ReDefiNER

Pull out the entities you define, using a plain-English rulebook, on your own CPU.

A 0.8B model that extracts and labels exact spans using your plain-English rulebook.

_{On CrossNER AI it scores 68.1 F1 on a CPU server, about 5 F1 behind Claude Haiku 5.5 (73.1) and 4.3 behind Gemini 3.1 Pro (72.4), given the same rulebook. How it compares, including where it loses}

Quickstart · Label mode · See a rule at work · Benchmarks · Code · Demo

_{45 seconds: one support ticket, three rulebook edits, a long rulebook, then the benchmarks. Every span in it is real output from this model. The 98-second version adds label mode.}

Same model, same ticket, two different jobs: the rulebooks and output as text

Privacy team

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] PERSON
[DESCRIPTION] A person's name.

[TYPE] ACCOUNT_ID
[DESCRIPTION] An account or customer number.

[TYPE] EMAIL
[DESCRIPTION] An email address.

[TYPE] PHONE
[DESCRIPTION] A phone number.

Output: Priya Raman PERSON · 88-41207 ACCOUNT_ID · priya.raman@outlook.com EMAIL · 617-555-0148 PHONE

Product team

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] PRODUCT
[DESCRIPTION] A named software product or app.

[TYPE] VERSION
[DESCRIPTION] A version number.

[TYPE] DEVICE
[DESCRIPTION] A named hardware model.

[TYPE] ERROR_CODE
[DESCRIPTION] An error code.

Output: Zoom PRODUCT · 6.2.5 VERSION · MacBook Air DEVICE · 1132 ERROR_CODE

Each rulebook lists the entity types you want, with a one-line description and any rules about what counts. The model finds and labels those spans in your text (extract mode), or types spans you already have (label mode). To change what it finds, you edit the rulebook; nothing is retrained. It runs on an ordinary CPU, with no GPU and no per-use fees, and once downloaded it works offline, so your text stays on your own servers: a fit for teams that can't send data outside, such as banks, hospitals and government. Free for commercial use under Apache 2.0.

Need it on your own data? We write and test rulebooks with you and run ReDefiNER on your own servers: jungroo.com/redefiner or hello@jungroo.com.

At a glance

Quickstart

The model has its own span and label heads, so it loads through the redefiner package rather than transformers alone.

pip install redefiner

from redefiner import Tagger

tagger = Tagger.from_pretrained("jungroo-ai/redefiner-0.8b")   # GPU if available, else CPU

rulebook = """[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] PERSON
[DESCRIPTION] A person's name.
PE1 Label the full name; leave out titles such as "Dr" or "Prof".

[TYPE] ORGANIZATION
[DESCRIPTION] A company, university or other institution.
"""

tagger.extract("Prof Ada Lovelace corresponded with the Royal Society.", rulebook)
# [{'start': 5, 'end': 17, 'text': 'Ada Lovelace', 'label': 'PERSON', 'confidence': 0.977},
#  {'start': 40, 'end': 53, 'text': 'Royal Society', 'label': 'ORGANIZATION', 'confidence': 0.489}]

Label mode, when you already have the spans (character offsets, end exclusive):

tagger.label("Ada Lovelace corresponded with the Royal Society.", [(0, 12), (35, 48)], rulebook)
# [{'start': 0, 'end': 12, 'text': 'Ada Lovelace', 'label': 'PERSON', 'confidence': 1.0,
#    'alternatives': [('ORGANIZATION', 0.0)]},
#  {'start': 35, 'end': 48, 'text': 'Royal Society', 'label': 'ORGANIZATION', 'confidence': 0.997,
#    'alternatives': [('PERSON', 0.003)]}]

Both calls also take a list of texts (and, for label, one list of spans per text) and return one result list per text. More in Label mode.

