Update README.md

README.md

@@ -21,32 +21,33 @@ library_name: transformers
 # Model Overview
 
 ## Description
+Qwen-2.5-Nemotron-32B-Reward is a reward model that assigns a numerical “reward” score to evaluate the quality of LLM-generated responses.  A higher reward on one conversation indicates better performance within that context, but does *not* translate across unrelated prompts.
 
-**Qwen-2.5-Nemotron-32B-Reward** is a mid-sized reward model built on **Qwen2.5-32B-Instruct** and fine-tuned using Bradley–Terry pairwise comparisons from the HelpSteer2 and HelpSteer3 datasets.
+This model is ready for commercial/non-commercial use.
 
-This model balances parameter count, inference latency, and scoring fidelity, making it suitable for production deployments.  
+## License/Terms of Use
+
+Use of this model is governed by the [NVIDIA Open Model License](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/).
 
+### Deployment Geography
 
-**Purpose:**  
-A specialized reward model that assigns a numerical “reward” score to assess the quality of LLM-generated responses.
+Global
 
-**Architecture & Training:**  
-- **Base:** Built on top of the Qwen2.5-32B-Instruct checkpoint.  
-- **Reward Framework:** Bradley–Terry pairwise comparison methodology.
-- **Training Data:** Human-annotated comparisons from HelpSteer2 & HelpSteer3.
+## Use Case
 
-**How It Works:**  
-- **Input:** An English dialogue (user ↔ assistant) of up to 8,192 tokens.  
-- **Output:** A single “reward” value assessing the last assistant response.
-- **Usage:** Use `AutoModelForSequenceClassification` for scoring.
+Qwen-2.5-Nemotron-32B-Reward labels an LLM-generated response to a user query with a reward score.
 
-> **Note:** Scores are only directly comparable for different answers to the *same* prompt. A higher reward on one conversation indicates better performance within that context, but does *not* translate across unrelated prompts.
+## Release Date:
 
-## License/Terms of Use
+HuggingFace 06/27/2025 via https://huggingface.co/nvidia/Qwen-2.5-Nemotron-32B-Reward
 
-Use of this model is governed by the [NVIDIA Open Model License](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/).
 
----
+## References
+
+* [HelpSteer3-Preference](https://arxiv.org/abs/2505.11475)
+* [HelpSteer2-Preference](https://arxiv.org/abs/2410.01257)
+* [Qwen2.5 Model Card](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct)  
+
 
 ## RM-Bench LeaderBoard
 
@@ -66,27 +67,39 @@ As of 29 May 2025, Qwen-2.5-Nemotron-32B-Reward is slightly lower on [RM-Bench](
 | **Qwen-2.5-Nemotron-32B-Reward**       | 61.7   | 74.5    | 76.2  |  82.1 | 70.3               |
 | [Llama-3.3-Nemotron-70B-Reward](https://huggingface.co/nvidia/Llama-3.3-Nemotron-70B-Reward)             | 70.8   | 76.5    | 82.1  | 66.7  | 73.7               |
 
----
 
-## Use Case
+## Model Architecture
 
-Qwen-2.5-Nemotron-32B-Reward labels an LLM-generated response to a user query with a reward score.
+**Architecture Type:** Transformer  
+**Network Architecture:** Qwen2.5
 
----
+We developed this model using [Qwen-2.5-32B-Instruct](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct) as its foundation. This model contains 32 billion parameters.
 
-## References
+## Input:
+**Input Type(s):** Text <br>
+**Input Format:** String <br>
+**Input Parameters:** One Dimensional (1D) <br>
+**Other Properties Related to Input:** Max of 128k tokens (but trained only on conversations up to 8K tokens) <br>
 
-* [HelpSteer3-Preference](https://arxiv.org/abs/2505.11475)
-* [HelpSteer2-Preference](https://arxiv.org/abs/2410.01257)
-* [Qwen2.5 Model Card](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct)  
+## Output:
+**Output Type(s):** Float <br>
+**Output Format:** One Single Float <br>
+**Output Parameters:** One-Dimensional (1D) <br>
+**Other Properties Related to Output:**  The float value represents the quality of the response, with a higher value representing higher quality. <br>
 
----
+Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA’s hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions. <br>
 
-## Model Architecture
+## Software Integration:
+**Runtime Engine(s):** <br>
+* [NeMo - 24.05.llama.3.1] <br>
+
+**Supported Hardware Microarchitecture Compatibility:** <br>
+* NVIDIA Ampere <br>
+* NVIDIA Hopper <br>
+* NVIDIA Turing <br>
+
+**Supported Operating System(s):** Linux <br>
 
-**Architecture Type:** Transformer  
-**Network:** Qwen2.5, 32B parameters  
-**Training:** Bradley–Terry preference training
 
