91,11 +84,18 @@ class MetaReferenceSafetyImpl(Safety):
def get_shield_impl(self, typ: MetaReferenceShieldType) -> ShieldBase:
cfg = self.config
if typ == MetaReferenceShieldType.llama_guard:
cfg = cfg.llama_guard_shield
assert (
cfg.llama_guard_shield is not None
cfg is not None
), "Cannot use LlamaGuardShield since not present in config"
model_dir = resolve_and_get_path(cfg.llama_guard_shield.model)
return LlamaGuardShield.instance(model_dir=model_dir)
return LlamaGuardShield(
model=cfg.model,
inference_api=self.inference_api,
excluded_categories=cfg.excluded_categories,
disable_input_check=cfg.disable_input_check,
disable_output_check=cfg.disable_output_check,
)
elif typ == MetaReferenceShieldType.jailbreak_shield:
assert (
cfg.prompt_guard_shield is not None