15 stories tagged llm.
Linear bias probes in LLM representations fail predictably when Fisher estimation error exceeds the eigengap, a spectral condition that makes probe instability diagnosable before deployment.
Between-day variance reaches 8.4 points, three times the 2.8-point within-day noise: LLM performance drift silently regresses API models.
IBM Granite 4.2 is a family of dense reasoning LLMs in 3B, 8B, and 30B sizes with native chain-of-thought and multi-stage agentic RL, all released under Apache 2.0.
Sub-2-bit LLM quantization puts a 250M model in 60 MB of disk and 80 MB of RAM, running at 400 tok/s on a laptop CPU. The method keeps recent tokens at full precision and compresses the rest, pointing toward a new class of lightweight deployments.
LLM instruction following breaks past 5 to 6 simultaneous constraints. Per-constraint pass rates decay at 72% times 0.922 per step, probe success below 2% by k=9.
Gemini 3.7 Flash is Google's coding and agent workhorse. It beats 3.6 Flash across every benchmark Google published and undercuts Claude Sonnet 5 on output token price by roughly two-thirds, at $0.75 per million input tokens.
Linear probes detect LLM errors that confidence signals miss, exposing a knowing-saying gap. Probe AUROC hits .931 versus .508 for logprob confidence within incorrect trials.
The Format Sensitivity Index measures how LLM benchmark scores shift when prompt wrappers and schema constraints change. The metric exposes a blind spot in model evaluation that developers ignore at their peril.
Scaling creativity-specific weights, CreativityNeuro improves LLM divergent thinking by up to 14 percentile points and reduces mode collapse without fine-tuning.
Models tend to converge on similar answers. In Springboards tests, Flint scored 7.47 distinct replies out of 10 against LLM groupthink.
Novel Search Space breaks an LLM out of its prior by ranking 80,000 dictionary words by embedding distance, banning the obvious neighbours, and forcing the model to brainstorm only from a surprising-but-related band.
Famous skincare brands were recommended 100 percent of the time in a new arXiv study, a measurable LLM recommendation bias toward incumbents.
Getting content cited in ChatGPT, Perplexity and AI Overviews takes specific tactics. What generative engine optimization rewards and what to change.
One open connector lets AI models call your tools and data instead of a custom integration per model. What MCP means for data engineers.
Snowflake Cortex AI exposes LLMs as SQL functions like AI_COMPLETE and AI_CLASSIFY, so you run inference without moving data out. It is billed per token by model, and cost swings roughly 40x between a small model and a frontier one, so model choice is the budget.