Nemotron-3-8B-Chat-4k-SteerLM

Model Overview

License

The use of this model is governed by the NVIDIA AI Foundation Models Community License Agreement.

Description

Nemotron-3-8B-SteerLM is an 8 billion parameter generative language model instruct-tuned on an 8B base model. It takes input with context length up to 4,096 tokens. The model has been customized using the SteerLM method developed by NVIDIA to allow for user control of model outputs during inference. ​

Key capabilities enabled by SteerLM: ​

• Dynamic steering of responses by specifying desired attributes like quality, helpfulness, and toxicity at inference time.
• Simplified training compared to RLHF techniques like fine-tuning and bootstrapping.

​ Nemotron-3-8B-SteerLM is part of Nemotron-3, which is a family of enterprise ready GPT-based decoder-only generative text models compatible with NVIDIA NeMo Framework. For other models in this collection, see the models page. ​

NVIDIA NeMo is an end-to-end, cloud-native framework to build, customize, and deploy generative AI models anywhere. It includes training and inferencing frameworks, guardrailing toolkits, data curation tools, and pretrained models, offering enterprises an easy, cost-effective, and fast way to adopt generative AI. ​

References

Announcement Blog

Model Architecture

Architecture Type: Transformer

​ Network Architecture: Generative Pre-Trained Transformer (GPT-3)

​ The SteerLM method involves the following key steps: ​

1. Train an attribute prediction model on human annotated data to evaluate response quality.
2. Use this model to annotate diverse datasets and enrich training data.
3. Perform conditioned fine-tuning to align responses with specified combinations of attributes.
4. (Optionally) Bootstrap training through model sampling and further fine-tuning.

​ SteerLM-8B applies this technique on top of the open-source NVIDIA GPT model architecture. It was pretrained on internet-scale data and then customized using OASST, HH-RLHF, Light, a subset of permissive licensed OpenPlatypus, and some internally collected SFT data.

Prompt Format

Single Turn

<extra_id_0>System
A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.
​
<extra_id_1>User
{prompt 1}
<extra_id_1>Assistant
<extra_id_2>quality:4,understanding:4,correctness:4,coherence:4,complexity:4,verbosity:4,toxicity:0,humor:0,creativity:0,violence:0,helpfulness:4,not_appropriate:0,hate_speech:0,sexual_content:0,fails_task:0,political_content:0,moral_judgement:0,lang:en

Multi-Turn or Few-shot

<extra_id_0>System
A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.
​
<extra_id_1>User
{prompt 1}
<extra_id_1>Assistant
<extra_id_2>quality:4,understanding:4,correctness:4,coherence:4,complexity:4,verbosity:4,toxicity:0,humor:0,creativity:0,violence:0,helpfulness:4,not_appropriate:0,hate_speech:0,sexual_content:0,fails_task:0,political_content:0,moral_judgement:0,lang:en
{response 1}
<extra_id_1>User
{prompt 2}
<extra_id_1>Assistant
<extra_id_2>quality:4,understanding:4,correctness:4,coherence:4,complexity:4,verbosity:4,toxicity:0,humor:0,creativity:0,violence:0,helpfulness:4,not_appropriate:0,hate_speech:0,sexual_content:0,fails_task:0,political_content:0,moral_judgement:0,lang:en

Example prompt formation code

PROMPT_TEMPLATE = """<extra_id_0>System
A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.
​
<extra_id_1>User
{prompt}
<extra_id_1>Assistant
<extra_id_2>quality:4,understanding:4,correctness:4,coherence:4,complexity:4,verbosity:4,toxicity:0,humor:0,creativity:0,violence:0,helpfulness:4,not_appropriate:0,hate_speech:0,sexual_content:0,fails_task:0,political_content:0,moral_judgement:0,lang:en"""
​
question = "Write a poem on NVIDIA in the style of Shakespeare"
prompt = PROMPT_TEMPLATE.format(prompt=question)
print(prompt)

​ Each of the properties (e.g. humor, toxicity…) can receive integer values in the range [0,4]. ​ ​

Software Integration

Runtime Engine(s): NVIDIA AI Enterprise ​

Toolkit: NeMo Framework ​

See NeMo inference container documentation for details on how to setup and deploy an inference server. ​

Sample Inference Code:

from nemo.deploy import NemoQuery

# In this case, we run inference on the same machine
nq = NemoQuery(url="localhost:8000", model_name="Nemotron-3-8B-Chat-4K-RLHF")

# See above for prompt format
output = nq.query_llm(prompts=[prompt], max_output_token=200, top_k=1, top_p=0.0, temperature=0.1)

# NOTE: Chat models require post-processing the output since the `NemoQuery` API
# does not support stopping generation on the special <extra_id_1> token.
output = [[s.split("<extra_id_1>", 1)[0].strip() for s in out] for out in output]

print(output)

Supported Hardware: ​

• H100
• A100 80GB, A100 40GB ​

Model Version(s)

​ Nemotron-3-8B-chat-4k-steerlm-BF16-1 ​

Dataset

NVIDIA models are trained on a diverse set of public and proprietary datasets. NVIDIA is committed to the responsible development of large language models and conducts reviews of all datasets included in training.

Evaluation

​ MT Bench Score

Intended use

​ The 8B-Chat-SteerLM model is for users who want to customize a model’s response during inference. ​

Ethical use

​ Technology can have a profound impact on people and the world, and NVIDIA is committed to enabling trust and transparency in AI development. NVIDIA encourages users to adopt principles of AI ethics and trustworthiness to guide your business decisions by following the guidelines in the NVIDIA AI Foundation Models Community License Agreement. ​

Limitations

• The model was trained on data that contains toxic language and societal biases originally crawled from the internet. Therefore, the model may amplify those biases and return toxic responses especially when prompted with toxic prompts.
• The model may generate answers that may be inaccurate, omit key information, or include irrelevant or redundant text producing socially unacceptable or undesirable text, even if the prompt itself does not include anything explicitly offensive.