Update README.md

README.md

@@ -21,32 +21,31 @@ library_name: transformers
 
 ## Description
 
-**Qwen-3-Nemotron-32B-Reward** is a mid-sized reward model built on **Qwen3-32B** and fine-tuned using Bradley–Terry pairwise comparisons collected in HelpSteer3.
-To **preserve Qwen-3’s native reasoning**, training employed a `"nothink"` instruction that separates the reasoning trace from the reward signal.
+Qwen-3-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.
 
-This model is optimized for commercial deployments requiring a balance between parameter count, inference latency, and scoring fidelity.
+This model is ready for commercial/non-commercial use.
 
+## License/Terms of Use
 
-**Purpose:**  
-A specialized reward model that assigns a numerical “reward” score to evaluate the quality of LLM-generated responses.
+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/).
 
-**Architecture & Training:**  
-- **Base:** Built on **Qwen3-32B**.  
-- **Training Data:** Human-annotated comparisons from HelpSteer3.  
-- **Framework:** Bradley–Terry pairwise methodology with `"nothink"` instruction.
+### Deployment Geography
 
-**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.
-- **Model Class:** `AutoModelForSequenceClassification`
+Global
 
-> **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.
+## Use Case
 
-## License/Terms of Use
+Qwen-3-Nemotron-32B-Reward assigns a reward score to an LLM-generated response in a user–assistant dialogue.
 
-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/).
+## Release Date:
+
+HuggingFace 06/27/2025 via https://huggingface.co/nvidia/Qwen-3-Nemotron-32B-Reward
+
+## References
+
+* [HelpSteer3-Preference](https://arxiv.org/abs/2505.11475)
+* [Qwen3 Model Card](https://huggingface.co/Qwen/Qwen3-32B)  
 
----
 
 ## RM-Bench LeaderBoard
 
@@ -70,26 +69,43 @@ As of 29 May 2025, Qwen-3-Nemotron-32B-Reward has comparable scores on [JudgeBen
 | [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               |
 | [Llama-3.3-Nemotron-70B-Reward-Multilingual](https://huggingface.co/nvidia/Llama-3.3-Nemotron-70B-Reward-Multilingual)  |66.2 | 71.4 | 82.1 |59.5 | 69.4|
 
----
 
-## Use Case
 
-Qwen-3-Nemotron-32B-Reward assigns a reward score to an LLM-generated response in a user–assistant dialogue.
 
----
 
-## References
-
-* [HelpSteer3-Preference](https://arxiv.org/abs/2505.11475)
-* [Qwen3 Model Card](https://huggingface.co/Qwen/Qwen3-32B)  
-
----
 
 ## Model Architecture
 
 **Architecture Type:** Transformer  
-**Network:** Qwen3, 32B parameters  
-**Training:** Bradley–Terry preference training
+**Network Architecture:** Qwen3
+
+We developed this model using [Qwen-3-32B](https://huggingface.co/Qwen/Qwen3-32B) as its foundation. This model contains 32 billion parameters.
+
+## 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>
+
+## 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>
+
+## 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>
+
 
 ## Quick Start
 
@@ -127,17 +143,74 @@ for response in [good_response, bad_response]:
 # reward for bad_response = -7.9765625
 ```
 
+## Model Version: 
+v1.0
+
+# Training, Testing and Evaluation Datasets: 
+
 ## Training Datasets:
 
-**Dataset Name:**  
-HelpSteer-3
+**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>
+* 38,459 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>
 
-**Dataset Links:**  
-https://huggingface.co/datasets/nvidia/HelpSteer2  
-https://huggingface.co/datasets/nvidia/HelpSteer3
+**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.
+
+## 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