Threshold and confidence. threshold (default 0.5) decides whether a piece of text counts as an entity at all. A span's confidence is that probability times the probability of its type, so spans below 0.5 confidence still come back, like "Royal Society" (0.489) above. The defaults are the setting every benchmark here uses, not one tuned to your data: label 20 to 50 of your own texts, run them, and choose the cutoff that scores best on them: [s for s in spans if s["confidence"] >= cutoff]. A higher cutoff or threshold gives fewer, surer spans; a higher none_bias (default 2) keeps more borderline ones.

Installing: Python version, NVIDIA GPUs, Linux without a GPU

Python 3.10 or newer. Runs on CPU, or on an NVIDIA GPU if your PyTorch build has CUDA (PyTorch's install selector gives the command for your system). On Linux with a GPU, install "redefiner[gpu]" instead: it adds flash-linear-attention's kernels, without which the GPU path is much slower. On Linux without a GPU, add --extra-index-url https://download.pytorch.org/whl/cpu to skip PyTorch's CUDA build, a download of several GB.

Tagger.from_pretrained(..., device="cpu") forces CPU. print(tagger.device) shows where the model runs (cpu or cuda:0); a warning says so when an NVIDIA GPU is present but PyTorch can't use it.

What rule PE1 does: without it, the same call keeps the title

without_pe1 = rulebook.replace('PE1 Label the full name; leave out titles such as "Dr" or "Prof".\n', "")
tagger.extract("Prof Ada Lovelace corresponded with the Royal Society.", without_pe1)
# [{'start': 0, 'end': 17, 'text': 'Prof Ada Lovelace', 'label': 'PERSON', 'confidence': 0.606},
#  {'start': 40, 'end': 53, 'text': 'Royal Society', 'label': 'ORGANIZATION', 'confidence': 0.424}]

How confidence works, and why "Royal Society" scores 0.489

In extract mode, each span's confidence is its span probability times its type probability. Two settings decide which spans come back: threshold (default 0.5) is the minimum span probability, and none_bias (default 2) is subtracted from the score of "none of these types" before the type is picked, so borderline spans get a type instead of being dropped. The confidence is computed without that bias, so a returned span can score below 0.5. Above, "Royal Society" is almost certainly a span (0.997), but the model weighs ORGANIZATION (0.491) and none of the types (0.508) about equally: 0.997 × 0.491 = 0.489. Filter on confidence if you only want the spans the model is sure of. In label mode, confidence is the probability of the chosen type among the rulebook's types.

Writing your own rulebook: start from a template (a blank layout; PII, news, product reviews and scientific text; and a skeleton for types that depend on context), and see How to write rules that work.

See a rule at work

Every card on this page is real output: the released int8 model on a laptop CPU, with nothing retrained between runs. Only the rulebook changes. Green lines were added to the rulebook; spans outlined in red disappear, spans outlined in green are new or changed. Try your own text and rules in the demo.

Keep "under" with the price: the rulebook and output as text

Before

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] BRAND
[DESCRIPTION] A brand name.

[TYPE] PRODUCT
[DESCRIPTION] A product line or model name.

[TYPE] COLOR
[DESCRIPTION] A colour.

[TYPE] SIZE
[DESCRIPTION] A size.

[TYPE] PRICE
[DESCRIPTION] A price or a price limit.

Output: red COLOR · nike BRAND · air max PRODUCT · 10 SIZE · $120 PRICE

After

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] BRAND
[DESCRIPTION] A brand name.

[TYPE] PRODUCT
[DESCRIPTION] A product line or model name.

[TYPE] COLOR
[DESCRIPTION] A colour.

[TYPE] SIZE
[DESCRIPTION] A size.

[TYPE] PRICE
[DESCRIPTION] A price or a price limit.
PR1 A price limit keeps its limit word in the span: "under $50", "below 30 dollars", "less than $200".

Output: red COLOR · nike BRAND · air max PRODUCT · 10 SIZE · under $120 PRICE

Ignore your own company and products: the rulebook and output as text

Before

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] COMPANY
[DESCRIPTION] A company name.

