mirror of
https://github.com/meta-llama/llama-stack.git
synced 2025-12-31 14:50:01 +00:00
## Summary
- Clean up datasets API to work with uri & files.
- Example client usage
```python
client.datasets.create(
uri="file://mydata.jsonl",
schema="jsonl_messages",
# optional
dataset_id="my_first_train_data"
)
# ecosystem enablement ( get huggingface dataset easily )
client.datasets.create(
uri="huggingface://tatsu-lab/alpaca",
uri_params={"split": "train"},
schema="jsonl_messages",
metadata={"label": "v/finetuning/test"},
)
# [ in the future with stronger story for authentication ]
client.datasets.create(
uri="s3://bucket/mydata.jsonl", <-- signed s3 url
uri_parms={...},
schema="jsonl_messages",
)
```
## Test Plan:
- Only API spec updates, implementation will be updated in follow PRs.
|
||
|---|---|---|
| .. | ||
| apis | ||
| cli | ||
| distribution | ||
| models/llama | ||
| providers | ||
| scripts | ||
| strong_typing | ||
| templates | ||
| __init__.py | ||
| env.py | ||
| log.py | ||
| schema_utils.py | ||