Judge results under a decision's objective
curl --request POST \
--url https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data '
{
"candidate": {
"run_ids": [
"<string>"
]
},
"base": {
"run_ids": [
"<string>"
]
},
"cases": {
"limit": 100,
"cursor": "<string>"
}
}
'import requests
url = "https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports"
payload = {
"candidate": { "run_ids": ["<string>"] },
"base": { "run_ids": ["<string>"] },
"cases": {
"limit": 100,
"cursor": "<string>"
}
}
headers = {
"Authorization": "Bearer <token>",
"Content-Type": "application/json"
}
response = requests.post(url, json=payload, headers=headers)
print(response.text)const options = {
method: 'POST',
headers: {Authorization: 'Bearer <token>', 'Content-Type': 'application/json'},
body: JSON.stringify({
candidate: {run_ids: ['<string>']},
base: {run_ids: ['<string>']},
cases: {limit: 100, cursor: '<string>'}
})
};
fetch('https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports', options)
.then(res => res.json())
.then(res => console.log(res))
.catch(err => console.error(err));<?php
$curl = curl_init();
curl_setopt_array($curl, [
CURLOPT_URL => "https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports",
CURLOPT_RETURNTRANSFER => true,
CURLOPT_ENCODING => "",
CURLOPT_MAXREDIRS => 10,
CURLOPT_TIMEOUT => 30,
CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
CURLOPT_CUSTOMREQUEST => "POST",
CURLOPT_POSTFIELDS => json_encode([
'candidate' => [
'run_ids' => [
'<string>'
]
],
'base' => [
'run_ids' => [
'<string>'
]
],
'cases' => [
'limit' => 100,
'cursor' => '<string>'
]
]),
CURLOPT_HTTPHEADER => [
"Authorization: Bearer <token>",
"Content-Type: application/json"
],
]);
$response = curl_exec($curl);
$err = curl_error($curl);
curl_close($curl);
if ($err) {
echo "cURL Error #:" . $err;
} else {
echo $response;
}package main
import (
"fmt"
"strings"
"net/http"
"io"
)
func main() {
url := "https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports"
payload := strings.NewReader("{\n \"candidate\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"base\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"cases\": {\n \"limit\": 100,\n \"cursor\": \"<string>\"\n }\n}")
req, _ := http.NewRequest("POST", url, payload)
req.Header.Add("Authorization", "Bearer <token>")
req.Header.Add("Content-Type", "application/json")
res, _ := http.DefaultClient.Do(req)
defer res.Body.Close()
body, _ := io.ReadAll(res.Body)
fmt.Println(string(body))
}HttpResponse<String> response = Unirest.post("https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports")
.header("Authorization", "Bearer <token>")
.header("Content-Type", "application/json")
.body("{\n \"candidate\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"base\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"cases\": {\n \"limit\": 100,\n \"cursor\": \"<string>\"\n }\n}")
.asString();require 'uri'
require 'net/http'
url = URI("https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports")
http = Net::HTTP.new(url.host, url.port)
http.use_ssl = true
request = Net::HTTP::Post.new(url)
request["Authorization"] = 'Bearer <token>'
request["Content-Type"] = 'application/json'
request.body = "{\n \"candidate\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"base\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"cases\": {\n \"limit\": 100,\n \"cursor\": \"<string>\"\n }\n}"
response = http.request(request)
puts response.read_body{
"objective": {
"source": "stored",
"revision": 2,
"objective_hash": "<string>",
"effective_hash": "<string>",
"unknown_outcomes": [
"<string>"
]
},
"base": {
"n": 1,
"loss": 1,
"loss_per_case": 1,
"target_met": true,
"macro_f1": 0.5,
"outcomes": [
{
"outcome": "<string>",
"expected": 1,
"given": 1,
"precision": 0.5,
"recall": 0.5,
"f1": 0.5
}
],
"by_mistake": [
{
"expected": "<string>",
"got": "<string>",
"n": 2,
"loss": 1,
"examples": [
"<string>"
]
}
],
"by_mistake_total": 1,
"by_question": [
{
"question": "<string>",
"loss": 1,
"share": 0.5,
