SAMPLE-001-US Sample report: illustrative figures Records through October 2026 21 October 2026

Training a transformer language model with attention over token embeddings for generative text

A transformer fine-tuning application routes to art unit 2122 at 78 percent, and Examiner 1 is the most likely to receive it at 15 percent, with nine other candidates covering most of the remaining mass. Sample report: the record behind the figures is illustrative.

This report predicts routing from the prosecution record and from the claims as drafted. The office's own classification and docketing decide the assignment; read every figure as a probability, not a promise. It is not legal advice.

Predicted art unit 2122
77.7% probability · TC 2100 Workgroup 2120; 3,299 filings in the window; allowance 68%; 10 active examiners.
Routing confidence 78/100
decisive Top-1 probability 77.7%; the five candidates hold 99% of the mass; entropy 0.85.
Most likely examiner Examiner 1
14.6% overall · 19% within 2122 Took 15.2% of the unit's new cases in the activity window (2 of 10 active examiners, 247 new cases).
Top-10 coverage 78%
ten of 58 candidates The chance that the assigned examiner is one of the ten listed below.
Probability of routing
2122 77.7% 2121 8.7% 3689 4.9% 2129 4.0% 3623 3.9% 0% 100%
Routing probability for each candidate unit, from the classification picture, the nearest recent filings and any claimed family. Records through October 2026.

The inventive symbol, G06N3/0455 for transformer architectures, has sent 97 percent of its recent first-listed filings to 2122, and the two additional symbols for training and machine learning in general spread across 2121, 2122 and the business method units without disturbing that lead. The forty nearest recent filings by title and abstract all sit in 2122. No parent application was claimed, so no family weight applies, and the unit is decided on the classification picture alone.

Inside 2122 the intake is flat: ten examiners took new cases in the activity window, the busiest at 17 percent and the lightest at 6 percent, and all ten carry every symbol in the picture. The ranking therefore follows intake share and recent docket fit rather than any single dominant examiner, and the one adjustment made, a 1.25 factor for Examiner 1 whose newest cases are transformer fine tuning, moves that examiner from second to first. The top-10 coverage of 78 percent equals the unit probability, because the ten names are the unit's whole active roster; the remaining 22 percent is the chance the application routes elsewhere.

Classification

The classification picture

Claim 1 recites computing attention over token embeddings with a transformer and generating text with a pretrained language model, so the inventive concept is the transformer architecture itself, classified at G06N3/0455. Training is claimed in the same independent claim and takes G06N3/08 as a second concept, and the machine learning context recited in claim 5 takes G06N20/00 at the lowest weight. The nearest-filing check carried all three symbols and listed G06N3/0455 and G06N20/00 as the most common first-listed symbols among the neighbours, so the draft stands. A claim 1 that recited the generated text's commercial use instead of the architecture would replace the inventive symbol with a G06Q symbol and move the application to 3689.

SymbolRoleWeightClaim basisWhere it routesSupport
G06N3/0455inventive1.00claim 1: transformer with attention2122 97%, 2121 1%, 2171 1%exact level; 309 first-listed, 609 carrying
G06N3/08additional0.50claim 1: training2122 30%, 2121 24%, 3689 24%exact level; 861 first-listed, 1952 carrying
G06N20/00additional0.30claim 5: machine learning generally2122 37%, 2121 31%, 3623 30%exact level; 755 first-listed, 1539 carrying
Routing

Where it routes

The unit is decisive at 78 percent. The second candidate, 2121, holds 9 percent on the strength of the training symbol and allows at 59 percent against 68 percent in 2122, so a routing miss toward 2121 costs about nine points of allowance. The business method unit 3689 appears at 5 percent only because G06N3/08 is carried by some of its filings; the picture gives it no first-listed support. Every candidate unit has an active roster.

Art unitProbabilityWorkgroupFilingsAllowance rateActive examinersEvidence
2122
77.7%
21203,29968%10picture 68%, nearest filings 100%, family none
2121
8.7%
21202,73959%14picture 12%, nearest filings 0%, family none
3689
4.9%
36802,17161%8picture 7%, nearest filings 0%, family none
2129
4.0%
21201,60860%13picture 6%, nearest filings 0%, family none
3623
3.9%
36202,13163%13picture 6%, nearest filings 0%, family none

Nearest recent filings

FiledArt unitExaminerTitleSimilarity
2024-01-232122Examiner 4transformer attention token for improved processing0.67
2024-02-132122Examiner 1transformer attention token for improved processing0.67
2024-03-072122Examiner 7token transformer attention for improved processing0.67
2024-05-292122Examiner 10transformer attention token for improved processing0.67
2025-01-202122Examiner 6attention token transformer for improved processing0.67
Examiners

The ten examiners, ranked

The long bar is the probability that the examiner receives the application; the short bar is the probability within the examiner's own unit, which here differs from the long bar by the 78 percent unit probability alone since all ten sit in 2122. One adjustment was made after reading the rosters: Examiner 1 carries a 1.25 factor because seven of the ten newest cases on that docket are transformer fine tuning, the closest dockets in the unit. No other docket read as closer or farther than the symbol statistics already said.

Probability of assignment Probability within the unit
1. Examiner 1 14.6% 2. Examiner 2 12.9% 3. Examiner 3 10.7% 4. Examiner 4 8.8% 5. Examiner 5 6.3% 6. Examiner 6 5.4% 7. Examiner 7 5.3% 8. Examiner 8 4.8% 9. Examiner 9 4.6% 10. Examiner 10 4.4% 0% 100%
The long bar is the probability that the examiner receives the application; the short bar is the probability within the examiner's own unit. Ten of 58 active examiners across the candidate units are shown.
RankExaminerArt unitProbabilityWhy
1Examiner 1212214.6%
19% within the unit
  • Took 15.2% of the unit's new cases in the activity window (2 of 10 active examiners, 247 new cases)
  • Holds 14.8% of the unit's recent cases carrying this picture's symbols or their main groups; portfolio carries G06N3/0455 (209), G06N3/08 (224), G06N20/00 (110)
  • Portfolio similarity 0.63, symbol qualification 100%
  • Newest case on the docket was filed this month
  • Fit adjustment x1.25: seven of the ten newest cases on the docket are transformer fine tuning, the closest dockets in the unit
2Examiner 2212212.9%
17% within the unit
  • Took 16.7% of the unit's new cases in the activity window (1 of 10 active examiners, 273 new cases)
  • Holds 15.6% of the unit's recent cases carrying this picture's symbols or their main groups; portfolio carries G06N3/0455 (197), G06N3/08 (210), G06N20/00 (120)
  • Portfolio similarity 0.64, symbol qualification 100%
  • Newest case on the docket was filed this month
3Examiner 3212210.7%
14% within the unit
  • Took 13.6% of the unit's new cases in the activity window (3 of 10 active examiners, 224 new cases)
  • Holds 14.2% of the unit's recent cases carrying this picture's symbols or their main groups; portfolio carries G06N3/0455 (307), G06N3/08 (134), G06N20/00 (176)
  • Portfolio similarity 0.64, symbol qualification 100%
  • Newest case on the docket was filed this month
4Examiner 421228.8%
11% within the unit
  • Took 11.6% of the unit's new cases in the activity window (4 of 10 active examiners, 185 new cases)
  • Holds 12.0% of the unit's recent cases carrying this picture's symbols or their main groups; portfolio carries G06N3/0455 (134), G06N3/08 (163), G06N20/00 (192)
  • Portfolio similarity 0.64, symbol qualification 100%
  • Newest case on the docket was filed this month
5Examiner 521226.3%
8% within the unit
  • Took 8.3% of the unit's new cases in the activity window (5 of 10 active examiners, 139 new cases)
  • Holds 9.0% of the unit's recent cases carrying this picture's symbols or their main groups; portfolio carries G06N3/0455 (114), G06N3/08 (123), G06N20/00 (152)
  • Portfolio similarity 0.65, symbol qualification 100%
  • Newest case on the docket was filed this month
6Examiner 621225.4%
7% within the unit
  • Took 8.0% of the unit's new cases in the activity window (6 of 10 active examiners, 134 new cases)
  • Holds 7.3% of the unit's recent cases carrying this picture's symbols or their main groups; portfolio carries G06N3/0455 (76), G06N3/08 (94), G06N20/00 (135)
  • Portfolio similarity 0.63, symbol qualification 100%
  • Newest case on the docket was filed this month
7Examiner 721225.3%
7% within the unit
  • Took 7.3% of the unit's new cases in the activity window (7 of 10 active examiners, 119 new cases)
  • Holds 7.5% of the unit's recent cases carrying this picture's symbols or their main groups; portfolio carries G06N3/0455 (57), G06N3/08 (93), G06N20/00 (159)
  • Portfolio similarity 0.64, symbol qualification 100%
  • Newest case on the docket was filed this month
8Examiner 821224.8%
6% within the unit
  • Took 7.1% of the unit's new cases in the activity window (8 of 10 active examiners, 117 new cases)
  • Holds 6.4% of the unit's recent cases carrying this picture's symbols or their main groups; portfolio carries G06N3/0455 (56), G06N3/08 (135), G06N20/00 (78)
  • Portfolio similarity 0.64, symbol qualification 100%
  • Newest case on the docket was filed this month
9Examiner 921224.6%
6% within the unit
  • Took 6.3% of the unit's new cases in the activity window (9 of 10 active examiners, 98 new cases)
  • Holds 6.9% of the unit's recent cases carrying this picture's symbols or their main groups; portfolio carries G06N3/0455 (70), G06N3/08 (94), G06N20/00 (114)
  • Portfolio similarity 0.64, symbol qualification 100%
  • Newest case on the docket was filed this month
10Examiner 1021224.4%
6% within the unit
  • Took 6.1% of the unit's new cases in the activity window (10 of 10 active examiners, 101 new cases)
  • Holds 6.4% of the unit's recent cases carrying this picture's symbols or their main groups; portfolio carries G06N3/0455 (91), G06N3/08 (62), G06N20/00 (114)
  • Portfolio similarity 0.65, symbol qualification 100%
  • Newest case on the docket was filed this month
Profiles

What the likely examiners do

As a group the ten allow between 51 and 80 percent of disposed cases, issue about two actions per case, and reach a first action in a little under fourteen months. The 101 share is the figure that separates them: it runs from 16 percent to 58 percent of cases with at least one eligibility rejection, and the leading examiner sits at the high end. The 103 share is uniform at 75 to 87 percent. Interview and 112 cells read n/a because the record carries no interview or 112 fields for this unit, and Examiner 2 has too few disposed cases for an allowance rate.

ExaminerAllowance rateActions per case101 share103 shareInterviewsMonths to first actionRead
Examiner 166%2.258%80%n/a14.3Allows 66 percent but raises 101 in 58 percent of cases; write the architectural improvement into claim 1 as a limitation, not an argument.
Examiner 2n/a2.041%80%n/a14.0No allowance rate yet, with 395 of 397 cases pending; a new docket, and 101 in 41 percent of first actions so far.
Examiner 362%2.236%82%n/a13.8Allows 62 percent with 101 in 36 percent; the obviousness rejection at 82 percent is the fight to prepare for.
Examiner 477%2.155%76%n/a14.1The strongest allowance in the top four at 77 percent, but 101 in 55 percent of cases; eligibility is decided on claim text here.
Examiner 551%2.240%79%n/a13.6The lowest allowance of the ten at 51 percent with 101 in 40 percent; expect a second action.
Examiner 677%2.018%87%n/a14.0Allows 77 percent with 101 in only 18 percent; obviousness at 87 percent is the ground that matters.
Examiner 758%2.251%75%n/a13.9Allows 58 percent with 101 in half of cases and the most actions per case of the ten.
Examiner 878%2.227%82%n/a13.5Allows 78 percent with 101 in 27 percent; a favourable draw if the art record is deep.
Examiner 980%2.216%76%n/a13.9The best allowance of the ten at 80 percent with 101 in 16 percent; the fastest to first action.
Examiner 1066%2.225%86%n/a14.2Allows 66 percent with 101 in a quarter of cases; obviousness at 86 percent.
Family

Continuation and family effects

What was claimed

Parent
None claimed
Relation
n/a
Parent's unit
n/a
Parent's examiner
n/a
Parent's status
n/a

What it does to the routing

No parent application was claimed, so the routing rests on the classification picture and the nearest recent filings alone.

Steering

Where else it could go

The lever on this application is whether claim 1 recites the architecture or the use. As drafted it recites the architecture and routes to 2122 at 78 percent with a 68 percent allowance rate. Recast around the commercial use of the generated text, with a G06Q30 inventive symbol, it routes to 3689 at 80 percent, where allowance runs lower and eligibility rejections are the norm. Recast around the network layer structure with G06N3/04 leading, it routes to 2121 at 80 percent, a unit that allows at 59 percent; that framing buys nothing. The draft is already in the best of the three positions, and the steering table shows the cost of drifting out of it.

FramingInventive symbolTop art unitProbabilityAllowance rateWhat the claims would say
As draftedG06N3/0455212277.7%68%Claim 1 recites computing attention over token embeddings with a transformer and generating text with the pretrained language model; the commercial use stays in the dependents.
Claimed on the commercial useG06Q30/0201368979.6%61%Claim 1 would recite selecting a promotion for a customer audience from the generated text, which puts the transaction in the independent claim and the application in a business method unit.
Claimed on the network structureG06N3/04212180.0%59%Claim 1 would recite the layer structure and the gradient update of the network, which leads with G06N3/04 and moves the application to 2121 without improving its position.
Calibration

How much to trust this

The measures come from predicting 300 recent filings held out from the record, each with its own classification picture. The unit stage is accurate on this record and the examiner stage carries the error: the leading examiner is right about one time in four, the top five about three times in five, and the ten listed cover the receiving examiner in 94 percent of cases. What the backtest does not measure is the drafting of the picture from the claims; published predictors working from claim text alone reach about half of cases at top 1 and four in five at top 5 across all units, and the decisive band on this record says the picture is not contested.

MeasureValueSampleNote
Art unit, top 1100%300 recent filings after 2026-04-02How often the leading unit was the unit that received the case.
Art unit, top 3100%300 recent filings after 2026-04-02How often the receiving unit was among the three leading units.
Art unit, top 5100%300 recent filings after 2026-04-02How often the receiving unit was among the five listed.
Examiner, top 124%300 recent filings after 2026-04-02How often the leading examiner received the case.
Examiner, top 561%300 recent filings after 2026-04-02How often the receiving examiner was among the five leading names.
Examiner, top 1094%300 recent filings after 2026-04-02How often the receiving examiner was among the ten listed.
Examiner, top 10, given the unit94%300 of those, within their own unitThe examiner stage alone: the ranking within the receiving unit, with the unit taken as correct.
Classification stepnot measuredn/aThe figures above take the classification picture as given. Drafting the picture from the claims is the harder step; published predictors working from claim text alone land near half of cases at top 1 and four in five at top 5 across all units, and the confidence card already reflects how contested this picture is.
Appendix

How the prediction is made

The appendix states how the office routes an application and how the probabilities in this report are formed from the record.

How the office routes an application

A new utility application is classified first. Every claimed concept receives a CPC symbol, and the symbols that carry the claimed invention are marked as the application's classification picture. An automated routing procedure then compares that picture with the portfolio of every examiner, the tally of symbols on the cases each examiner has already acted on, and weighs the size of each examiner's new docket, the backlog waiting in the same symbols, and whether the application continues an earlier one. The examiner who best fits receives the case. The art unit is the consequence of that match, not a separate decision.

How the probabilities are formed

The classification picture is drafted from the claims the way the office would draft it. Each symbol carries an observed distribution over art units from recent filings, blended with the units of the nearest recent filings by text, and with the parent's unit when a continuing relationship is claimed. Inside each candidate unit every examiner with new cases in the activity window is scored from the signals in the table, the scores are normalized to a probability within the unit, and the final figure for an examiner is the unit probability times the within-unit probability. An analyst's reading of each examiner's recent docket may adjust a score by a bounded factor, and every adjustment is stated in the reasons.

The signals

SignalWhat it measuresWhere it enters
Classification pictureThe CPC symbols the claims would receive, with the inventive symbol weighted above the additional ones.The art unit distribution and the examiner match.
Nearest recent filingsRecent applications whose titles and abstracts read closest to this one, and the units that received them.The art unit distribution.
FamilyA claimed parent, its unit and its examiner, weighted the way continuing applications are weighted in routing.Both the unit and the examiner.
Intake shareThe examiner's share of the unit's new cases in the activity window, which already reflects docket balancing.The examiner score.
Symbol shareThe examiner's share of the unit's recent cases carrying this picture's symbols.The examiner score.
Portfolio similarityThe cosine similarity between the picture and the examiner's symbol tally, at symbol, main group and subclass level.The examiner score.
QualificationThe weighted share of the picture's symbols that appear in the examiner's portfolio.The examiner score.
RecencyHow recently the examiner last received a new case; an examiner without new cases in the activity window is not a candidate.The examiner score and the candidate set.
Applicant historyEarlier cases of the same applicant with the examiner, which tend to share a classification picture.A bounded multiplier on the examiner score.
Analyst fitA reading of the examiner's recent docket against this application, bounded to between half and twice the score.A stated adjustment within the unit.