LCDD creates sparse carriers for SFT behaviors that SFT-Eraser can reverse, with ablations showing the sparse structure enables causal control.
Title resolution pending
24 Pith papers cite this work. Polarity classification is still indexing.
24
Pith papers citing it
citation-role summary
background 2
baseline 1
citation-polarity summary
representative citing papers
WildChat releases a dataset of 1 million ChatGPT conversations with timestamps, demographics, and headers, claimed to be the most diverse and multilingual such resource available.
SHADOWMASK backdoors MDLMs by replacing the all-mask terminal distribution with a trigger-mask mixture prior, achieving near-100% attack success on DiT and LLaDA-8B models across multiple datasets while resisting fine-tuning and some defenses.
DuoAttention identifies retrieval heads requiring full KV cache and streaming heads using constant-length cache to reduce memory and latency in long-context LLM inference.
Magpie synthesizes 300K high-quality alignment instructions from Llama-3-Instruct via auto-regressive prompting on partial templates, enabling fine-tuned models to match official instruct performance on AlpacaEval, ArenaHard, and WildBench.
Iterative self-rewarding via LLM-as-Judge in DPO training on Llama 2 70B improves instruction following and self-evaluation, outperforming GPT-4 on AlpacaEval 2.0.
EvoPrompt uses LLMs to run evolutionary operators on populations of prompts, outperforming human-engineered prompts by up to 25% on BIG-Bench Hard tasks across 31 datasets.
Fine-tuning a 65B model on 1,000 high-quality examples produces output that humans rate as good as or better than GPT-4 in 43% of cases, indicating most capabilities come from pretraining.
WizardLM uses LLM-driven iterative rewriting to generate complex instruction data and fine-tunes LLaMA to reach over 90% of ChatGPT capacity on 17 of 29 evaluated skills.
LLaMA-Adapter turns frozen LLaMA 7B into a capable instruction follower using only 1.2M new parameters and zero-init attention, matching Alpaca while extending to image-conditioned reasoning on ScienceQA and COCO.
Refusal suppression via difference-in-means ablation equals projection onto a linear probe's decision boundary, and a controlled evasion attack optimizing confidence past the boundary achieves SOTA success rates on 15 models.
The paper introduces a dynamical model that decomposes alignment updates in LLM fine-tuning into rebound and driving forces and predicts a rehearsal priming effect.
Gradient-informed placement of LoRA parameters recovers full performance under GRPO while random placement does not, due to differences in gradient rank and stability across training regimes.
NCO enables efficient online pattern matching for negative hard and regex constraints in LLM decoding to prevent forbidden content without state explosion.
StreamingLLM lets finite-window LLMs generalize to infinite-length sequences by retaining initial-token KV states as attention sinks, enabling stable streaming inference up to 4M tokens.
ToRA trains language models on interactive tool-use trajectories with imitation learning and output shaping to integrate reasoning and external tools, yielding 13-19% gains on math datasets and new highs like 44.6% on MATH for a 7B model.
MAmmoTH models trained via hybrid CoT-PoT instruction tuning on MathInstruct outperform prior open-source LLMs by 16-32% average accuracy on nine math datasets, reaching 33% and 44% on MATH for 7B and 34B scales.
Baseline defenses including perplexity-based detection, input preprocessing, and adversarial training offer partial robustness to text adversarial attacks on LLMs, with challenges arising from weak discrete optimizers.
GPT-4 as an LLM judge achieves over 80% agreement with human preferences on MT-Bench and Chatbot Arena, matching human agreement levels and providing a scalable evaluation method.
A 13B model called Orca learns detailed reasoning from GPT-4 explanation traces and reaches parity with ChatGPT on Big-Bench Hard while outperforming other 13B models.
FINCH is a loss-adaptive learning-rate schedule that reduces forgetting by 93% on average during LLM fine-tuning while matching standard task performance across several benchmarks.
QWHA proposes Walsh-Hadamard Transform adapters with adaptive initialization for quantization-aware PEFT, claiming better low-bit accuracy and faster training than low-rank or other FT-based baselines.
IFEval is a new benchmark of 25 verifiable instruction types and ~500 prompts for objective, reproducible evaluation of LLMs' instruction-following abilities.
LoRA-FA freezes LoRA's A matrix and trains only B with gradient corrections to approximate full fine-tuning gradients more closely.
citing papers explorer
-
WildChat: 1M ChatGPT Interaction Logs in the Wild
WildChat releases a dataset of 1 million ChatGPT conversations with timestamps, demographics, and headers, claimed to be the most diverse and multilingual such resource available.
-
DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads
DuoAttention identifies retrieval heads requiring full KV cache and streaming heads using constant-length cache to reduce memory and latency in long-context LLM inference.
-
Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing
Magpie synthesizes 300K high-quality alignment instructions from Llama-3-Instruct via auto-regressive prompting on partial templates, enabling fine-tuned models to match official instruct performance on AlpacaEval, ArenaHard, and WildBench.
-
Self-Rewarding Language Models
Iterative self-rewarding via LLM-as-Judge in DPO training on Llama 2 70B improves instruction following and self-evaluation, outperforming GPT-4 on AlpacaEval 2.0.