NLP & Language
Research on NLP & Language: explanations, evidence, limitations and connections between papers.
65 results
Trace the evidence path before changing document question answering
A wrong answer can arise even when useful evidence exists: a system may keep searching instead of reading, lose an author's visual mark, present a page unclearly, or consult the wrong output file. The studies suggest tracing where evidence became hard to find, interpret, or use, then testing a change at that point. Each study establishes an outcome for its own task and evaluation.
Parallel compilation speeds finite-action language model control
A compiler evaluates finite-action language model choices in parallel, matching serial control on most questions in four QASPER splits.
RIFT keeps draft caches useful as generation grows
RIFT combines recent and selected context in draft caches, improving accepted-token measures in specified Qwen settings.
Identity matching weakens when names switch scripts
A cross-script identity test finds Qwen misses same-person name matches; shifting its answer cutoff explains much, but not all.
Training guardrails across formats improves safety judgments
UniGuard trains safety guardrails across aligned input formats and improves classification and confidence on three unfamiliar-interface tests.
A benchmark reveals when language models set reports aside
A two-urn test separates declined reports from trust, revealing information loss hidden by average reliability ratings.
Compact encoders face a trade-off in how they read options
A controlled Crivo comparison found higher decision-test skill for separate answer reading from a reranker, but not a clear raw-encoder advantage.
One language model learns retrieval and answer generation
ULME links retrieval and answer generation in one language model; tests show stronger retrieval scores on three Qwen3 sizes.
Models solve fewer problems after a structural change makes them NP-complete
A paired coding benchmark finds models solve NP-complete problems less reliably than polynomial-time counterparts, even with more time.
DynaWeave turns static tool tasks into changing scenarios
DynaWeave turns static tool tasks into changing scenarios; trained Qwen3.5 models improved across three dynamic tests.
Models that solve best are not always the best task authors
Environment Duels finds that strong language model task solvers are not necessarily the strongest authors of useful challenges.
Why preference ranking can improve as text generation collapses
ECHO traces how shared hidden-state shifts can preserve response rankings while disrupting text generation in tested language models.
A clause-based test exposes conflicts with professional ethics rules
TenetBench tests answers against named professional rules; most evaluated model rows often conflicted with the rule on improper requests.
A document parser preserves visual cues agents need to find requirements
A formatting-aware parser helps an answering model recover author-marked requirements in a real-policy-document task.
ProVe speeds token generation by checking proposed blocks
ProVe checks multi-token proposals, matching greedy output while increasing measured generation speed in six Qwen instruction tests.
A coordinated system makes low-bit language models generate in blocks
BitMTP coordinates low-bit block drafting and runtime execution, raising measured generation speed while keeping tested mean scores close to baselines.
Agents can drop old cache states without rewriting their history
Cache engineering lets language-model agents drop old cached states; tests report lower cost and improved accuracy in a bounded setup.
A repaired teacher helps agents finish more tasks
A teacher-score repair method improved task completion across the tested text-agent models and environments.
Monsoon builds speech data that improves conversational recognition
Monsoon collects verified spontaneous speech; its Whisper models reduce errors on independently collected conversational tests.
TokenGate cuts long-prompt processing while preserving accuracy
TokenGate selects prompt tokens for later processing, reducing prefill time and memory without accuracy loss across five tested tasks.
A system helps visual agents gather and reuse evidence
OpenVLHarness helps visual agents gather and reuse tool evidence; Qwen3-VL-8B scored higher on average across 22 visual task datasets.
Measure the consequence before trusting a model signal
A byte model can be confident that the next symbol is a digit while missing the calculation that determines its value. A structured judgment can fit its required format while changing a program’s action. Several accurate observers can still fail together. These studies suggest a practical sequence: identify the decision, then test the output or behavior on which it depends. The relevant endpoint differs across computed answers, program decisions, and the benefit of acting with shared errors. Each finding belongs to its own tested setting.
Match cache changes to the cost your workload pays
A smaller cache can save stored memory, while selective reads can reduce the information fetched for each new query. Removing visual entries during answer generation changes a third cost. A quality score is useful for choosing among these approaches only if its test exposes the phase being changed. The studies point to a practical first step: identify the constrained resource, then test the corresponding method on the intended workload. Their classification, language-modeling, and visual-answer results describe different outcomes, so they do not yield a shared ranking.
Non-English requests more often enter retrieval
A multilingual retrieval-agent evaluation finds higher tool admission for non-English harmful requests in a controlled URL setting.