mirror of
https://github.com/meta-llama/llama-stack.git
synced 2025-10-04 04:04:14 +00:00
finetuning
This commit is contained in:
Ashwin Bharambe
parent
956f07b04c
commit
69ecf55de2
5 changed files with 1334 additions and 28 deletions
|
|
@ -172,13 +172,20 @@ components:
|
|||
tool_response_delta:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
response:
|
||||
type: string
|
||||
content:
|
||||
oneOf:
|
||||
- type: string
|
||||
- $ref: '#/components/schemas/Attachment'
|
||||
- items:
|
||||
oneOf:
|
||||
- type: string
|
||||
- $ref: '#/components/schemas/Attachment'
|
||||
type: array
|
||||
tool_name:
|
||||
type: string
|
||||
required:
|
||||
- tool_name
|
||||
- response
|
||||
- content
|
||||
type: object
|
||||
violation:
|
||||
additionalProperties: false
|
||||
|
|
@ -274,13 +281,20 @@ components:
|
|||
items:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
response:
|
||||
type: string
|
||||
content:
|
||||
oneOf:
|
||||
- type: string
|
||||
- $ref: '#/components/schemas/Attachment'
|
||||
- items:
|
||||
oneOf:
|
||||
- type: string
|
||||
- $ref: '#/components/schemas/Attachment'
|
||||
type: array
|
||||
tool_name:
|
||||
type: string
|
||||
required:
|
||||
- tool_name
|
||||
- response
|
||||
- content
|
||||
type: object
|
||||
type: array
|
||||
uuid:
|
||||
|
|
@ -690,6 +704,224 @@ components:
|
|||
- text_delta
|
||||
title: streamed completion response.
|
||||
type: object
|
||||
CreateDatasetRequest:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
dataset:
|
||||
$ref: '#/components/schemas/Dataset'
|
||||
uuid:
|
||||
type: string
|
||||
required:
|
||||
- uuid
|
||||
- dataset
|
||||
title: Request to create a dataset.
|
||||
type: object
|
||||
Dataset:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
columns:
|
||||
additionalProperties:
|
||||
enum:
|
||||
- dialog
|
||||
- text
|
||||
- media
|
||||
- number
|
||||
- json
|
||||
type: string
|
||||
type: object
|
||||
content_url:
|
||||
$ref: '#/components/schemas/URL'
|
||||
metadata:
|
||||
additionalProperties:
|
||||
oneOf:
|
||||
- type: 'null'
|
||||
- type: boolean
|
||||
- type: number
|
||||
- type: string
|
||||
- type: array
|
||||
- type: object
|
||||
type: object
|
||||
required:
|
||||
- columns
|
||||
- content_url
|
||||
- metadata
|
||||
title: Dataset to be used for training or evaluating language models.
|
||||
type: object
|
||||
FinetuningJobLogStream:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
job_uuid:
|
||||
type: string
|
||||
log_lines:
|
||||
items:
|
||||
type: string
|
||||
type: array
|
||||
required:
|
||||
- job_uuid
|
||||
- log_lines
|
||||
title: Stream of logs from a finetuning job.
|
||||
type: object
|
||||
FinetuningJobStatusResponse:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
completed_at:
|
||||
format: date-time
|
||||
type: string
|
||||
job_uuid:
|
||||
type: string
|
||||
resources_allocated:
|
||||
additionalProperties:
|
||||
oneOf:
|
||||
- type: 'null'
|
||||
- type: boolean
|
||||
- type: number
|
||||
- type: string
|
||||
- type: array
|
||||
- type: object
|
||||
type: object
|
||||
scheduled_at:
|
||||
format: date-time
|
||||
type: string
|
||||
started_at:
|
||||
format: date-time
|
||||
type: string
|
||||
status:
|
||||
enum:
|
||||
- running
|
||||
- completed
|
||||
- failed
|
||||
- scheduled
|
||||
type: string
|
||||
required:
|
||||
- job_uuid
|
||||
- status
|
||||
title: Status of a finetuning job.