"examples": [
"<string>"
]
}
],
"missing_criterion": {
"n": 1,
"loss": 1,
"examples": [
"<string>"
]
},
"failed": {
"n": 1,
"loss": 1
},
"near_cuts": [
{
"aggregation": "<string>",
"at": 123,
"lower_it": {
"n": 1,
"loss": 1
},
"raise_it": {
"n": 1,
"loss": 1
},
"examples": [
"<string>"
]
}
],
"severe": {
"n": 1,
"human": 1
},
"cases": [
{
"expected": "<string>",
"got": "<string>",
"failed": true,
"loss": 1,
"severe": true,
"human": true,
"margin": {
"value": 123,
"cut": 123,
"distance": 1,
"adjacent": true
},
"attributed": true,
"fixing_flips": [
"<string>"
],
"flips_total": 1,
"answer_conflicts": [
"<string>"
],
"suspects": [
"<string>"
],
"answers": {},
"evaluation_id": "<string>",
"item_id": "<string>"
}
],
"cases_total": 1,
"attributed": 1
},
"candidate": {
"n": 1,
"loss": 1,
"loss_per_case": 1,
"target_met": true,
"macro_f1": 0.5,
"outcomes": [
{
"outcome": "<string>",
"expected": 1,
"given": 1,
"precision": 0.5,
"recall": 0.5,
"f1": 0.5
}
],
"by_mistake": [
{
"expected": "<string>",
"got": "<string>",
"n": 2,
"loss": 1,
"examples": [
"<string>"
]
}
],
"by_mistake_total": 1,
"by_question": [
{
"question": "<string>",
"loss": 1,
"share": 0.5,
"examples": [
"<string>"
]
}
],
"missing_criterion": {
"n": 1,
"loss": 1,
"examples": [
"<string>"
]
},
"failed": {
"n": 1,
"loss": 1
},
"near_cuts": [
{
"aggregation": "<string>",
"at": 123,
"lower_it": {
"n": 1,
"loss": 1
},
"raise_it": {
"n": 1,
"loss": 1
},
"examples": [
"<string>"
]
}
],
"severe": {
"n": 1,
"human": 1
},
"cases": [
{
"expected": "<string>",
"got": "<string>",
"failed": true,
"loss": 1,
"severe": true,
"human": true,
"margin": {
"value": 123,
"cut": 123,
"distance": 1,
"adjacent": true
},
"attributed": true,
"fixing_flips": [
"<string>"
],
"flips_total": 1,
"answer_conflicts": [
"<string>"
],
"suspects": [
"<string>"
],
"answers": {},
"evaluation_id": "<string>",
"item_id": "<string>"
}
],
"cases_total": 1,
"attributed": 1
},
"accepted": true,
"reasons": [
"<string>"
],
"new_severe_human": [
"<string>"
],
"environment": "production",
"decision_id": "<string>",
"excluded": {
"unlabeled": 1,
"failed_transient": 1,
"base_only": 1,
"outcome_not_in_result": 1,
"no_base": 1
},
"cases": {
"filter": {
"kind": "all",
"expected": "<string>",
"got": "<string>",
"question": "<string>",
"at": 123
},
"total": 1,
"rows": [
{
"item_id": "<string>",
"expected": "<string>",
"got": "<string>",
"loss": 1,
"severe": true,
"base": {
"got": "<string>",
"loss": 1,
"severe": true
}
}
],
"next_cursor": "<string>"
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}objective
Judge results under a decision's objective
Judges one version’s labeled results under the decision’s objective and, with a base, compares the two on the same items: the loss report a preview returns. Each mistake’s loss is its severity under the objective (every mistake normal when none is set), weighted so each expected outcome counts equally however many items it has; loss_per_case is their mean. accepted (lower loss per case, no new severe mistake on a human-labeled case) and reasons say whether the candidate is better. Each side is runs by run_ids (from evaluate with items, one structure per side) or results passed inline with their structure or structure_hash. Items match by item_id; the candidate’s labels are used; unlabeled items, outcomes neither version can produce, items the base lacks and retryable failures on either side are left out and counted in excluded; a permanent failure counts as a mistake. List a resend’s run beside the original: an item’s result that didn’t fail replaces its failed one. The report lists the worst mistakes, with example ids on every roll-up row. Send cases to page through each item’s loss: filter.kind is all, wrong (default), right, failed, severe, changed, cell, question, near_cut or missing_criterion. Runs no engine and spends nothing; the word is loss, not money. An unknown or expired run is 404 RUN_UNKNOWN. A version whose result is a value, not outcomes, is 400 OUTCOMES_INVALID.
