Author

Samrath Singh Chadha

h-index 1 View on Semantic Scholar

We have 5 of 5 papers

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Preprint Sep 2026

A Heckler in the Hidden State: Correctness Signals in Diffusion Language Models

Diffusion language models generate code by repeatedly updating a partially masked sequence. We ask whether their internal activations encode code correctness and whether that information can improve generation. Across six diffusion models, linear probes distinguish passing from failing attempts, with the strongest read...

A. Miglani, Samrath Singh Chadha, Kevin Li et al. · 0 citations

#small language model Preprint Sep 2026

The Editor Has Read-Only Access: Correctness Signals in Diffusion Language Models

Across six diffusion models, linear probes distinguish passing from failing attempts, with the strongest reads generally appearing beyond the early layers, and probe point estimates offer no consistent advantage in comparisons with model confidence.

A. Miglani, Samrath Singh Chadha, Kevin Li et al. · 0 citations

#machine learning Preprint Sep 2026

Decodable In-Context State and Model Output Across Training

This work follows probe accuracy, model output, and steering response across public pretraining and post-training checkpoints and explains why decodability on errors alone cannot establish discarded output information.

Manas Venkata Sai Ravulapalli, Samrath Singh Chadha · 0 citations

#machine learning Preprint Sep 2026

In-Context Binding Capacity in Language Models

This work derives why interference can lower measured capacity by reducing single-binding recall even when the load-dependent recall profile is unchanged, and characterize capacity at the model's query interface.

Manas Venkata Sai Ravulapalli, Samrath Singh Chadha · 0 citations

#machine learning Preprint Sep 2026

Legible Failures: Detecting and Repairing In-Context Binding Errors

Where recent studies report that probe-detected errors are resistant to interventions, it is found that in-context binding is a setting in which probes are actionable.

Manas Venkata Sai Ravulapalli, Samrath Singh Chadha, Abhinav M. Hari · 1 citation