eval-harness

arm

cnn_lead3_dm200_v1

One config on cnn_dailymail_dm200, scored one way; its repeats averaged per item.

dataset
cnn_dailymail_dm200
scored against
gold/cnn_dailymail_dm200 gold
goal
Short news summaries (CNN)
runs
1
headline
coverage 0.370
last run
2026-10-14 13:41

Inspect itWith the 5 arms scored the same way

01metrics

metricmeanitemsbetter
coverage0.370200higher
compression0.148200neutral
concision0.334200higher
cost_usd$0.000000200lower
fmt_bullets0.000200neutral
fmt_label0.000200neutral
fmt_markdown0.000200neutral
fmt_narration0.020200neutral
fmt_overlong0.200200neutral
fmt_paragraphs0.000200neutral
grounding1.000200higher
latency_ms0.123200lower
length_vs_reference1.464200neutral
reference_words57.9200neutral
rouge10.422200higher
rougeLsum0.382200higher
summary_words77.3200neutral

02cost and speed

cost per item
$0.000000
speed per item
p50 0.1 · p95 0.27 ms

Per item, over its items: cost as dollars and tokens (a local model is free but still uses tokens), speed as the median and the slow tail (p95). A goal that names this arm weighs them in its decision.

03runs

runmadeitemscosthow
cnn_lead3_dm200_v1_20261014T134148Z2026-10-14 13:41200$0.000000measured

04config

config cnn_lead3_dm200_v1 · adapter examples/summarization-cnn-dailymail/adapter.py · credentials are never shown

family
lead
price_tier
free
provider
lead_k
sentences
3

05conditions

data.items_sha256
0cf85d1eb3b9176ac6e49f3ab83761e271aeb0f87b18709f230815a4e213865b
deployment.model
lead_3
deployment.source
params
fingerprint.version
5
library.numpy
2.4.6
library.openai
3.19.2
library.rouge-score
0.1.2
library.torch
2.14.0
library.transformers
4.55.4
machine.cpu_count
4
machine.platform
Linux/x86_64
machine.processor
x86_64
machine.python
3.11.15
machine.release
6.18.44-fc-v114
reference.sha256
7fc293bc2d9661e3b23d2bb1af8981f5e4bab10e4ab3c83b36e03f571fcef8cc
scorer.id
examples/summarization-cnn-dailymail/adapter.py
scorer.sha256
2955537740425c943bf6eb342ebf0c2dac2379d13b4e32ec5005d27227be6f15