llm-kb
← Back to research
Paper

Self-Guided Test-Time Training Improves Long-Context LLM Accuracy

A new arxiv paper proposes Self-Guided Test-Time Training (SG-TTT) to improve long-context utilization in LLMs without requiring labeled data. The method uses the model's own predictions to generate pseudo-labels for fine-tuning on the test context, addressing accuracy degradation in long inputs. This approach is more efficient than full TTT and shows promise for practical deployment.

0 engagement·1 source·Fri, Jul 10, 2026, 01:45 PM
The paper, posted on arxiv (2026-07-10), addresses the problem that LLMs with extended context windows still fail to effectively use long inputs, leading to accuracy drops. SG-TTT adapts model parameters at test time using the test context itself, generating pseudo-labels from the model's own outputs to guide fine-tuning. This avoids the expense of full TTT on the entire context. The method is designed to be self-supervised, requiring no external labels. While specific benchmarks and model names are not provided in the excerpt, the approach targets a key practical challenge in deploying long-context LLMs.

Entities

Self-Guided Test-Time Training(concept)Test-Time Training(concept)Long-Context LLMs(concept)

Related

PaperFri, Jul 10, 2026, 09:31 AM

Researchers identify asymmetric generalization problem in LLM unlearning benchmarks

A new arXiv paper argues that existing machine unlearning benchmarks for LLMs suffer from under-forgetting and over-forgetting due to an asymmetric generalization problem. The authors propose that evaluation must cover diverse query formulations of target facts to reliably measure knowledge removal while preserving unrelated capabilities.

0 engagement·1 source·arxiv
arXiv
PaperTue, Jul 7, 2026, 05:26 PM

Two papers propose token-adaptive KV cache compression for long-context LLMs

Two arXiv papers from July 7, 2026 introduce token-adaptive KV cache compression methods for long-context LLM inference. DepthWeave-KV factorizes key/value states across neighboring layers using shared low-rank bases with token-specific residuals. FreqDepthKV uses shared low-frequency depth components and sparse high-frequency residuals, with an online probe assigning attention heads to different cache modes. Both aim to reduce memory bandwidth while preserving retrieval and reasoning quality.

0 engagement·2 sources·arxiv
arXiv
PaperThu, Jul 9, 2026, 09:53 PM

NL-PAC paper introduces framework for handling specification ambiguity in LLM supervision

A new arXiv paper proposes NL-PAC, a framework that addresses specification ambiguity when LLMs provide labels or evaluations from natural language instructions. The framework uses a fixed model's thresholded decoding law to define admissible labels and candidate targets, showing that additional labels reduce sampling error but cannot resolve identification problems from ambiguous specifications.

0 engagement·1 source·arxiv
arXiv
PaperFri, Jul 10, 2026, 10:55 AM

Super-Tuning: Activation-Aware Pruning Reused for Sparse Fine-Tuning

A new arxiv paper proposes Super, a sparse PEFT method that reuses Wanda-style activation-weighted magnitude scores from pruning to select a small trainable support, and Supra, a hybrid adapter combining sparse updates with LoRA. This approach reduces memory and compute for fine-tuning LLMs while maintaining performance.

0 engagement·1 source·arxiv
arXiv
CommunitySun, Jul 12, 2026, 12:12 PM

Developer seeks feedback on fine-tuning LoRA for conversation state extraction in long LLM chats

A developer is working on a side project to improve AI conversation continuity by training a small model to extract structured conversation state from chat chunks, rather than relying on summarization. They are seeking feedback on their approach involving fine-tuning a LoRA, dataset design, and long-context systems.

19 engagement·2 sources·reddit