CLARITI: Disambiguation in Research

• CLARITI is a term applied to distinct research systems spanning clinical dialogue, software clarification, LLM reinforcement learning, and chest X-ray imaging.
• The clinical variant (CLARITY) uses a hybrid finite state machine and LLM microservices to achieve precise patient routing and emergency triage.
• The software and RL versions leverage reward-driven pipelines and structured constraints to enhance clarification quality and reasoning consistency.

Searching arXiv for the term and closely related variants to ground the article in the current literature. CLARITI is not a single canonical research artifact in the current arXiv literature. Rather, the label appears as a spelling variant, acronymic shorthand, or stylistic near-homograph for several technically distinct systems: CLARITY, a clinical assistant for routing, inference, and triage in telehealth; CLARITI, an 8B clarification module for software engineering tasks; CLARity, a reinforcement-learning framework for improving reasoning consistency in expert LLMs trained on MCQs; and CLARiTy, a vision transformer for chest X-ray classification and weakly supervised localization (Shaposhnikov et al., 2 Oct 2025, Vijayvargiya et al., 16 Apr 2026, Lin et al., 10 Oct 2025, Statheros et al., 18 Dec 2025). The term therefore requires contextual disambiguation: in medical dialogue systems it denotes a production clinical assistant, in software engineering it denotes a reward-trained clarification model, in LLM training it denotes a consistency-aware RL framework, and in medical imaging it denotes a class-localizing transformer architecture.

1. Terminological scope and disambiguation

Within the supplied literature, the string “CLARITI” is used in more than one way. In the clinical dialogue paper, the paper itself uses the name CLARITY, and the query form “CLARITI” is described as a spelling or acronym variant referring to the same system rather than to a separate module (Shaposhnikov et al., 2 Oct 2025). In the software-engineering paper, CLARITI is the canonical name and is expanded as “CLARIficiation with TIered rewards” (Vijayvargiya et al., 16 Apr 2026). In the chest X-ray paper, the query form “CLARITI” is explicitly tied to CLARiTy, expanded as “Class Localizing and Attention Refining Image Transformer” (Statheros et al., 18 Dec 2025). A further near-match, CLARity, names a reinforcement-learning framework for expert LLMs and expands to “Consistency-aware Learning with Data-Augmented 2-Stage Reinforcement Strategy” (Lin et al., 10 Oct 2025).

This naming convergence is substantive rather than merely orthographic. Each system operates in a different problem class, uses a different architecture, and is evaluated on different tasks. A plausible implication is that “CLARITI” functions in practice as an overloaded research label rather than a stable identifier for one model family.

A further source of confusion is CLARO, which is unrelated to these systems. CLARO is an 8-channel ASIC for the LHCb RICH upgrade and concerns high-rate photon readout rather than dialogue systems, clarification modules, or medical imaging (Cassina, 2015).

2. CLARITY as a clinical assistant for routing, inference, and triage

CLARITY is a production-grade, AI-driven medical dialogue system deployed as a virtual clinical assistant on a large national inter-hospital telehealth platform (Shaposhnikov et al., 2 Oct 2025). Its stated goals are patient-to-specialist routing, structured clinical consultations, and severity assessment with emergency escalation. It is designed to replace part of the traditional “patient → general practitioner → specialist” pathway by collecting anamnesis, generating diagnostic hypotheses, returning a ranked specialist list with short explanations, and triggering emergency escalation when indicated.

Its architecture is explicitly hybrid. A Finite State Machine (FSM) serves as the central chat manager, while multiple LLM-based and ML-based microservices provide text generation, medical specialty selection, moderation, emergency detection, readiness estimation, and question detection (Shaposhnikov et al., 2 Oct 2025). The FSM is formally defined as a Mealy-type machine,

$$M = (Q, \Sigma, \Omega, C, T, \delta, \lambda, q_0, q_d, q_{ca}, F),$$

with bounded internal execution controlled by N_ATTEMPTS. High-level states include Initialization, Information Collection, Diagnosis, Moderation, Emergency, and Free Dialogue. This bounded state-machine design is described as central to controlling LLM behavior, because the LLM is always called within state-specific templates and bounded transitions rather than in an unconstrained conversational loop (Shaposhnikov et al., 2 Oct 2025).

The information-collection subsystem combines a question generator, dialogue retrieval from a vector database, and a relevance classifier trained on 2,500 labeled examples. If cosine similarity to a historical dialogue exceeds 0.965, cached questions are reused; redundant questions are removed with a cosine cutoff of 0.86; and the relevance classifier attains precision 0.84 for the “relevant” label (Shaposhnikov et al., 2 Oct 2025). The routing core is the Medical Specialty Selector, which generates candidate diagnoses, maps each diagnosis to a specialty, and then produces brief explanations. Its evaluation uses pairwise precision and recall over multisets of specialist labels. Reported routing results are Precision@1 = 77±0.3%, Recall@2 = 92±0.4%, and Recall@3 = 96±0.4%, summarized in the paper as “77% precision for first recommendations” and “96% recall for top-3 recommendations” (Shaposhnikov et al., 2 Oct 2025).

Triage is operationalized through an emergency detector and a hard-coded emergency state. The emergency detector uses the full chat and a feature pipeline combining TF-IDF, one-hot critical words, and a binary LLM criticality feature, followed by PCA with $n_c = 423$ and histogram-based gradient boosting (Shaposhnikov et al., 2 Oct 2025). In deployment, the threshold is tuned to keep FPR = 0.02; reported performance is Precision = 0.72, Recall = 0.49, and $F_1 = 0.59$. When an emergency is flagged, the FSM transitions to a scripted protocol such as “Call 103 immediately,” preventing the LLM from overriding escalation (Shaposhnikov et al., 2 Oct 2025).

The deployment scale is notable. The second pilot processed 55,856 dialogues, with 92.3% progression after dialogue initiation, 54.7% of dialogues yielding diagnosis plus referral, and appointment conversions of 26.8% for online consults and 5.6% for offline consults (Shaposhnikov et al., 2 Oct 2025). The first pilot reported 1,534 initiated consultations, 62.13% ending with specialist recommendation plus diagnostic hypothesis, a mean appointment duration of 2:13 minutes, and referrals to 70 different specialties. The paper further reports that CLARITY is up to 3× shorter than a human GP consultation and describes its first-attempt routing precision as surpassing typical human-level performance in initial routing (Shaposhnikov et al., 2 Oct 2025).

3. CLARITI as a reward-driven clarification module for software engineering

In software engineering, CLARITI denotes a learned clarification module rather than a coding agent (Vijayvargiya et al., 16 Apr 2026). It sits upstream of a fixed downstream coding system, receives an underspecified issue, generates a small set of clarification questions, obtains answers from a simulated user, and passes the enriched context to Seed OSS 36B in OpenHands. The model itself is an 8B-parameter LLM trained specifically for question generation in software engineering tasks (Vijayvargiya et al., 16 Apr 2026).

The motivation is that real software issues are underspecified along several axes, but missing information is not uniformly valuable. The paper defines an information need as information absent from the issue description that is required for an agent to produce a correct patch without relying on unverifiable assumptions (Vijayvargiya et al., 16 Apr 2026). From expert annotations on SWE-Bench Verified, six categories are used: Error Information, Expected Behavior, Implementation Details, External References, Reproduction Steps, and Version/Environment. Clarification is formulated as a single-turn, set-generation problem: the model outputs a set of questions all at once rather than engaging in multi-turn ask–answer–ask interaction (Vijayvargiya et al., 16 Apr 2026).

The paper’s empirical design is driven by two analyses. First, it uses Shapley attribution on 1,500 underspecified variants of 500 SWE-Bench Verified issues to quantify which information categories most affect downstream success. Mean absolute SHAP values establish an impact hierarchy led by Error Information at approximately 0.183, followed by Implementation Details at approximately 0.0982, with the remaining categories lower but non-zero (Vijayvargiya et al., 16 Apr 2026). Second, it studies question answerability by comparing generated questions against both underspecified and fully specified issue descriptions. Distributional analysis yields four strategic themes associated with answerable questions: Ground in Evidence, Demand Specificity, Minimize Scope, and Ensure Actionability (Vijayvargiya et al., 16 Apr 2026).

These findings are operationalized as a four-stage reward pipeline trained with Group Relative Policy Optimization (GRPO) on top of Qwen3-8B, with a frozen SFT checkpoint as reference model and Qwen3-32B as judge (Vijayvargiya et al., 16 Apr 2026). The four reward components are:

• Non-redundancy
• Diversity
• Answerability
• Task relevance

The final reward is

$$r_{\text{final}} = \frac{1}{4}\Big( r_{\text{non-redundancy}} + r_{\text{diversity}} + r_{\text{answerability}} + r_{\text{relevance}} \Big).$$

Rejection filtering is applied stage-wise: if a generation fails Stage 1, 2, or 3, later stages are not evaluated and the total reward is set to zero (Vijayvargiya et al., 16 Apr 2026). Relevance is category-weighted according to the SHAP-derived impact hierarchy; answerability is estimated by asking whether the question can be answered from the fully specified original issue.

The main downstream results are reported on 250 underspecified issues. No clarification yields 22.4% success, GPT-5 nano reaches 29.6%, GPT-5 reaches 35.6% with 5.1 questions per issue, CLARITI reaches 36.8% with 3.0 questions per issue, and the full issue upper bound is 41.6% (Vijayvargiya et al., 16 Apr 2026). The paper therefore reports that CLARITI matches or slightly exceeds the GPT-5 clarification baseline while using 41% fewer questions, and that it recovers approximately 88% of the fully specified performance ceiling. Its answerability score is 0.373 and its relevance score is 0.622, both measured on clarification sets (Vijayvargiya et al., 16 Apr 2026).

Category allocation explains much of this efficiency. CLARITI assigns 26.4% of its questions to Error Information, compared with 10.2% for GPT-5 and 6.0% for GPT-5 nano (Vijayvargiya et al., 16 Apr 2026). The ablations are also strong: SFT only yields 24.2% success, removing the answerability stage yields 24.4%, removing the relevance stage yields 32.0%, and the full four-stage system reaches 36.8% (Vijayvargiya et al., 16 Apr 2026). The paper interprets this as evidence that both answerability and relevance are necessary; high-relevance but unanswerable questions degrade performance, and merely plausible question generation without RL is insufficient.

4. CLARity as a consistency-aware RL framework for expert LLMs

A different system, spelled CLARity, is a reinforcement-learning framework for training expert LLMs in scarce-data domains such as law and medicine (Lin et al., 10 Oct 2025). Its starting point is the observation that standard outcome-based RL on multiple-choice questions can improve answer accuracy while degrading reasoning quality, especially the consistency between chain-of-thought and final answer. On the legal MCQ dataset JEC-QA, the paper reports that after standard GRPO-based RL, the proportion of responses with logical fallacies or inconsistencies rises from 7% to 31% (Lin et al., 10 Oct 2025).

The central construct is reasoning consistency: whether the options endorsed in the model’s reasoning match the options chosen in its final answer (Lin et al., 10 Oct 2025). To exploit this, CLARity uses a small general-purpose LLM as a parser rather than as a domain expert. The expert model outputs > ... and <answer> ... </answer>, and a separate Qwen2.5-7B-Instruct instance parses the reasoning into a set such as {A}, {ABD}, or {}. The core consistency reward is

$$R_{\text{consistency}} = \mathbb{I}_{\text{Ans}_{\text{parsed}} = \text{Ans}_{\text{gen}}},$$

while answer correctness is

$$R_{\text{ans}} = \mathbb{I}_{\text{Ans}_{\text{gen}} = A_i}.$$

In Stage 2, the implementation uses 0 for consistent outputs and $-1$ for inconsistent ones, plus a similar small penalty for malformed format (Lin et al., 10 Oct 2025).

The training pipeline is explicitly two-stage. Stage 1 (Refine) shapes outputs into explicit option-wise reasoning using only a format reward and a structure reward; it does not optimize correctness (Lin et al., 10 Oct 2025). Stage 2 (Monitor) drops the rigid per-option structural constraint, retains the <think>/<answer> format, and adds consistency and answer-correctness rewards:

$$R = 0.1 R_{\text{format}} + 1.0 R_{\text{consistency}} + 1.0 R_{\text{ans}}.$$

This is combined with dynamic data reformulation, which deconstructs easy MCQs into propositions, rewrites them with DeepSeek-V3, and regroups them into harder composite MCQs according to a curriculum based on estimated pass rate (Lin et al., 10 Oct 2025).

The framework is evaluated with Qwen2.5-3B-Instruct and Qwen2.5-7B-Instruct as expert models on JEC-QA and MedQA-USMLE (Lin et al., 10 Oct 2025). The primary metrics are Accuracy (Acc), Consistency ratio (Cons%), and Acc$^+$, the fraction of responses that are both correct and consistent. Averaged over law and medical MCQs, the reported values for the 7B model are:

Method	Acc$^+$	Cons%	Acc
Vanilla Qwen2.5-7B	35.5	90.5	39.4
Standard MCQ RL	51.4	75.3	61.1
PRM RL	51.1	82.3	57.8
CLARity$n_c = 423$0	58.9	91.8	61.3

These results support the paper’s headline claim that CLARity improves consistency by 16.5 points and Acc$n_c = 423$1 by 7.5 points over standard MCQ RL while maintaining essentially the same high accuracy (Lin et al., 10 Oct 2025). The paper further reports that, on validation variants involving iterative judgment and shuffled MCQs, CLARity$n_c = 423$2 achieves average Acc$n_c = 423$3 = 52.0 and Cons% = 94.6, close to GPT-4o at 52.2 and 95.2, respectively (Lin et al., 10 Oct 2025). Human evaluations also show gains in correctness, professionalism, and readability, despite supervision being based only on consistency (Lin et al., 10 Oct 2025).

A notable practical claim is cost-effectiveness. The reward model is a general-purpose 7B model with no domain-specific fine-tuning. Very small reward models perform poorly: reported inconsistency detection rates against a DeepSeek-V3 ground truth are 21% for Qwen2.5-0.5B, 40% for Qwen2.5-1.5B, and 94% for Qwen2.5-7B (Lin et al., 10 Oct 2025). This suggests a minimum viable reward-model capacity, but still at substantially lower cost than large PRM-style supervision.

5. CLARiTy as a vision transformer for chest X-ray pathology classification and localization

In medical imaging, CLARiTy is a pure Vision Transformer–based architecture for multi-label classification of 14 thoracic pathologies and weakly supervised localization of 8 pathologies on chest radiographs (Statheros et al., 18 Dec 2025). It is designed for the common setting in which image-level labels are abundant but region-level annotations are scarce. The model’s principal idea is to use ViT self-attention as a class-specific localization signal and then refine it via a dedicated module, SegmentCAM, using explicit anatomical priors (Statheros et al., 18 Dec 2025).

The backbone is ViT-S/16 with 12 transformer blocks, 6 attention heads per layer, embedding dimension $n_c = 423$4, and patch size $n_c = 423$5 (Statheros et al., 18 Dec 2025). Two configurations are reported:

• CLARiTy-S-16-224: input $n_c = 423$6, patch grid $n_c = 423$7
• CLARiTy-S-16-512: input $n_c = 423$8, patch grid $n_c = 423$9

Instead of a single CLS token, CLARiTy uses 14 class-specific tokens, one per pathology (Statheros et al., 18 Dec 2025). The self-attention matrix is partitioned so that the class→patch block yields class-specific attention maps, which are averaged over the last $F_1 = 0.59$0 layers:

$F_1 = 0.59$1

At the output, class-token logits $F_1 = 0.59$2 are combined with SegmentCAM foreground logits $F_1 = 0.59$3 by averaging:

$F_1 = 0.59$4

SegmentCAM operates on the final patch tokens and has two roles: foreground segmentation and background activation suppression (Statheros et al., 18 Dec 2025). It produces heatmaps $F_1 = 0.59$5, feature maps $F_1 = 0.59$6, and classwise foreground masks $F_1 = 0.59$7. Three losses constrain the heatmaps using automatically generated anatomical masks: mask proximity $F_1 = 0.59$8, mask confinement $F_1 = 0.59$9, and foreground area $$r_{\text{final}} = \frac{1}{4}\Big( r_{\text{non-redundancy}} + r_{\text{diversity}} + r_{\text{answerability}} + r_{\text{relevance}} \Big).$$0. A background activation suppression term,

$$r_{\text{final}} = \frac{1}{4}\Big( r_{\text{non-redundancy}} + r_{\text{diversity}} + r_{\text{answerability}} + r_{\text{relevance}} \Big).$$1

penalizes signal retained in background-only features for positive classes (Statheros et al., 18 Dec 2025). The model also uses an orthogonal class token (OCT) loss over the last $$r_{\text{final}} = \frac{1}{4}\Big( r_{\text{non-redundancy}} + r_{\text{diversity}} + r_{\text{answerability}} + r_{\text{relevance}} \Big).$$2 transformer layers and a learned attention pooling head instead of global average pooling. The full objective combines class-token classification loss, CAM losses, and OCT regularization (Statheros et al., 18 Dec 2025).

Training uses NIH ChestX-ray14 with 112,120 frontal CXRs, 30,805 patients, and 14 pathology labels (Statheros et al., 18 Dec 2025). Bounding boxes exist for 880 images and 984 boxes across 8 pathologies, used only for localization evaluation (Statheros et al., 18 Dec 2025). CLARiTy-S-16-224 is initialized from DINO self-supervised pretraining, while CLARiTy-S-16-512 is initialized from the 224 model with interpolated positional embeddings. The system is further distilled from a ConvNeXtV2-B teacher, whose reported NIH validation Macro AUC is 0.833 at 224 and 0.850 at 512 (Statheros et al., 18 Dec 2025).

The main test results separate classification from localization. For classification, Macro AUC is 0.799 for CLARiTy-S-16-224 and 0.818 for CLARiTy-S-16-512 (Statheros et al., 18 Dec 2025). For localization on the NIH localization test subset, Macro IoU [email protected] is 0.303 for the 224 model and 0.318 for the 512 model (Statheros et al., 18 Dec 2025). The paper highlights that the 50.7% gain in the abstract refers to the relative improvement of CLARiTy-S-16-224 over ThoraX-PriorNet-224 at IoU 0.5, and that CLARiTy-S-16-512 improves over ThoraX-PriorNet-512 by 46.5% relative at the same threshold (Statheros et al., 18 Dec 2025). Particularly large gains are reported for small pathologies such as nodules and masses: at IoU 0.5, the 512 model reaches 0.393 for nodules and 0.419 for masses (Statheros et al., 18 Dec 2025).

The computational footprint remains moderate for the lower-resolution model. CLARiTy-S-16-224 has 27.03 M parameters and 5.99 GFLOPs, whereas CLARiTy-S-16-512 has 27.35 M parameters and 37.84 GFLOPs (Statheros et al., 18 Dec 2025). The paper explicitly positions the 224 variant as suitable for lower-resource settings while still surpassing baselines decisively in localization (Statheros et al., 18 Dec 2025).

6. Cross-cutting patterns, distinctions, and common misconceptions

Although these systems share a near-identical name, their methodological cores are different. CLARITY uses a hybrid FSM + LLM + classical ML microservices design for safe medical dialogue orchestration (Shaposhnikov et al., 2 Oct 2025). CLARITI in software engineering uses a single-turn clarification policy trained with tiered intrinsic rewards derived from empirical analyses of information impact and user answerability (Vijayvargiya et al., 16 Apr 2026). CLARity for expert LLM training uses consistency-aware RL plus dynamic MCQ reformulation to prevent the collapse of reasoning quality under answer-only rewards (Lin et al., 10 Oct 2025). CLARiTy in radiology is a pure ViT augmented with multiple class tokens, SegmentCAM, anatomical priors, and distillation for classification and weakly supervised localization (Statheros et al., 18 Dec 2025).

There are, however, recurring design motifs. One is the use of structured constraints around a flexible model. In CLARITY, the FSM bounds LLM behavior (Shaposhnikov et al., 2 Oct 2025). In software-engineering CLARITI, the four-stage reward pipeline with rejection filtering constrains question generation before optimizing task relevance (Vijayvargiya et al., 16 Apr 2026). In CLARity, the refine-then-monitor pipeline constrains output structure before applying consistency-based supervision (Lin et al., 10 Oct 2025). In CLARiTy, anatomical priors and foreground-background decomposition constrain what kinds of activations are accepted as plausible localization (Statheros et al., 18 Dec 2025). This suggests a shared design philosophy: expressive models are embedded within explicit control structures rather than left fully unconstrained.

A second recurring motif is efficiency under scarce or weak supervision. CLARITY relies on 2,500 expert-annotated dialogues yet is deployed at national scale (Shaposhnikov et al., 2 Oct 2025). Software-engineering CLARITI uses intrinsic proxies instead of downstream outcome reward during RL because full coding-agent rollouts are too expensive (Vijayvargiya et al., 16 Apr 2026). CLARity avoids full process reward models and instead supervises only reasoning consistency with a small reward model (Lin et al., 10 Oct 2025). CLARiTy trains localization without pathology bounding boxes, using only image-level labels plus automatically generated anatomical masks (Statheros et al., 18 Dec 2025). A plausible implication is that the shared naming convergence is accompanied by a shared engineering bias toward extracting high-value supervision from limited annotation budgets.

A common misconception would be to treat “CLARITI” as one research platform spanning telehealth, software engineering, LLM reasoning, and radiology. The literature does not support that interpretation. These are separate artifacts with distinct problem formulations, datasets, and evaluation protocols (Shaposhnikov et al., 2 Oct 2025, Vijayvargiya et al., 16 Apr 2026, Lin et al., 10 Oct 2025, Statheros et al., 18 Dec 2025). The only safe use of the label is therefore context-specific: in a telehealth setting it most likely refers to CLARITY, in a software-engineering clarification setting to CLARITI, in reasoning-focused RL to CLARity, and in chest radiography to CLARiTy.