Dream-7B-Instruct
Dream-7B-Instruct is an instruction-tuned version of a 7-billion-parameter diffusion language model (dLLM) used for text generation and reasoning tasks.
Timeline (last 90 days)
The researchers conducted experiments on two models, LLaDA-8B-Instruct and Dream-7B-Instruct.
Compared with a strong adaptive allocation baseline, this method achieved absolute accuracy improvements of up to 3.73 percentage points (LLaDA) and 7.33 percentage points (Dream) on MATH-500.
Compared with a strong adaptive allocation baseline, this method achieved absolute coverage improvements of up to 11.39 percentage points (LLaDA) and 10.98 percentage points (Dream) on HumanEval.
All Facts (3)
The researchers conducted experiments on two models, LLaDA-8B-Instruct and Dream-7B-Instruct.
50%UnverifiedCompared with a strong adaptive allocation baseline, this method achieved absolute accuracy improvements of up to 3.73 percentage points (LLaDA) and 7.33 percentage points (Dream) on MATH-500.
50%UnverifiedCompared with a strong adaptive allocation baseline, this method achieved absolute coverage improvements of up to 11.39 percentage points (LLaDA) and 10.98 percentage points (Dream) on HumanEval.
50%