On-policy distillation has an extrapolation cliff at closed-form lambda*(p,b,c) set by teacher modal probability, warm-start mass, and clip strength, past which training shifts from format-preserving to format-collapsing.
Guiding LLMs The Right Way: Fast, Non-Invasive Constrained Generation
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
citation-role summary
background 3
method 1
citation-polarity summary
verdicts
UNVERDICTED 8polarities
background 4representative citing papers
NetNomos is a multi-stage framework that extracts, filters, and enforces first-order logic rules in generative ML models for networking tasks including telemetry imputation, traffic forecasting, and synthetic trace generation.
TruncProof lets LLMs generate syntactically valid JSON within strict token limits by approximating completion token counts via LL(1) parser lookahead.
TeCoD improves Text-to-SQL execution accuracy by up to 36% over in-context learning and cuts latency 2.2x on matched queries by extracting templates from historical pairs and enforcing them with constrained decoding.
Post-generation grammar and schema filtering on top of confidence scoring raises syntactic validity and execution success for Text2Cypher but increases empty outputs and lowers coverage.
Agent Mentor analyzes semantic trajectories in agent logs to identify undesired behaviors and derives corrective prompt instructions, yielding measurable accuracy gains on benchmark tasks across three agent setups.
CircuitSynth generates synthetic data with guaranteed schema validity and improved rare case coverage by using a distilled logical diagram from LLMs combined with convex optimization.
Cross-family LLM differences outweigh within-family scaling on orthographic constraint puzzles, with modest human difficulty calibration but systematic failures on atypical common words.
citing papers explorer
-
The Extrapolation Cliff in On-Policy Distillation of Near-Deterministic Structured Outputs
On-policy distillation has an extrapolation cliff at closed-form lambda*(p,b,c) set by teacher modal probability, warm-start mass, and clip strength, past which training shifts from format-preserving to format-collapsing.
-
Making Logic a First-Class Citizen in Generative ML for Networking
NetNomos is a multi-stage framework that extracts, filters, and enforces first-order logic rules in generative ML models for networking tasks including telemetry imputation, traffic forecasting, and synthetic trace generation.
-
TruncProof: A Guardrail for LLM-based JSON Generation under Token-Length Constraints
TruncProof lets LLMs generate syntactically valid JSON within strict token limits by approximating completion token counts via LL(1) parser lookahead.
-
Reliable Answers for Recurring Questions: Boosting Text-to-SQL Accuracy with Template Constrained Decoding
TeCoD improves Text-to-SQL execution accuracy by up to 36% over in-context learning and cuts latency 2.2x on matched queries by extracting templates from historical pairs and enforcing them with constrained decoding.
-
Extending Confidence-Based Text2Cypher with Grammar and Schema Aware Filtering
Post-generation grammar and schema filtering on top of confidence scoring raises syntactic validity and execution success for Text2Cypher but increases empty outputs and lowers coverage.
-
Agent Mentor: Framing Agent Knowledge through Semantic Trajectory Analysis
Agent Mentor analyzes semantic trajectories in agent logs to identify undesired behaviors and derives corrective prompt instructions, yielding measurable accuracy gains on benchmark tasks across three agent setups.
-
CircuitSynth: Reliable Synthetic Data Generation
CircuitSynth generates synthetic data with guaranteed schema validity and improved rare case coverage by using a distilled logical diagram from LLMs combined with convex optimization.
-
Orthographic Constraint Satisfaction and Human Difficulty Alignment in Large Language Models
Cross-family LLM differences outweigh within-family scaling on orthographic constraint puzzles, with modest human difficulty calibration but systematic failures on atypical common words.