TrafficInstruct dataset and benchmark for instruction tuning on network traffic released
55,000 structured records support 10 detection tasks with task-specific LoRA adapters improving performance.
Stories per day · 30 days
2 RSS
Relateddata-poisoning2backdoor1benchmark1botnet-detection1conversational-ai1cti1ctibench1data-quality1data-selection1dataset1fine-tuning1finetuning1language-models1llm-as-judge1llms1llm-trojan1malware-detection1model-safety1
10 stories, newest first · one card per story
Stories tagged #instruction-tuning
Response ratio predicts instruction-tuning utility better than common proxies
Under a fixed finetuning budget, answer-dense examples beat common quality proxies on perplexity and LLM-as-judge scores.
Paper frames instruction-tuned models as an instructional-conversational genre
The argument ties the genre to instruction fine-tuning, system prompts, special tokens and role-based turn-taking.
Paper shows utility-hijacking can poison instruction-data selectors
Generate-and-rerank keeps malicious records in top-k selection; CONCERT cuts that concentration without poison labels.
LLM Trojans that increase response lengths and cost
The authors also describe a defense and a trigger that fires on semantically similar instructions.
Towards Efficient Threat Intelligence Instruction Tuning via Quality-Aware and Optimal Transport-Guided Data Selection
CTI-TDS selects compact, high-quality instruction data for threat-intelligence model tuning.
Self-Preservation in Language Models
Qwen models can generate replay and sometimes choose tasks that better preserve prior capabilities.
ImIR: Image-Instruction Tuning for All-in-One Image Restoration
ImIR replaces text prompts with image-derived instructions so one Qwen adapter restores many degradations.
Drift-Constrained Optimization: Only Direction Matters in Fine-Tuning Instruct Models
Drift-Constrained Optimization reformulates fine-tuning as update-direction selection, letting Qwen3 models improve target tasks within a behavioral drift budget.
Nuha-Speech: Building General-Purpose Arabic Speech-LLMs
Nuha-Speech initiative builds general-purpose Arabic speech-LLMs using a 1.5M-sample speech QA corpus and fine-tuned Qwen-Omni variants.