[TYPE] PRODUCT
[DESCRIPTION] A named product or device.

Output: Acme COMPANY · Sonos Arc PRODUCT · Acme Home PRODUCT · Sonos COMPANY · Acme COMPANY

After

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] COMPANY
[DESCRIPTION] A company name.

[TYPE] PRODUCT
[DESCRIPTION] A named product or device.

[TYPE] NOT_AN_ENTITY
NE1 Our own company and products: "Acme", "Acme Home". Never label them.

Output: Sonos Arc PRODUCT · Sonos COMPANY

The next card uses an example rulebook for a travel desk, with 11 types, written the way the model reads best: general rules, precedence rules that settle conflicts, numbered type rules with "query" -> span examples, and what never to extract. "Before" is the same rulebook with only its types and descriptions.

On 21 test queries the full rulebook got 19 exactly right, against 11 with types and descriptions only and 9 with type names only. Both misses came from one rule: a single code with the wrong name ("my ticket number is ZK48TQ") keeps the type of its name; the format rule works when both kinds of code appear together. The layout is explained in Structure for hard rules, and docs/templates/hard-rules.txt is a blank copy to fill in.

The travel-desk rulebook (docs/examples/travel-desk-rulebook.txt)

[RULES] Travel desk rulebook. Label each entity in the traveller's message with one of the types below.
G1 Label all entities in the message, not only the first one.
G2 Keep each span as short as the entity itself ("a flight to Oslo" -> "Oslo").
G3 When two types fit, the precedence rules below decide.
[PRECEDENCE]
P1 An airport name or a three-letter airport code is AIRPORT, never ORIGIN or DESTINATION, even after "from" or "to".
P2 Direction decides between ORIGIN and DESTINATION: a place after "from", "out of" or "leaving" is ORIGIN; after "to", "into", "arriving in" or "back to" is DESTINATION.
P3 The person who will fly is TRAVELLER; the person who arranges or pays is BOOKER.
P4 A date after "leave", "depart" or "out on" is DEPARTURE_DATE; after "return", "back on" or "fly home" is RETURN_DATE.
P5 An airline name followed by a programme word ("Miles", "Club", "Rewards") is LOYALTY_PROGRAM; any other airline name is AIRLINE.
P6 The reference format decides, whatever the traveller calls it: a code that mixes letters and digits is BOOKING_REF, even when called a ticket number; a long code of digits only is TICKET_NUMBER, even when called a booking reference.
[TYPE] TRAVELLER
[DESCRIPTION] A person who will take the trip.
TR1 "Book Wanda on the morning flight" -> "Wanda"; "seats for Ilse and Gus" -> "Ilse", "Gus".
[TYPE] BOOKER
[DESCRIPTION] A person who arranges or pays for someone else's trip.
BK1 "Petra is booking for her son Leon" -> "Petra" is BOOKER, "Leon" is TRAVELLER.
BK2 "charge it to Viktor's card" -> "Viktor".
[TYPE] ORIGIN
[DESCRIPTION] The city or country the trip starts from.
OR1 "from Porto" -> "Porto"; "leaving Seoul" -> "Seoul".
[TYPE] DESTINATION
[DESCRIPTION] The city or country the trip goes to.
DS1 "to Nairobi" -> "Nairobi"; "back to Quito" -> "Quito".
[TYPE] AIRPORT
[DESCRIPTION] A named airport or a three-letter airport code.
AI1 "from YYZ" -> "YYZ"; "into Changi" -> "Changi".
[TYPE] DEPARTURE_DATE
[DESCRIPTION] The date the outbound trip starts.
DD1 "leave on 4 June" -> "4 June"; "I went out on 2 May" -> "2 May".
[TYPE] RETURN_DATE
[DESCRIPTION] The date of the return trip.
RD1 "back on 11 June" -> "11 June".
[TYPE] AIRLINE
[DESCRIPTION] An airline the traveller flies with.
AL1 "on Skyvane" -> "Skyvane".
[TYPE] LOYALTY_PROGRAM
[DESCRIPTION] An airline's points or membership programme.
LY1 The span keeps the programme word ("my Skyvane Club number" -> "Skyvane Club").
[TYPE] BOOKING_REF
[DESCRIPTION] A booking reference: a short code of letters and digits.
BR1 "reference QX7PLM" -> "QX7PLM"; "my ticket number is HB2RWA" -> "HB2RWA" is BOOKING_REF.
[TYPE] TICKET_NUMBER
[DESCRIPTION] An e-ticket number: a long code of digits only.
TN1 "ticket 2201234567890" -> "2201234567890"; "booking ref 1259876543210" -> "1259876543210" is TICKET_NUMBER.
[NOT_AN_ENTITY]
NE1 Places the traveller wants to avoid ("not via Doha", "no stop in Frankfurt").
NE2 Request phrases ("can you", "please") and generic words ("flight", "seat", "trip").

Six more examples: a rule that overrules the words in the text, pronouns, a split type, a new type, and two more travel-desk cards

Hard rules. Rules that overrule the words in the text, or ask for something the model leaves out by default.

The reference format decides, not what the customer calls it: the rulebook and output as text

Before

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] ORDER_ID
[DESCRIPTION] An order number.

[TYPE] INVOICE_ID
[DESCRIPTION] An invoice number.

Output: INV-7731 ORDER_ID · #D-55102 INVOICE_ID

After

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] ORDER_ID
[DESCRIPTION] An order number.

[TYPE] INVOICE_ID
[DESCRIPTION] An invoice number.
[RULES]
The format decides, whatever the customer calls it: a reference starting with # is ORDER_ID ("#B-11872"), even when the customer calls it an invoice; a reference starting with INV- is INVOICE_ID ("INV-2025-0099"), even when the customer calls it an order. This overrides everything above.

Output: INV-7731 INVOICE_ID · #D-55102 ORDER_ID

Count pronouns as people: the rulebook and output as text

Before

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] PERSON
[DESCRIPTION] A person's name.

Output: Maria PERSON

After

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] PERSON
[DESCRIPTION] A person's name.
PE1 Also label the pronouns that refer to a person: "she", "he", "they", "her", "him", "them".

Output: Maria PERSON · She PERSON · her PERSON · her PERSON

More everyday edits

Separate programming languages from other skills: the rulebook and output as text

Before

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] SKILL
[DESCRIPTION] A technical skill: a programming language, framework, platform or tool.

Output: Python SKILL · Go SKILL · AWS SKILL · Kubernetes SKILL · PostgreSQL SKILL · React SKILL

After

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] SKILL
[DESCRIPTION] A technical skill: a programming language, framework, platform or tool.

[TYPE] LANGUAGE
[DESCRIPTION] A programming language, such as "Java" or "Rust".
LA1 Programming languages are LANGUAGE, never SKILL.

Output: Python LANGUAGE · Go LANGUAGE · AWS SKILL · Kubernetes SKILL · PostgreSQL SKILL · React SKILL

Add a type in two lines: the rulebook and output as text

Before

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] COMPANY
[DESCRIPTION] A company name.

[TYPE] PERSON
[DESCRIPTION] A person's name.

[TYPE] PRODUCT
[DESCRIPTION] A product name.

[TYPE] DATE
[DESCRIPTION] A date or day.

Output: Apple COMPANY · Tim Cook PERSON · iPhone PRODUCT · Tuesday DATE

After

[RULES]
Extract every entity of the supplied types. Return exact spans from the text.

[TYPE] COMPANY
[DESCRIPTION] A company name.

[TYPE] PERSON
[DESCRIPTION] A person's name.

[TYPE] PRODUCT
[DESCRIPTION] A product name.

[TYPE] DATE
[DESCRIPTION] A date or day.

[TYPE] TICKER
[DESCRIPTION] A stock ticker symbol, such as "TSLA" or "MSFT".

Output: Apple COMPANY · AAPL TICKER · Tim Cook PERSON · iPhone PRODUCT · Tuesday DATE

More from the travel-desk rulebook (the same rulebook as above)

The quoted examples inside the rules ("under $50", "Java", "TSLA", "#B-11872") are not in the texts: the model applies the rule, it doesn't copy it. Hard rules work when they are concrete, as here; a rule that asks the model to judge what didn't happen ("only cities the writer actually visited, not ones they plan to") is not followed yet. See Limitations.

Benchmarks

Against frontier models

Five frontier models got the same rulebooks as ReDefiNER, on two of our test sets. ReDefiNER is 4 to 11 F1 behind on extraction and 10 to 15 points behind on labeling. It costs far less to run: with the rulebook cached, about 85 times less than Claude Haiku 5.5 on an RTX A5000 GPU, and about 9 times less on a 2-vCPU server with no GPU.

• Tasks. Extraction: the 431 CrossNER AI test sentences, micro-F1 on exact spans and types. One request is GoLLIE's CrossNER AI guidelines as a rulebook, instructions and one sentence, about 2,000 tokens. Labeling: the 358 S1MB Few-NERD spans, accuracy.
• 68.1 here, 66.7 below. Every model here got the rulebook with its types in seed-0 order (the one in benchmarks/reproduce/), on which ReDefiNER scores 68.1. The GoLLIE table below reports 66.7, the mean over three type orders. Likewise, 71.8 here is Few-NERD alone; the S1MB row below, 72.0, is the mean of Few-NERD and Gretel PII.
• Seconds per request. Frontier models: API wall time per extraction request. ReDefiNER on GPU: one text per request with a 2,000-token rulebook, measured on an NVIDIA A10 and on an RTX A5000 with the same result. On CPU: one extraction request at a time on a 2-vCPU cloud server, with the CrossNER AI rulebook. "Cached" means the rulebook was already loaded; "new" includes reading it first.
• Cost per 1,000 requests. Frontier models: 2,000 input tokens plus the model's measured output tokens, at its per-token prices; in the cached column the input is billed at the cache-read price (10% of the input price where none is listed). ReDefiNER: seconds per request times the hourly rental price, with the machine kept busy: $0.16 an hour for the RTX A5000, $0.0586 for the 2-vCPU server.

Method and limits: the frontier comparison.

Against research models

Entity-level F1 (exact boundary and type) unless marked; S1MB is accuracy. Full tables, protocols and caveats are on the benchmark page, with every rulebook it reports in benchmarks/rulebooks/.

All numbers use one fixed setting (int8, span threshold 0.5, NONE bias 2, 256-token query cap). Baseline numbers are published results unless marked as our run.

The 13 GoLLIE zero-shot NER sets, with GoLLIE's own guidelines

GoLLIE's guideline text and five examples per type, converted into a rulebook with nothing added; scored with GoLLIE's code. ReDefiNER scores are the mean over 3 orders of the types in the rulebook (for CrossNER AI, the seed-0 order alone gives 68.1, the score in the frontier comparison above); GoLLIE scores are from its paper.

ReDefiNER leads GoLLIE 34B on 9 of the 13 sets with about 1/40 of its parameters. The largest loss is WikiEvents, which labels pronouns and common nouns under the ACE05 scheme GoLLIE was trained on.

Open NER: the same 7 CrossNER and MIT test sets

Mean F1 over 6,470 sentences, with each type given as a one-line description or by its name only. For our runs of other models, the first number uses the model's default threshold and the second a threshold chosen on the dev sets; ReDefiNER uses its one fixed release setting.

ReDefiNER leads when types are described, also against the tuned baselines. With names only it trails GLiNER large, NuNER Zero-span and GNER. Descriptions hurt both NuNER models, which were trained on short type names; their labels are lower-cased as NuNER's card requires. NuNER Zero's own reported 68.0 on these sets is token-level F1, which gives credit for partly right spans; scored that way, our run gives 67.9. The Policy-Linter takes free-text rules at inference, like ReDefiNER, but scores each word on its own, so multi-word names often break apart.

S1MB span labeling: Few-NERD and Gretel PII

Accuracy on the benchmark's own cases. Rows not marked "our run" are from the S1MB leaderboard. The d1 decision models got S1MB's records as they are, through their system_one interface. Settings, prompts and every other caveat are on the benchmark page.

Limitations

• Rules against a strong prior are mostly ignored. Relabelling a person's name as something other than a person, asking for common nouns as entities, or "only revenue amounts, not every money amount" mostly has no effect. Pronouns do work when they are quoted in a short list under the type.
• Types that depend on context need structure. Precedence rules and numbered rules with worked examples help (see Structure for hard rules); rules that need counting or judging what didn't happen are still mostly ignored.
• Names-only rulebooks are weaker than descriptive ones; always write a one-line description per type.
• Abstaining is weak. When a type's condition isn't clearly met, the model tends to label anyway.
• Spans up to 16 tokens. Long texts are split into parts that are each read on their own, so the same name can get different labels in different parts of a document.
• English only; tested on English benchmarks only.

Training data

ReDefiNER was trained on synthetic rule-following data across many domains: each example pairs a text with a rulebook and the spans that rulebook implies, and many examples come in pairs that differ in only one rule. Labels come from existing annotations, some of them written by language models; a larger teacher model wrote parts of the rulebook text and audited the labels before training. The data is not released. No benchmark test set was used in training. Benchmark dev splits were used only to write and select the rulebooks of the rows marked "tuned on dev"; CrossNER, MIT and S1MB test scores were tracked during development and informed the choice of the released checkpoint (see Disclosures on the benchmark page).

Intended use

ReDefiNER is meant for extracting entities from English text when you need control over what counts: your own entity types, document fields, search-query slots, and research on rule-following extraction.

• Check outputs before acting on them in legal, medical, financial or other high-stakes work. The model can miss or mislabel spans, and its confidences are model scores, not guarantees.
• Don't rely on it as the only safeguard when finding personal data for redaction or compliance; it will miss some.
• Don't use it to profile or track people without a legal basis and their consent.

Architecture

Files

• model.safetensors: the merged int8 weights (decoder linear layers and token embedding in int8 with fp32 scales; span and label heads).
• config.json: model and decoding settings.
• Tokenizer files from Qwen3.5-0.8B.
• LICENSE and NOTICE: Apache-2.0, and the modifications made to Qwen3.5-0.8B.
• benchmarks/: the full benchmark page, every rulebook it reports, and two reproduce spot checks.
• docs/: the rule-writing guide, rulebook templates, the example travel-desk rulebook, and the images in "See a rule at work".

Licence

Apache-2.0. The weights are a modified version of Qwen3.5-0.8B (Apache-2.0); NOTICE lists the modifications. The GoLLIE guideline text in benchmarks/ comes from GoLLIE (Apache-2.0). The benchmark datasets keep their own licences: test sentences and quoted examples from them (in benchmarks/ and the demo) are covered by those licences, not by Apache-2.0; see Datasets and licences.

Citation

@misc{redefiner2026,
  title  = {ReDefiNER: Rule-Defined Named Entity Recognition},
  author = {{Cibe Hariharan K} and {Sethuraman T A} and {Jungroo AI Labs}},
  year   = {2026},
  url    = {https://huggingface.co/jungroo-ai/redefiner-0.8b},
  note   = {Release 1.0}
}

Credits

Created by Jungroo AI Labs. For help putting it to work on your own data, see jungroo.com/redefiner or write to hello@jungroo.com.