----------------------- REVIEW 1 ---------------------
SUBMISSION: 21
TITLE: RecWizard: A Toolkit for Conversational Recommendation with Modular, Portable Models and Interactive User Interface
AUTHORS: Zeyuan Zhang, Tanmay Laud, Zihang He, Xiaojie Chen, Xinshuang Liu, Zhouhang Xie, Julian McAuley and Zhankui He

----------- Relevance -----------
SCORE: 4 (good)
----------- Significance and originality -----------
SCORE: 4 (good)
----------- Paper evaluation -----------
SCORE: 4 (good)
----- TEXT:
The authors propose RecWizward, a modular toolkit for conversational recommender systems.

Strengths:
- RecWizard builds on HuggingFace, which has become the de-facto standard for distributing models, and transformers is the go-to library for language models. These design choices should ensure easy use and integration in ongoing projects.
- Modularity by design is another benefit, allowing flexible research and changes to the CRS
- Debug Mode at a low level enables investigation of failure modes, limitations, and overall debugging of the pipeline.

Weaknesses:
-  An empirical evaluation of the system and different module combinations would be helpful but are not strictly necessary for a demo
----------- Video evaluation -----------
SCORE: 4 (good)
----- TEXT:
Overall, the video is well made. It gives a nice overview of the system and its implementation and demonstrates some exemplary use cases.
The authors could consider adding one more complex usage example that showcases some of the more intricate details of the model.
Additionally, the earlier slides are of noticeably lower resolution, which may be distracting for some viewers.
----------- Overall evaluation -----------
SCORE: 2 (accept)
----- TEXT:
RecWizard is a nice solution for a relevant application. The HF integration should make it easily usable for researchers and other stakeholders.

An interactive demo or Jupyter Notebook of the pipeline beyond the provided codebase might be beneficial for demonstration purposes.



----------------------- REVIEW 2 ---------------------
SUBMISSION: 21
TITLE: RecWizard: A Toolkit for Conversational Recommendation with Modular, Portable Models and Interactive User Interface
AUTHORS: Zeyuan Zhang, Tanmay Laud, Zihang He, Xiaojie Chen, Xinshuang Liu, Zhouhang Xie, Julian McAuley and Zhankui He

----------- Relevance -----------
SCORE: 4 (good)
----------- Significance and originality -----------
SCORE: 4 (good)
----------- Paper evaluation -----------
SCORE: 5 (excellent)
----- TEXT:
This is the best structured demo paper I have read across all demo papers I have reviewed, providing all relevant information in these two pages, and also the supplementary material shows a thorough comparison to existing approaches.
----------- Video evaluation -----------
SCORE: 4 (good)
----- TEXT:
Also the video is well structured and explained. There was a minor point that made me wonder, which was the use of <\entity \> in the text prompts.
----------- Overall evaluation -----------
SCORE: 3 (strong accept)
----- TEXT:
This demo seems to be a good contribution to the community, the paper is well-written and the video demos all functionalities sufficiently. Also, the code is provided on GitHub.