Skip to content

Conversation

dineshyv
Copy link
Contributor

@dineshyv dineshyv commented Nov 13, 2024

❯llama-stack-client models register meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo
Successfully registered model meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo
❯ llama-stack-client models list
+----------------------+--------------------+---------------------------------------------+------------+
| identifier           | provider_id        | provider_resource_id                        | metadata   |
+======================+====================+=============================================+============+
| meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo| remote::together-0 | meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo | {}         |
+----------------------+--------------------+---------------------------------------------+------------+
| meta-llama/Llama-3.2-3B-Instruct-Turbo  | remote::together-0 | meta-llama/Llama-3.2-3B-Instruct-Turbo      | {}         |
+----------------------+--------------------+---------------------------------------------+------------+
❯ llama-stack-client models register dineshyv-model --provider-model-id=meta-llama/Meta-Llama-Guard-3-8B
Successfully registered model dineshyv-model
❯ llama-stack-client models list
+----------------------------------------------+--------------------+----------------------------------------------+------------+
| identifier                                   | provider_id        | provider_resource_id                         | metadata   |
+==============================================+====================+==============================================+============+
+----------------------------------------------+--------------------+----------------------------------------------+------------+
| dineshyv-model                               | remote::together-0 | meta-llama/Meta-Llama-Guard-3-8B             | {}         |
+----------------------------------------------+--------------------+----------------------------------------------+------------+


llama-stack-client shields register --shield-id llama-guard --provider-shield-id Llama-Guard-3-8B
identifier: llama-guard
params: {}
provider_id: inline::llama-guard-0
provider_resource_id: Llama-Guard-3-8B
type: shield

❯ llama-stack-client shields list
+--------------+----------+-----------------------+------------------------+--------+
| identifier   | params   | provider_id           | provider_resource_id   | type   |
+==============+==========+=======================+========================+========+
| llama-guard  | {}       | inline::llama-guard-0 | Llama-Guard-3-8B       | shield |
+--------------+----------+-----------------------+------------------------+--------+

 llama-stack-client datasets register \
  --dataset-id "alpaca" \
  --url "https://huggingface.co/datasets/tatsu-lab/alpaca" \
  --schema '{"instruction": {"type": "string"}, "input": {"type": "string"}, "output": {"type": "string"}}' --provider-id huggingface-0
❯ llama-stack-client datasets list
+--------------+---------------+---------------------------------------------------------------------------------+---------+
| identifier   | provider_id   | metadata                                                                        | type    |
+==============+===============+=================================================================================+=========+
| mmlu         | huggingface-0 | {'path': 'llama-stack/evals', 'name': 'evals__mmlu__details', 'split': 'train'} | dataset |
+--------------+---------------+---------------------------------------------------------------------------------+---------+
| alpaca       | huggingface-0 | {}                                                                              | dataset |
+--------------+---------------+---------------------------------------------------------------------------------+---------+

 llama-stack-client eval_tasks register \
    --eval-task-id "my-eval-task" \
    --dataset-id "alpaca" \
    --scoring-functions "accuracy" \
    --scoring-functions "f1_score" \
    --metadata '{"description": "My evaluation task", "version": "1.0"}'

llama-stack-client eval_tasks list
+--------------+---------------------+---------------------------------------------------------+------------------+------------------------+------------------------------------------------+-----------+
| dataset_id   | identifier          | metadata                                                | provider_id      | provider_resource_id   | scoring_functions                              | type      |
+==============+=====================+=========================================================+==================+========================+================================================+===========+
| mmlu         | meta-reference-mmlu | {}                                                      | meta-reference-0 | meta-reference-mmlu    | ['basic::regex_parser_multiple_choice_answer'] | eval_task |
+--------------+---------------------+---------------------------------------------------------+------------------+------------------------+------------------------------------------------+-----------+
| alpaca       | my-eval-task        | {'description': 'My evaluation task', 'version': '1.0'} | meta-reference-0 | my-eval-task           | ['accuracy', 'f1_score']                       | eval_task |
+--------------+---------------------+---------------------------------------------------------+------------------+------------------------+------------------------------------------------+-----------+

Failure cases:

llama-stack-client models register dineshyv-model --provider-model-id=Llama-Guard-78979
Failed to register model: Error code: 400 - {'detail': 'Invalid value: Unknown model: `Llama-Guard-78979`'}

@dineshyv dineshyv changed the title model create Resource create Nov 13, 2024
@dineshyv dineshyv merged commit 7e862d8 into main Nov 14, 2024
3 checks passed
@dineshyv dineshyv deleted the dineshyv/create branch November 14, 2024 04:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants