11 results

Preprint

Calibrated Is Not Informative: What a Fall in Expected Calibration Error Shows About LLM Confidence, Why Verbalised Scores Fail on Resolution as Well as Level, and the Recalibration Baseline the Ensembling Results Omit

Pranay Mahendrakar – 2026

(c) 2026 Pranay Mahendrakar. Licensed under CC BY 4.0. Life of Research A common finding in work on language-model confidence is that a model asked to state its own confidence in one pass is badly overconfident, and that some added procedure - sampling several answers, a separate critique pass, a…

Cite Related research LOR-00133 · 13 views · 10 downloads Select “Calibrated Is Not Informative: What a Fall in Expected…” for export

PDF352.5 KB

Preprint

A Starting Score Is Not a Ceiling: When Accuracy Before Reinforcement Learning Predicts Accuracy After It, Why Each Proposed Replacement Sees a Different Failure, and the Axis Test That Decides Which Proxy to Trust

Pranay Mahendrakar – 2026

(c) 2026 Pranay Mahendrakar. Licensed under CC BY 4.0. Life of Research Reasoning models are usually built in two stages, supervised fine-tuning and then reinforcement learning against a verifier, and the first stage is usually judged by the benchmark accuracy of its checkpoint. An August 2026…

Cite Related research LOR-00129 · 10 views · 11 downloads Select “A Starting Score Is Not a Ceiling: When Accuracy Before…” for export

PDF326 KB

Preprint

A Misalignment Rate Is Not a Model Property: Capacity, Propensity, Efficacy and Targeting in Multi-Agent LLM Commerce, Why a Null Capability Correlation Does Not Contradict the Scaling of Deceptive Skill, and the Design That Would Separate Them

Pranay Mahendrakar – 2026

(c) 2026 Pranay Mahendrakar. Licensed under CC BY 4.0. Life of Research As language-model agents begin to trade with one another on behalf of separate principals, a natural worry is that the strongest agents will mislead, pressure or collude against the weakest. A 2026 study of 2,583 emails…

Cite Related research LOR-00128 · 12 views · 23 downloads Select “A Misalignment Rate Is Not a Model Property: Capacity…” for export

PDF400.9 KB

Preprint

A Satisfied Constraint Is Not a True Answer: What Reasoning Changes When an Instruction Cannot Be Met Truthfully, Why 'Detection-Resistant' Names a Property of the Checker, and the Feasibility-Controlled Test That Would Settle the Trade

Pranay Mahendrakar – 2026

(c) 2026 Pranay Mahendrakar. Licensed under CC BY 4.0. Life of Research A language model asked for something its knowledge cannot honestly supply has four ways out: it can break the instruction openly, decline, bend a real fact until it fits, or invent one. A January 2026 study reports that…

Cite Related research LOR-00124 · 12 views · 21 downloads Select “A Satisfied Constraint Is Not a True Answer: What…” for export

PDF407.4 KB

Journal article

Who Spoke Next Is Not Who Should Have: What Zero-Shot Failure on Multi-Party Turn-Taking Benchmarks Measures, Why the Human Reference Sits Beside the Models, and What Would Show That Knowing When to Speak Must Be Trained

Pranay Mahendrakar – 2026

(c) 2026 Pranay Mahendrakar. Licensed under CC BY 4.0. An assistant that joins a group conversation has to decide, at every pause, whether to take the floor. Two 2026 benchmarks report that language models prompted zero-shot make this decision badly and that fine-tuning repairs it, and both…

Cite Related research LOR-00111 · 13 views · 30 downloads Select “Who Spoke Next Is Not Who Should Have: What Zero-Shot…” for export

PDF420.3 KB

Journal article

Three Rounds on Emergent Analogy: What the Webb, Hodel-West and Lewis-Mitchell Exchange Settled, How Its Third Round Tested a System While Still Claiming the Model, and Whether 'Counting' Names the Step That Analogy Theory Calls Inference

Pranay Mahendrakar – 2026

(c) 2026 Pranay Mahendrakar. Licensed under CC BY 4.0. In 2023 a large language model was reported to solve text-based analogy problems zero-shot at or above the level of college students. Two critiques followed. They showed that performance on letter-string analogies collapses when the alphabet…

Cite Related research LOR-00070 · 18 views · 33 downloads Select “Three Rounds on Emergent Analogy: What the Webb…” for export

PDF439.3 KB

Journal article

Does a Model Forget Differently When the Data Is Its Own? RL's Retention Advantage and Model Collapse Are Claims About the Same Loop, and No Study Has Measured Both

Pranay Mahendrakar – 2026

(c) 2026 Pranay Mahendrakar. Licensed under CC BY 4.0. Two literatures make claims about what happens when a language model trains on data that resembles its own distribution. One asks whether reinforcement learning forgets a model's prior capabilities less than supervised fine-tuning does, and…

Cite Related research LOR-00052 · 15 views · 34 downloads Select “Does a Model Forget Differently When the Data Is Its Own?…” for export

PDF9.3 MB

Journal article

Faithful to What? Four Instruments for Chain-of-Thought Faithfulness Disagree With Each Other, and the One Ground-Truth Check Run So Far Found Most of Them Near Chance

Pranay Mahendrakar – 2026

(c) 2026 Pranay Mahendrakar. Licensed under CC BY 4.0. A chain-of-thought (CoT) trace is called "faithful" when it accurately represents the computation that produced the model's answer, as opposed to a plausible-sounding story invented after the fact (Jacovi and Goldberg, 2020). Whether published…

Cite Related research LOR-00050 · 38 views · 71 downloads Select “Faithful to What? Four Instruments for Chain-of-Thought…” for export

PDF573.3 KB

Journal article

Poison Below the Base Rate: What the Near-Constant Poisoning Result Establishes About Detection, and Four Things It Does Not

Pranay M. Mahendrakar – 2026

(c) 2026 Pranay Mahendrakar. Licensed under CC BY 4.0. A widely cited 2025 result reports that backdooring a language model through its training data takes a near-constant number of poisoned documents rather than a constant fraction of the corpus, and the field has read that as retiring the…

Cite Related research LOR-00036 · 12 views · 26 downloads Select “Poison Below the Base Rate: What the Near-Constant…” for export

PDF441.6 KB

Journal article

The Crossover Does Not Carry the Claim: Pass@k Inversion, the Reasoning Boundary, and Five Premises the RLVR Debate Leaves Unstated

Pranay M. Mahendrakar – 2026

(c) 2026 Pranay Mahendrakar. Licensed under CC BY 4.0. A model trained with reinforcement learning from verifiable rewards beats its base model at one sample and loses to it at many. The field has read that crossover as a measurement of the reasoning boundary and has spent two years arguing about…

Cite Related research All 2 versions: LOR-00034 LOR-00072 · 14 views · 13 downloads Select “The Crossover Does Not Carry the Claim: Pass@k Inversion…” for export

PDF461 KB