 ## Quick Start
 
@@ -120,17 +133,100 @@ for response in [good_response, bad_response]:
 # reward for bad_response = -7.53515625
 ```
 
+## Model Version: 
+v1.0
+
+# Training, Testing and Evaluation Datasets: 
+
 ## Training Datasets:
 
-**Dataset Name:**  
-HelpSteer-2/-3
+**Dataset Name:** HelpSteer3 <br>
+**Dataset Link:** https://huggingface.co/datasets/nvidia/HelpSteer3
 
-**Dataset Links:**  
-https://huggingface.co/datasets/nvidia/HelpSteer2  
-https://huggingface.co/datasets/nvidia/HelpSteer3
+**Data Collection Method by dataset** <br>
+* [Hybrid: Human, Synthetic] <br>
+
+**Labeling Method by dataset** <br>
+* [Human] <br>
+
+**Properties:** <br>
+* 38,459 prompts, each with a pair of responses as well as human preferences between the pair of responses.
+
+**Dataset Name:** HelpSteer2 <br>
+**Dataset Link:** https://huggingface.co/datasets/nvidia/HelpSteer2
+
+**Data Collection Method by dataset** <br>
+* [Hybrid: Human, Synthetic] <br>
+
+**Labeling Method by dataset** <br>
+* [Human] <br>
+
+**Properties:** <br>
+* 6,766 prompts, each with a pair of responses as well as human preferences between the pair of responses.
+
+
+## Testing Datasets:
+
+**Dataset Name:** HelpSteer3 <br>
+**Dataset Link:** https://huggingface.co/datasets/nvidia/HelpSteer3
+
+**Data Collection Method by dataset** <br>
+* [Hybrid: Human, Synthetic] <br>
+
+**Labeling Method by dataset** <br>
+* [Human] <br>
+
+**Properties:** <br>
+* 2,017 prompts, each with a pair of responses as well as human preferences between the pair of responses.
+
+
+**Dataset Name:** HelpSteer2 <br>
+**Dataset Link:** https://huggingface.co/datasets/nvidia/HelpSteer2
+
+**Data Collection Method by dataset** <br>
+* [Hybrid: Human, Synthetic] <br>
+
+**Labeling Method by dataset** <br>
+* [Human] <br>
+
+**Properties:** <br>
+* 352 prompts, each with a pair of responses as well as human preferences between the pair of responses.
+
+## Evaluation Datasets
+
+**Dataset Name:** RM-Bench <br>
+**Dataset Link:** https://huggingface.co/datasets/THU-KEG/RM-Bench
+
+**Data Collection Method by dataset** <br>
+* [Hybrid: Human, Synthetic] <br>
+
+**Labeling Method by dataset** <br>
+* [Hybrid: Human, Synthetic] <br>
+
+**Properties:** <br>
+* 1,327 prompts, each with three pairs of responses as well as preferences between the pair of responses.
+
+
+**Dataset Name:** JudgeBench <br>
+**Dataset Link:** https://huggingface.co/datasets/ScalerLab/JudgeBench
+
+**Data Collection Method by dataset** <br>
+* [Hybrid: Human, Synthetic] <br>
+
+**Labeling Method by dataset** <br>
+* [Hybrid: Human, Synthetic] <br>
+
+**Properties:** <br>
+* 350 prompts, each with a pair of responses as well as preferences between the pair of responses.
+
+# Inference:
+**Engine:** PyTorch <br>
+**Test Hardware:** H100, A100 80GB, A100 40GB <br>
 
 ## Ethical Considerations:
 NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications.  When downloaded or used in accordance with our terms of service, developers should work with their supporting model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse. 
+For more detailed information on ethical considerations for this model, please see the Model Card++ [Explainability](explainability.md), [Bias](bias.md), [Safety & Security](safety.md), and [Privacy](privacy.md) Subcards.  
+
 Please report security vulnerabilities or NVIDIA AI Concerns [here](https://www.nvidia.com/en-us/support/submit-security-vulnerability/).
 
 ## Citation
@@ -148,12 +244,13 @@ If you find this model useful, please cite the following works:
       url={https://arxiv.org/abs/2505.11475}, 
 }
 
-@misc{wang2024helpsteer2,
-      title={Help{S}teer2: Open-source dataset for training top-performing reward models}, 
-      author={Zhilin Wang and Yi Dong and Olivier Delalleau and Jiaqi Zeng and Gerald Shen and Daniel Egert and Jimmy J. Zhang and Makesh Narsimhan Sreedhar and Oleksii Kuchaiev},
-      year={2024},
-      eprint={2406.08673},
+@misc{wang2025helpsteer2preferencecomplementingratingspreferences,
+      title={HelpSteer2-Preference: Complementing Ratings with Preferences}, 
+      author={Zhilin Wang and Alexander Bukharin and Olivier Delalleau and Daniel Egert and Gerald Shen and Jiaqi Zeng and Oleksii Kuchaiev and Yi Dong},
+      year={2025},
+      eprint={2410.01257},
       archivePrefix={arXiv},
-      primaryClass={id='cs.CL' full_name='Computation and Language' is_active=True alt_name='cmp-lg' in_archive='cs' is_general=False description='Covers natural language processing. Roughly includes material in ACM Subject Class I.2.7. Note that work on artificial languages (programming languages, logics, formal systems) that does not explicitly address natural-language issues broadly construed (natural-language processing, computational linguistics, speech, text retrieval, etc.) is not appropriate for this area.'}
+      primaryClass={cs.LG},
+      url={https://arxiv.org/abs/2410.01257}, 
 }
 ```