|
||||
type: object
|
||||
FinetuningTrainRequest:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
algorithm:
|
||||
enum:
|
||||
- full
|
||||
- lora
|
||||
- qlora
|
||||
- dora
|
||||
type: string
|
||||
algorithm_config:
|
||||
oneOf:
|
||||
- $ref: '#/components/schemas/LoraFinetuningConfig'
|
||||
- additionalProperties: false
|
||||
properties:
|
||||
alpha:
|
||||
type: integer
|
||||
apply_lora_to_mlp:
|
||||
type: boolean
|
||||
apply_lora_to_output:
|
||||
type: boolean
|
||||
lora_attn_modules:
|
||||
items:
|
||||
type: string
|
||||
type: array
|
||||
rank:
|
||||
type: integer
|
||||
required:
|
||||
- lora_attn_modules
|
||||
- apply_lora_to_mlp
|
||||
- apply_lora_to_output
|
||||
- rank
|
||||
- alpha
|
||||
type: object
|
||||
- additionalProperties: false
|
||||
properties:
|
||||
alpha:
|
||||
type: integer
|
||||
apply_lora_to_mlp:
|
||||
type: boolean
|
||||
apply_lora_to_output:
|
||||
type: boolean
|
||||
lora_attn_modules:
|
||||
items:
|
||||
type: string
|
||||
type: array
|
||||
rank:
|
||||
type: integer
|
||||
required:
|
||||
- lora_attn_modules
|
||||
- apply_lora_to_mlp
|
||||
- apply_lora_to_output
|
||||
- rank
|
||||
- alpha
|
||||
type: object
|
||||
dataset:
|
||||
$ref: '#/components/schemas/Dataset'
|
||||
hyperparam_search_config:
|
||||
additionalProperties:
|
||||
oneOf:
|
||||
- type: 'null'
|
||||
- type: boolean
|
||||
- type: number
|
||||
- type: string
|
||||
- type: array
|
||||
- type: object
|
||||
type: object
|
||||
job_uuid:
|
||||
type: string
|
||||
logger_config:
|
||||
additionalProperties:
|
||||
oneOf:
|
||||
- type: 'null'
|
||||
- type: boolean
|
||||
- type: number
|
||||
- type: string
|
||||
- type: array
|
||||
- type: object
|
||||
type: object
|
||||
model:
|
||||
enum:
|
||||
- llama3_8b
|
||||
- llama3_70b
|
||||
type: string
|
||||
optimizer_config:
|
||||
$ref: '#/components/schemas/OptimizerConfig'
|
||||
training_config:
|
||||
$ref: '#/components/schemas/TrainingConfig'
|
||||
validation_dataset:
|
||||
$ref: '#/components/schemas/Dataset'
|
||||
required:
|
||||
- job_uuid
|
||||
- model
|
||||
- dataset
|
||||
- validation_dataset
|
||||
- algorithm
|
||||
- algorithm_config
|
||||
- optimizer_config
|
||||
- training_config
|
||||
- hyperparam_search_config
|
||||
- logger_config
|
||||
title: Request to finetune a model.
|
||||
type: object
|
||||
LoraFinetuningConfig:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
alpha:
|
||||
type: integer
|
||||
apply_lora_to_mlp:
|
||||
type: boolean
|
||||
apply_lora_to_output:
|
||||
type: boolean
|
||||
lora_attn_modules:
|
||||
items:
|
||||
type: string
|
||||
type: array
|
||||
rank:
|
||||
type: integer
|
||||
required:
|
||||
- lora_attn_modules
|
||||
- apply_lora_to_mlp
|
||||
- apply_lora_to_output
|
||||
- rank
|
||||
- alpha
|
||||
type: object
|
||||
Message:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
|
|
@ -735,13 +967,20 @@ components:
|
|||
items:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
response:
|
||||
type: string
|
||||
content:
|
||||
oneOf:
|
||||
- type: string
|
||||
- $ref: '#/components/schemas/Attachment'
|
||||
- items:
|
||||
oneOf:
|
||||
- type: string
|
||||
- $ref: '#/components/schemas/Attachment'
|
||||
type: array
|
||||
tool_name:
|
||||
type: string
|
||||
required:
|
||||
- tool_name
|
||||
- response
|
||||
- content
|
||||
type: object
|
||||
type: array
|
||||
required:
|
||||
|
|
@ -750,6 +989,27 @@ components:
|
|||
- tool_calls
|
||||
- tool_responses
|
||||
type: object
|
||||
OptimizerConfig:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
lr:
|
||||
type: number
|
||||
lr_min:
|
||||
type: number
|
||||
optimizer_type:
|
||||
enum:
|
||||
- adam
|
||||
- adamw
|
||||
- sgd
|
||||
type: string
|
||||
weight_decay:
|
||||
type: number
|
||||
required:
|
||||
- optimizer_type
|
||||
- lr
|
||||
- lr_min
|
||||
- weight_decay
|
||||
type: object
|
||||
RewardScoringRequest:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
|
|
@ -885,6 +1145,32 @@ components:
|
|||
title: Response from the synthetic data generation. Batch of (prompt, response,
|
||||
score) tuples that pass the threshold.
