phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-06-27 18:50:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Add support for Structured Output / Guided decoding (#281)

Browse source 
Added support for structured output in the API and added a reference implementation for meta-reference.

A few notes:

* Two formats are specified in the API: Json schema and EBNF based grammar
* Implementation only supports Json for now
We use lm-format-enhancer to provide the implementation right now but may change this especially because BNF grammars aren't supported by that library.
Fireworks has support for structured output and Together has limited supported for it too. Subsequent PRs will add these changes. We would like all our inference providers to provide structured output for llama models since it is an extremely important and highly sought-after need by the developers.
This commit is contained in:
Ashwin Bharambe 2024-10-22 12:53:34 -07:00 • committed by GitHub
parent 4c3d33e6f4
commit c06718fbd5
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
16 changed files with 257 additions and 25 deletions
Download patch file Download diff file Expand all files Collapse all files

38
llama_stack/providers/registry/inference.py
View file

@ -9,36 +9,36 @@ from typing import List
from llama_stack.distribution.datatypes import * # noqa: F403
META_REFERENCE_DEPS = [
"accelerate",
"blobfile",
"fairscale",
"torch",
"torchvision",
"transformers",
"zmq",
"lm-format-enforcer",
]
def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.inference,
provider_type="meta-reference",
pip_packages=[
"accelerate",
"blobfile",
"fairscale",
"torch",
"torchvision",
"transformers",
"zmq",
],
pip_packages=META_REFERENCE_DEPS,
module="llama_stack.providers.impls.meta_reference.inference",
config_class="llama_stack.providers.impls.meta_reference.inference.MetaReferenceInferenceConfig",
),
InlineProviderSpec(
api=Api.inference,
provider_type="meta-reference-quantized",
pip_packages=[
"accelerate",
"blobfile",
"fairscale",
"fbgemm-gpu==0.8.0",
"torch",
"torchvision",
"transformers",
"zmq",
],
pip_packages=(
META_REFERENCE_DEPS
+ [
"fbgemm-gpu==0.8.0",
]
),
module="llama_stack.providers.impls.meta_reference.inference",
config_class="llama_stack.providers.impls.meta_reference.inference.MetaReferenceQuantizedInferenceConfig",
),