Inference-PPL

OpenCompass metric that averages token NLL only on labeled positions, shipped with a local cn-reasoning-val example rather than a fixed public test.

Also known as: inference-ppl, OpenCompass InferencePPL

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.
Categorygeneration
Subcategorytoken-weighted negative log-likelihood on labeled spans of a reasoning corpus
Page statusunknown
Metricaverage_ppl (token-weighted mean of labeled-span NLL)
Directionlower_is_better
PublisherOpenCompass

What it measures

Inference-PPL is a language-modeling score, not a multiple-choice exam. OpenCompass feeds each example's text through InferencePPLOnlyInferencer and asks the model for a tokenwise loss on labeled positions, then averages those losses. The config README says the method is meant for reasoning corpora, where only the "answer" or other marked span should contribute. The bundled example file is named cn-reasoning-val.jsonl under ./data/inference_ppl. The same inferencer can point at other jsonl files with a text field.

Task format

Zero-shot completion over raw {text}. The dataset loader requires a local jsonl named {name}.jsonl (default name cn-reasoning-val), drops empty texts, and sets output_column to None. The inferencer then calls get_ppl_tokenwise_from_template(entry, label) and stores per-example ppl and token_len. AverageInferencePPLEvaluator returns sum(ppl) / sum(token_len).

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub