File size: 1,903 Bytes
3c14026 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 | ---
license: mit
base_model: Dream-org/Dream-v0-Instruct-7B
tags:
- dllm-prm-gap
- discrete-diffusion
- gsm8k
- process-reward-model
- dream-7b
pipeline_tag: text-classification
library_name: peft
---
<h1 style="text-align: left; font-size: 1.6em; margin-bottom: 0.75em;">
<span style="color:#E69F00; font-weight:bold;">dLLM</span><span style="color:#0072B2; font-weight:bold;"> PRM</span><span style="color:#009E73; font-weight:bold;"> Gap</span> · Bidirectional PRM
</h1>
<div style="text-align: left; margin-bottom: 18px;">
<a href="https://github.com/dLLM-PRM-Gap/dLLM-PRM-Gap" target="_blank">💻 Code</a>
•
<a href="https://huggingface.co/collections/YanZhanPKU/dllm-prm-gap" target="_blank">🤗 Collection</a>
•
<a href="https://arxiv.org/abs/2609.35472" target="_blank">📄 Paper</a>
</div>
## News
- **2026-09**: Accepted at **NeurIPS 2026**.
Public adapter release for **dLLM PRM Gap**. bidirectional attention with mask-aware mean pooling over intermediate denoising states
> This repository contains only trainable adapter parameters and the reward head. It does not include the base model. The release configuration is recorded in `config.json`; the arXiv paper defines the release scope and citation.
## Files
- `adapter.safetensors` — compact adapter weights
- `config.json` — public base-model id, architecture, and provenance
## Load
```python
from huggingface_hub import snapshot_download
from prm.checkpointing import load_diffusion_prm
path = snapshot_download("YanZhanPKU/dLLM-PRM-Gap-bidir-dream7b")
model, report = load_diffusion_prm(
checkpoint=path,
model_path="Dream-org/Dream-v0-Instruct-7B",
local_files_only=False,
)
```
**Role:** process-reward model used by PRM Guided, Hybrid, and the snapshot diagnostics.
Paper: [https://arxiv.org/abs/2609.35472](https://arxiv.org/abs/2609.35472).
|