POST
/
v1
/
decisions
/
{decision_id}
/
loss-reports
Judge results under a decision's objective
curl --request POST \
--url https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data '
{
"candidate": {
"run_ids": [
"<string>"
]
},
"base": {
"run_ids": [
"<string>"
]
},
"cases": {
"limit": 100,
"cursor": "<string>"
}
}
'import requests
url = "https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports"
payload = {
"candidate": { "run_ids": ["<string>"] },
"base": { "run_ids": ["<string>"] },
"cases": {
"limit": 100,
"cursor": "<string>"
}
}
headers = {
"Authorization": "Bearer <token>",
"Content-Type": "application/json"
}
response = requests.post(url, json=payload, headers=headers)
print(response.text)const options = {
method: 'POST',
headers: {Authorization: 'Bearer <token>', 'Content-Type': 'application/json'},
body: JSON.stringify({
candidate: {run_ids: ['<string>']},
base: {run_ids: ['<string>']},
cases: {limit: 100, cursor: '<string>'}
})
};
fetch('https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports', options)
.then(res => res.json())
.then(res => console.log(res))
.catch(err => console.error(err));<?php
$curl = curl_init();
curl_setopt_array($curl, [
CURLOPT_URL => "https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports",
CURLOPT_RETURNTRANSFER => true,
CURLOPT_ENCODING => "",
CURLOPT_MAXREDIRS => 10,
CURLOPT_TIMEOUT => 30,
CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
CURLOPT_CUSTOMREQUEST => "POST",
CURLOPT_POSTFIELDS => json_encode([
'candidate' => [
'run_ids' => [
'<string>'
]
],
'base' => [
'run_ids' => [
'<string>'
]
],
'cases' => [
'limit' => 100,
'cursor' => '<string>'
]
]),
CURLOPT_HTTPHEADER => [
"Authorization: Bearer <token>",
"Content-Type: application/json"
],
]);
$response = curl_exec($curl);
$err = curl_error($curl);
curl_close($curl);
if ($err) {
echo "cURL Error #:" . $err;
} else {
echo $response;
}package main
import (
"fmt"
"strings"
"net/http"
"io"
)
func main() {
url := "https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports"
payload := strings.NewReader("{\n \"candidate\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"base\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"cases\": {\n \"limit\": 100,\n \"cursor\": \"<string>\"\n }\n}")
req, _ := http.NewRequest("POST", url, payload)
req.Header.Add("Authorization", "Bearer <token>")
req.Header.Add("Content-Type", "application/json")
res, _ := http.DefaultClient.Do(req)
defer res.Body.Close()
body, _ := io.ReadAll(res.Body)
fmt.Println(string(body))
}HttpResponse<String> response = Unirest.post("https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports")
.header("Authorization", "Bearer <token>")
.header("Content-Type", "application/json")
.body("{\n \"candidate\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"base\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"cases\": {\n \"limit\": 100,\n \"cursor\": \"<string>\"\n }\n}")
.asString();require 'uri'
require 'net/http'
url = URI("https://api.moonbounce.ai/v1/decisions/{decision_id}/loss-reports")
http = Net::HTTP.new(url.host, url.port)
http.use_ssl = true
request = Net::HTTP::Post.new(url)
request["Authorization"] = 'Bearer <token>'
request["Content-Type"] = 'application/json'
request.body = "{\n \"candidate\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"base\": {\n \"run_ids\": [\n \"<string>\"\n ]\n },\n \"cases\": {\n \"limit\": 100,\n \"cursor\": \"<string>\"\n }\n}"
response = http.request(request)
puts response.read_body{
"objective": {
"source": "stored",
"revision": 2,
"objective_hash": "<string>",
"effective_hash": "<string>",
"unknown_outcomes": [
"<string>"
]
},
"base": {
"n": 1,
"loss": 1,
"loss_per_case": 1,
"target_met": true,
"macro_f1": 0.5,
"outcomes": [
{
"outcome": "<string>",
"expected": 1,
"given": 1,
"precision": 0.5,
"recall": 0.5,
"f1": 0.5
}
],
"by_mistake": [
{
"expected": "<string>",
"got": "<string>",
"n": 2,
"loss": 1,
"examples": [
"<string>"
]
}
],
"by_mistake_total": 1,
"by_question": [
{
"question": "<string>",
"loss": 1,
"share": 0.5,
"examples": [
"<string>"
]
}
],
"missing_criterion": {
"n": 1,
"loss": 1,
"examples": [
"<string>"
]
},
"failed": {
"n": 1,
"loss": 1
},
"near_cuts": [
{
"aggregation": "<string>",
"at": 123,
"lower_it": {
"n": 1,
"loss": 1
},
"raise_it": {