|
||||
type: object
|
||||
TrainingConfig:
|
||||
additionalProperties: false
|
||||
properties:
|
||||
batch_size:
|
||||
type: integer
|
||||
enable_activation_checkpointing:
|
||||
type: boolean
|
||||
fsdp_cpu_offload:
|
||||
type: boolean
|
||||
memory_efficient_fsdp_wrap:
|
||||
type: boolean
|
||||
n_epochs:
|
||||
type: integer
|
||||
n_iters:
|
||||
type: integer
|
||||
shuffle:
|
||||
type: boolean
|
||||
required:
|
||||
- n_epochs
|
||||
- batch_size
|
||||
- shuffle
|
||||
- n_iters
|
||||
- enable_activation_checkpointing
|
||||
- memory_efficient_fsdp_wrap
|
||||
- fsdp_cpu_offload
|
||||
type: object
|
||||
URL:
|
||||
format: uri
|
||||
pattern: ^(https?://|file://|data:)
|
||||
|
|
@ -989,6 +1275,98 @@ paths:
|
|||
description: Normal completion response. **OR** streamed completion response.
|
||||
tags:
|
||||
- Inference
|
||||
/datasets/create:
|
||||
post:
|
||||
parameters: []
|
||||
requestBody:
|
||||
content:
|
||||
application/json:
|
||||
schema:
|
||||
$ref: '#/components/schemas/CreateDatasetRequest'
|
||||
required: true
|
||||
responses:
|
||||
'200':
|
||||
description: OK
|
||||
tags:
|
||||
- Datasets
|
||||
/datasets/delete:
|
||||
delete:
|
||||
parameters:
|
||||
- in: query
|
||||
name: dataset_id
|
||||
required: true
|
||||
schema:
|
||||
type: string
|
||||
responses:
|
||||
'200':
|
||||
description: OK
|
||||
tags:
|
||||
- Datasets
|
||||
/datasets/get:
|
||||
get:
|
||||
parameters:
|
||||
- in: query
|
||||
name: dataset_id
|
||||
required: true
|
||||
schema:
|
||||
type: string
|
||||
responses:
|
||||
'200':
|
||||
content:
|
||||
application/json:
|
||||
schema:
|
||||
$ref: '#/components/schemas/Dataset'
|
||||
description: OK
|
||||
tags:
|
||||
- Datasets
|
||||
/finetuning/job/logs:
|
||||
get:
|
||||
parameters:
|
||||
- in: query
|
||||
name: job_uuid
|
||||
required: true
|
||||
schema:
|
||||
type: string
|
||||
responses:
|
||||
'200':
|
||||
content:
|
||||
application/json:
|
||||
schema:
|
||||
$ref: '#/components/schemas/FinetuningJobLogStream'
|
||||
description: OK
|
||||
tags:
|
||||
- Finetuning
|
||||
/finetuning/job/status:
|
||||
get:
|
||||
parameters:
|
||||
- in: query
|
||||
name: job_uuid
|
||||
required: true
|
||||
schema:
|
||||
type: string
|
||||
responses:
|
||||
'200':
|
||||
content:
|
||||
application/json:
|
||||
schema:
|
||||
$ref: '#/components/schemas/FinetuningJobStatusResponse'
|
||||
description: OK
|
||||
tags:
|
||||
- Finetuning
|
||||
/finetuning/text_generation/train:
|
||||
post:
|
||||
parameters: []
|
||||
requestBody:
|
||||
content:
|
||||
application/json:
|
||||
schema:
|
||||
$ref: '#/components/schemas/FinetuningTrainRequest'
|
||||
required: true
|
||||
responses:
|
||||
'200':
|
||||
description: OK
|
||||
tags:
|
||||
- Finetuning
|
||||
/reward_scoring/score:
|
||||
post:
|
||||
parameters: []
|
||||
|
|
@ -1030,10 +1408,12 @@ security:
|
|||
servers:
|
||||
- url: http://llama.meta.com
|
||||
tags:
|
||||
- name: RewardScoring
|
||||
- name: Inference
|
||||
- name: SyntheticDataGeneration
|
||||
- name: RewardScoring
|
||||
- name: AgenticSystem
|
||||
- name: Finetuning
|
||||
- name: Inference
|
||||
- name: Datasets
|
||||
- description: <SchemaDefinition schemaRef="#/components/schemas/ShieldConfig" />
|
||||
name: ShieldConfig
|
||||
- description: <SchemaDefinition schemaRef="#/components/schemas/AgenticSystemCreateRequest"
|
||||
|
|
@ -1072,6 +1452,27 @@ tags:
|
|||
<SchemaDefinition schemaRef="#/components/schemas/AgenticSystemExecuteResponseStreamChunk"
|
||||
/>'
|
||||
name: AgenticSystemExecuteResponseStreamChunk
|
||||
- description: 'Request to create a dataset.
