Hugging Face Bio releases Carbon-X, a 14.8B DNA model
The open chat model, started from Qwen3-14B, reports 66.7% on five DNA benchmarks with thinking disabled.
The open chat model, started from Qwen3-14B, reports 66.7% on five DNA benchmarks with thinking disabled.
Waypoint-OPD gates RL policy-shift distillation and leads math and code scores for Qwen3 students.
LayerTrim prunes reasoning LLMs without training, cutting cost while preserving long-form accuracy.
Task context changes which facts shared language-model heads use and how they influence later computation.
A learned answer selector on Qwen3-14B improves TriviaQA coverage and lowers scored-answer errors versus confidence filtering.
SparseDecoding prunes LLMs using decoding activations, speeding generation up to 1.48x.
Fixed starting-token cues let base models recover much of the math and coding gains from RL training.
The cue Okay lifts Olmo-3-7B MATH-500 pass@1 from 42% to 78%, recovering much of RL's gain.
SlideDP scales host-resident multi-GPU LLM fine-tuning, beating offload baselines and nearing FSDP2 on Qwen3-14B.
Re-derivability, not extra stages, decides how staged LLM agents recover after an upstream fault.
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.
Today 245 readers · 261 views · 66 reads · all time 199K views