[KongchangAI]
ConceptDiffusion Verifier

DiVR

An independent verifier for diffusion language models that guides regeneration decisions by predicting the final correctness of a partially generated sequence. It is plug-and-play.

Timeline (last 90 days)

Nov 13

DiVR (Diffusion Verifier) is a standalone verifier used to predict whether a partially generated sequence will ultimately reach the correct answer.

Unverified50%
Nov 13

DiVR's training data comes from the final correctness of the model's own trajectories, that is, letting the model generate a large number of trajectories and recording whether each trajectory ultimately succeeds, then using these results to train the verifier.

Unverified50%
Nov 13

At inference time, DiVR decides when to regenerate a block and which candidate to keep.

Unverified50%
Nov 13

As a plug-and-play external module, DiVR does not modify the generator itself, so it is compatible with different generators without retraining the generative model.

Unverified50%
Nov 13

The experiments use LLaDA-8B-Base on five reasoning benchmarks, and DiVR raises the average accuracy of the fine-tuned generator from 35.3% to 43.4%.

Unverified50%
Nov 13

DiVR costs 1.9-3.5 times that of greedy decoding.

Unverified50%
Nov 13

Search-based baseline methods cost 3.8-7.4 times as much but still achieve lower accuracy than DiVR.

Unverified50%
Nov 13

DiVR is complementary to post-training methods such as supervised fine-tuning (SFT) and reinforcement learning (RL).

Unverified50%
Nov 13

DiVR can generalize to other generators, including Dream.

Unverified50%
Nov 13

The core insight of DiVR is that error correction for diffusion language models should not rely solely on token-level confidence, but should be verified starting from sequence-level correctness.

Unverified50%

All Facts (10)

Source Articles