jeff-adapter-equivalent

Same meaning check. Checks whether two items have the same meaning within a defined scope, allowing for differences in wording, units and format. Returns yes, no or unclear.

A LoRA adapter for jeff-base v1.4, Jeff's small open decision model (a fine-tune of Qwen3.5-0.8B). You give it a situation and a question with named options; it returns a calibrated probability for every option from one forward pass, with no generated text to parse.

• Adapter ID: jeff.reasoning.equivalent, version 1.4.0; short names: oss-equivalent, equivalent
• Catalogue page: jeffhub.ai/catalogue/jeff.reasoning.equivalent
• Code, server and SDKs: github.com/firelex/jeff
• Base it needs: mstrasser/jeff-base at revision v1.4. The engine refuses an adapter trained on another base, so this adapter loads only on that revision.

When to use it

• You need to compare records or statements expressed differently, such as names, amounts or dates in different formats.

When not to use it

• You need to determine which item is better or more complete; this adapter only checks equivalence.

What it decides

What a call sends:

Field	Required	What it holds
setting (context)	yes	Who is asking: a sentence naming the team and organisation.
scope (context)	yes	What counts when comparing: a sentence saying which details matter (for example every fact, number, date and name).
item_a (context)	yes	The first item: a text (a rule, sentence, record or value).
item_b (input, fields)	yes	The second item, to compare with item_a within the scope.

jeff.reasoning.equivalent.equivalent (classify, choice)

Instructions: Do these two items mean the same thing, within the stated scope? Answer yes when every detail that matters under the scope is the same, even if the wording, the order of things whose order does not matter, the units or the formats differ. Answer no when at least one detail that matters differs, or is in one item but not the other. Answer unclear only when sameness depends on something neither item settles, such as a date or a time of day that can be read two ways, or a number given without the unit or sign that decides what it means.

Labels:

• no: No: they differ in a detail that matters.
• unclear: Unclear: sameness depends on something neither item settles.
• yes: Yes: they mean the same within the scope.

How it was measured

Base: Jeff v1.4; adapter trained 2026-10-25.

Test	Rows	Accuracy
PAWS, a public paraphrase benchmark (public)	1,000	93.4%
the in-house test (generator)	2,000	98.5%
the in-house test in domains never seen in training (generator)	1,931	97.8%

Accuracy is the share of test rows whose top answer was right. Kinds: public, a public benchmark; real, held-out real documents; generator, an in-house test from the same generator as the training data (it measures less than a public or real test does); mixed, several kinds; unknown, not recorded.

When it is confident (from the calibration set): the threshold is the confidence from which an answer counts as confident, precision the share of confident answers that were right, coverage the share of answers that were confident.

Function	Threshold	Precision	Coverage	Calibration rows
equivalent	0.499	98.4%	100.0%	2,000

How to call it

Install the Python SDK and the engine (see the install guide; until the packages are on PyPI, install them from github.com/firelex/jeff).

The example below is "Two listings of the same paperback". Every fact the scope names agrees between the two book listings, so the answer is yes.

On your machine (Jeff.local downloads this adapter and the base from Hugging Face the first time):

import jeff

client = jeff.Jeff.local(["equivalent"])

decide = client.function("jeff.reasoning.equivalent.equivalent", context={
    "setting": "The data team of an online bookshop compares supplier listings before merging duplicate products.",
    "scope": "Every fact that identifies the edition counts (title, author, format, page count and year); wording and punctuation do not.",
    "item_a": "The Lighthouse Keeper's Daughter by Ana Rivas, paperback, 312 pages, published 2021.",
})

decision = decide("Ana Rivas, The Lighthouse Keeper's Daughter (paperback edition, 2021), 312 pp.")

print(decision.value, decision.confidence)

Through the Jeff API (an API key from jeffhub.ai):

import os

import jeff

with jeff.Jeff.remote("https://api.jeffhub.ai", token=os.environ["JEFF_API_KEY"], timeout_s=30) as client:
    decide = client.function("jeff.reasoning.equivalent.equivalent", context={
        "setting": "The data team of an online bookshop compares supplier listings before merging duplicate products.",
        "scope": "Every fact that identifies the edition counts (title, author, format, page count and year); wording and punctuation do not.",
        "item_a": "The Lighthouse Keeper's Daughter by Ana Rivas, paperback, 312 pages, published 2021.",
    })

    decision = decide("Ana Rivas, The Lighthouse Keeper's Daughter (paperback edition, 2021), 312 pp.")

    print(decision.value, decision.confidence)

curl -s https://api.jeffhub.ai/v1/decide \
  -H "Authorization: Bearer $JEFF_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"calls": [{"function": "jeff.reasoning.equivalent.equivalent", "input": "Ana Rivas, The Lighthouse Keeper'\''s Daughter (paperback edition, 2021), 312 pp.", "context": {"setting": "The data team of an online bookshop compares supplier listings before merging duplicate products.", "scope": "Every fact that identifies the edition counts (title, author, format, page count and year); wording and punctuation do not.", "item_a": "The Lighthouse Keeper'\''s Daughter by Ana Rivas, paperback, 312 pages, published 2021."}}]}'

Recorded answer (2026-11-03): yes with confidence 0.971.

To download the weights yourself:

hf download mstrasser/jeff-base --revision v1.4 --local-dir jeff-base
hf download mstrasser/jeff-adapter-equivalent --revision v1.4 --local-dir adapters/equivalent

Files

File	What it is
adapter_model.safetensors, adapter_config.json	The LoRA weights and their configuration
readout.safetensors	The adapter's readout over the answer codes
decision_config.json	Answer codes, temperature, prompt layout and the base it was trained on (the SHA-256 of the base's weights, which the engine checks)
manifest.yaml	The adapter's manifest: its functions, labels, fields and calibration, as the SDKs read it
test.jsonl	The held-out test set the numbers above come from
calibration.jsonl	The calibration set the thresholds come from
LICENSE	The Apache License 2.0, for the Qwen3.5-0.8B notice below

Licence

Adapter licence: Apache-2.0

Qwen3.5-0.8B notice: these weights were modified from Qwen3.5-0.8B by the Jeff project: jeff-base is a fine-tune of Qwen3.5-0.8B, and this adapter was trained on top of it. Qwen3.5-0.8B is Copyright 2026 Alibaba Cloud and licensed under the Apache License, Version 2.0; a copy of that licence is in LICENSE.

Jeff is an independent project, not affiliated with or endorsed by the makers of Qwen.