Human Feedback and Eval Paper Explorer

Match reason: Keyword overlap 1/1 across title and protocol fields.

• Experimental results demonstrate that SparseBalance achieves up to a 1.33\times end-to-end speedup while still improving the long-context capability by 0.46\% on the LongBench benchmark.

Match reason: Keyword overlap 1/1 across title and protocol fields.

• Extensive experiments across multiple benchmarks demonstrate the effectiveness of our approach: it outperforms a strong baseline by an average margin of 7.7\% (improving from 46.3\% to 54.0\%) across Loogle, Loong, LongBench-v2,…

Match reason: Keyword overlap 1/1 across title and protocol fields.

• Experimental results on the LongBench and RULER benchmarks demonstrate that StructKV effectively preserves long-range dependencies and retrieval robustness.

Match reason: Keyword overlap 1/1 across title and protocol fields.

• Extensive experiments across various models and benchmarks demonstrate that KVSlimmer consistently outperforms SOTA methods.

Match reason: Keyword overlap 1/1 across title and protocol fields.

• Across long-context benchmarks including LongBench and Needle-in-a-Haystack, MAGE achieves near-lossless accuracy with a fraction of the KV budget while delivering up to 3-4x end-to-end speedup, consistently outperforming AR-oriented sparse…

Match reason: Keyword overlap 1/1 across title and protocol fields.

• However, it often relies on gold-standard answers or explicit evaluation rubrics provided by powerful teacher models or human experts, which are costly and time-consuming.
• In this work, we investigate unsupervised approaches to enhance the long-context capabilities of LLMs, eliminating the need for heavy human annotations or teacher models' supervision.

Match reason: Keyword overlap 1/1 across title and protocol fields.

• Across multiple instruction-tuned and reasoning models, DYSCO consistently improves performance on challenging long-context reasoning benchmarks, yielding relative gains of up to 25% on MRCR and LongBenchV2 at 128K context length with…

Match reason: Keyword overlap 1/1 across title and protocol fields.

• Abstract shows limited direct human-feedback or evaluation-protocol detail; use as adjacent methodological context.

Match reason: Keyword overlap 1/1 across title and protocol fields.

• We test KVTC with Llama 3, Mistral NeMo, and R1-Qwen 2.5 models across benchmarks including AIME25, GSM8K, LiveCodeBench, LongBench, MATH-500, MMLU, Qasper and RULER.

Match reason: Keyword overlap 1/1 across title and protocol fields.

• Across mathematical reasoning (GSM8K, MATH-500, AIME24), procedural generation (LongProc), conversational long-memory benchmarks (LongMemEval), and long-context understanding (LongBenchV2 and SCBench), TRIM-KV consistently outperforms…
• Qualitative analyses further reveal that learned retention scores align with human intuition, naturally recovering heuristics such as sink tokens, sliding windows, and gist compression without explicit design.

Match reason: Matched by broad semantic/index fallback.

• We systematically study continued pretraining, supervised finetuning, and preference optimization for 24B and 32B parameter models, backed by extensive LC evaluations and ablations to bridge this gap, and achieve state-of-the-art…
• In addition to this, our key findings include: (i) training on context lengths that match evaluation context lengths outperforms training on longer contexts, (ii) training and evaluating with page indices provides a simple, high-impact…