|
||||
|
||||
|
||||
<SchemaDefinition schemaRef="#/components/schemas/CreateDatasetRequest" />'
|
||||
name: CreateDatasetRequest
|
||||
- description: 'Dataset to be used for training or evaluating language models.
|
||||
|
||||
|
||||
<SchemaDefinition schemaRef="#/components/schemas/Dataset" />'
|
||||
name: Dataset
|
||||
- description: 'Status of a finetuning job.
|
||||
|
||||
|
||||
<SchemaDefinition schemaRef="#/components/schemas/FinetuningJobStatusResponse"
|
||||
/>'
|
||||
name: FinetuningJobStatusResponse
|
||||
- description: 'Stream of logs from a finetuning job.
|
||||
|
||||
|
||||
<SchemaDefinition schemaRef="#/components/schemas/FinetuningJobLogStream" />'
|
||||
name: FinetuningJobLogStream
|
||||
- description: <SchemaDefinition schemaRef="#/components/schemas/ChatCompletionRequest"
|
||||
/>
|
||||
name: ChatCompletionRequest
|
||||
|
|
@ -1127,10 +1528,25 @@ tags:
|
|||
|
||||
<SchemaDefinition schemaRef="#/components/schemas/RewardScoringResponse" />'
|
||||
name: RewardScoringResponse
|
||||
- description: 'Request to finetune a model.
|
||||
|
||||
|
||||
<SchemaDefinition schemaRef="#/components/schemas/FinetuningTrainRequest" />'
|
||||
name: FinetuningTrainRequest
|
||||
- description: <SchemaDefinition schemaRef="#/components/schemas/LoraFinetuningConfig"
|
||||
/>
|
||||
name: LoraFinetuningConfig
|
||||
- description: <SchemaDefinition schemaRef="#/components/schemas/OptimizerConfig"
|
||||
/>
|
||||
name: OptimizerConfig
|
||||
- description: <SchemaDefinition schemaRef="#/components/schemas/TrainingConfig" />
|
||||
name: TrainingConfig
|
||||
x-tagGroups:
|
||||
- name: Operations
|
||||
tags:
|
||||
- AgenticSystem
|
||||
- Datasets
|
||||
- Finetuning
|
||||
- Inference
|
||||
- RewardScoring
|
||||
- SyntheticDataGeneration
|
||||
|
|
@ -1149,10 +1565,18 @@ x-tagGroups:
|
|||
- CompletionRequest
|
||||
- CompletionResponse
|
||||
- CompletionResponseStreamChunk
|
||||
- CreateDatasetRequest
|
||||
- Dataset
|
||||
- FinetuningJobLogStream
|
||||
- FinetuningJobStatusResponse
|
||||
- FinetuningTrainRequest
|
||||
- LoraFinetuningConfig
|
||||
- Message
|
||||
- OptimizerConfig
|
||||
- RewardScoringRequest
|
||||
- RewardScoringResponse
|
||||
- ShieldConfig
|
||||
- SyntheticDataGenerationRequest
|
||||
- SyntheticDataGenerationResponse
|
||||
- TrainingConfig
|
||||
- URL
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue