Dual-Track CoT: Budget-Aware Stepwise Guidance for Small LMs
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908997047549952 |
|---|---|
| author | Chatterjee, Sagnik Patil, Atharva Ramesh, Sricharan |
| author_facet | Chatterjee, Sagnik Patil, Atharva Ramesh, Sricharan |
| contents | Large Language Models (LLMs) solve many reasoning tasks via chain-of-thought (CoT) prompting, but smaller models (about 7 to 8B parameters) still struggle with multi-step reasoning under tight compute and token budgets. Existing test time reasoning methods such as self consistency (sampling multiple rationales and voting), Tree-of-Thoughts (search over intermediate thoughts), and critique revise loops improve performance, but often at high token cost and without fine-grained step-level control. This project1 aims to address that gap: can Small Language Models (SLMs) reason reliably using the same or fewer tokens? This question is both scientific and practical. Scientifically, it probes whether process supervision and simple test-time controls (such as token budgets and rejection of redundant steps) can substitute for model scale or large sampling counts. Practically, many deployments (on-device, low-latency, or cost-constrained settings) cannot afford huge models or dozens of sampled rationales per query. A method that improves SLM reasoning at fixed cost would therefore be directly useful. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_25039 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Dual-Track CoT: Budget-Aware Stepwise Guidance for Small LMs Chatterjee, Sagnik Patil, Atharva Ramesh, Sricharan Computation and Language Artificial Intelligence Large Language Models (LLMs) solve many reasoning tasks via chain-of-thought (CoT) prompting, but smaller models (about 7 to 8B parameters) still struggle with multi-step reasoning under tight compute and token budgets. Existing test time reasoning methods such as self consistency (sampling multiple rationales and voting), Tree-of-Thoughts (search over intermediate thoughts), and critique revise loops improve performance, but often at high token cost and without fine-grained step-level control. This project1 aims to address that gap: can Small Language Models (SLMs) reason reliably using the same or fewer tokens? This question is both scientific and practical. Scientifically, it probes whether process supervision and simple test-time controls (such as token budgets and rejection of redundant steps) can substitute for model scale or large sampling counts. Practically, many deployments (on-device, low-latency, or cost-constrained settings) cannot afford huge models or dozens of sampled rationales per query. A method that improves SLM reasoning at fixed cost would therefore be directly useful. |
| title | Dual-Track CoT: Budget-Aware Stepwise Guidance for Small LMs |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2604.25039 |