POST
/
api
/
parea
/
v1
/
experiments
p = Parea(api_key="PAREA_API_KEY")  # replace with your API key

exp_uuids = p.list_experiment_uuids(ListExperimentUUIDsFilters())
[
  {
    "created_at": "2024-04-15 22:19:13",
    "is_public": false,
    "metadata": {
      "dataset": "dev 123"
    },
    "name": "greeting",
    "num_samples": 6,
    "pinned_stats": [
      {
        "operation": "mean",
        "value": 0.176471,
        "var1": "levenshtein"
      },
      {
        "operation": "mean",
        "value": 0.00002,
        "var1": "Cost"
      },
      {
        "operation": "mean",
        "value": 11,
        "var1": "Total Tokens"
      },
      {
        "operation": "mean",
        "value": 9,
        "var1": "Output Tokens"
      },
      {
        "operation": "mean",
        "value": 2,
        "var1": "Input Tokens"
      },
      {
        "operation": "mean",
        "value": 0.65273,
        "var1": "Latency"
      }
    ],
    "project_uuid": "...",
    "run_name": "bijou-kail",
    "status": "completed",
    "uuid": "..."
  }
]
p = Parea(api_key="PAREA_API_KEY")  # replace with your API key

exp_uuids = p.list_experiment_uuids(ListExperimentUUIDsFilters())

Authorizations

x-user-id
string
header
required

Body

application/json
project_name
string | null

Name of the project to filter experiments by.

metadata_filter
object | null

Filters experiments to contain the given metadata. The metadata_filter is a dictionary where the keys are the metadata keys and the values are the metadata values to filter by.

experiment_name_filter
string | null

Filters experiments to contain the given string in their name.

run_name_filter
string | null

Filters experiments to contain the given string in their run name.

experiment_uuids
string[] | null

Filters experiments to contain the given UUIDs.

Response

200
application/json
Successful Response
name
string
required

Name of the experiment

project_uuid
string
required

UUID of the project associated with this experiment

uuid
string
required

UUID of the created experiment

created_at
string
required

Timestamp the experiment was created

status
enum<string>
required

Status of the experiment

Available options:
pending,
running,
completed,
failed
pinned_stats
object[]
required

List of population/dataset-level statistics. Typically contains mean of any reported scores, latencies, etc.

run_name
string | null

Name of the run. Must be unique for each experiment and can only contain alphanumeric characters, hyphens, and underscores.

is_public
boolean
default:
false

If experiment should be public

metadata
object | null

Any additional key-value pairs which provide context or are useful for filtering.

num_samples
integer | null

Number of samples in the experiment