"n": 1,
"loss": 1
},
"examples": [
"<string>"
]
}
],
"severe": {
"n": 1,
"human": 1
},
"cases": [
{
"expected": "<string>",
"got": "<string>",
"failed": true,
"loss": 1,
"severe": true,
"human": true,
"margin": {
"value": 123,
"cut": 123,
"distance": 1,
"adjacent": true
},
"attributed": true,
"fixing_flips": [
"<string>"
],
"flips_total": 1,
"answer_conflicts": [
"<string>"
],
"suspects": [
"<string>"
],
"answers": {},
"evaluation_id": "<string>",
"item_id": "<string>"
}
],
"cases_total": 1,
"attributed": 1
},
"candidate": {
"n": 1,
"loss": 1,
"loss_per_case": 1,
"target_met": true,
"macro_f1": 0.5,
"outcomes": [
{
"outcome": "<string>",
"expected": 1,
"given": 1,
"precision": 0.5,
"recall": 0.5,
"f1": 0.5
}
],
"by_mistake": [
{
"expected": "<string>",
"got": "<string>",
"n": 2,
"loss": 1,
"examples": [
"<string>"
]
}
],
"by_mistake_total": 1,
"by_question": [
{
"question": "<string>",
"loss": 1,
"share": 0.5,
"examples": [
"<string>"
]
}
],
"missing_criterion": {
"n": 1,
"loss": 1,
"examples": [
"<string>"
]
},
"failed": {
"n": 1,
"loss": 1
},
"near_cuts": [
{
"aggregation": "<string>",
"at": 123,
"lower_it": {
"n": 1,
"loss": 1
},
"raise_it": {
"n": 1,
"loss": 1
},
"examples": [
"<string>"
]
}
],
"severe": {
"n": 1,
"human": 1
},
"cases": [
{
"expected": "<string>",
"got": "<string>",
"failed": true,
"loss": 1,
"severe": true,
"human": true,
"margin": {
"value": 123,
"cut": 123,
"distance": 1,
"adjacent": true
},
"attributed": true,
"fixing_flips": [
"<string>"
],
"flips_total": 1,
"answer_conflicts": [
"<string>"
],
"suspects": [
"<string>"
],
"answers": {},
"evaluation_id": "<string>",
"item_id": "<string>"
}
],
"cases_total": 1,
"attributed": 1
},
"accepted": true,
"reasons": [
"<string>"
],
"new_severe_human": [
"<string>"
],
"environment": "production",
"decision_id": "<string>",
"excluded": {
"unlabeled": 1,
"failed_transient": 1,
"base_only": 1,
"outcome_not_in_result": 1,
"no_base": 1
},
"cases": {
"filter": {
"kind": "all",
"expected": "<string>",
"got": "<string>",
"question": "<string>",
"at": 123
},
"total": 1,
"rows": [
{
"item_id": "<string>",
"expected": "<string>",
"got": "<string>",
"loss": 1,
"severe": true,
"base": {
"got": "<string>",
"loss": 1,
"severe": true
}
}
],
"next_cursor": "<string>"
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}{
"error": {
"type": "invalid_request",
"message": "<string>",
"reason": "<string>",
"param": "<string>",
"details": {
"issues": [
{
"param": "<string>",
"code": "<string>",
"message": "<string>"
}
]
}
}
}Authorizations
An API key, sent as Authorization: Bearer <key>. Each operation lists the scope its key needs: decisions:evaluate (evaluate and read evaluations), decisions:feedback (your app's feedback) or decisions:develop (everything else, including owner feedback and the review queue). An operation listing no scope takes any scoped key. A key with no scope, or without the operation's, gets 403 INSUFFICIENT_SCOPE.
Path Parameters
Pattern:
^[a-z][a-z0-9_.-]{0,127}$Body
application/json
One version's results: runs by id, or results inline with the structure they ran on.
- Option 1
- Option 2
- Option 3
Show child attributes
Show child attributes
The version to compare with: same items, by item_id.
- Option 1
- Option 2
- Option 3
Show child attributes
Show child attributes
Page through each item's loss. Without it the report has no rows.
Show child attributes
Show child attributes
Response
The loss report
What this run scored with.
Show child attributes
Show child attributes
The base's report; null when there is no base.
Show child attributes
Show child attributes
Show child attributes
Show child attributes
The candidate lowers loss per case and adds no new severe mistake on a human-labeled case. False without a base.
Why it is not accepted; empty when it is.
Human-labeled cases the candidate gets severely wrong that the base did not.
The environment of the key that made the call: production or test. Evaluations, feedback, reviews and suggestions are separate per environment; registered structures are shared.
Available options:
production, test Pattern:
^[a-z][a-z0-9_.-]{0,127}$Items left out of the report, by why.
Show child attributes
Show child attributes
With cases in the request: one page of rows.
Show child attributes
Show child attributes