YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/model-cards#model-card-metadata)

Model Card for SWE-LLM

Disclaimer: This model has not been trained with regard to safety and bias, and might therefore emit behaviour that is unfavorable. Note that this is a very early version of the model and that future versions of SWE-LLM will include wider and more extensive training data to make safer, more knowledgeable and more capable.

Model Details

Model Description

SWE-LLM is a fine-tuned version of Llama 3 70B, trained on a carefully curated dataset consisting of over 12,700 Swedish texts, corresponding to over 13.3 million tokens. The model has been trained on A100 GPUs and is now available for the public to test for free at https://swe-llm.se/.

  • Developed by: VISS.AI
  • Model type: Language Model
  • Language(s) (NLP): Swedish

Uses

Direct Use

SWE-LLM can be used directly for generating high-quality Swedish text, including articles, technical reports, and various types of content. The model can also be utilized for conversational purposes in chat applications.

Downstream Use

SWE-LLM can be fine-tuned for specific tasks such as sentiment analysis, customer service enhancement, and efficient translation of content into Swedish.

Out-of-Scope Use

The model should not be used for generating harmful or biased content, and care should be taken to avoid misuse in sensitive applications.

Bias, Risks, and Limitations

While SWE-LLM is trained on a diverse set of Swedish texts, it may still exhibit biases present in the training data. Users should be cautious and critical of the outputs, especially in sensitive or high-stakes contexts.

Recommendations

Users should be aware of the potential biases and limitations of the model. It is recommended to continuously monitor and evaluate the outputs to ensure they meet the desired quality and ethical standards.

How to Get Started with the Model

Use the code below to get started with the model.

[More Information Needed]

Training Details

Training Data

The model was trained on a dataset of over 12,700 Swedish texts, corresponding to roughly 13.3 million tokens. The dataset includes a variety of text types, ensuring a broad competence in Swedish language generation.

Environmental Impact

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

  • Hardware Type: A100 GPUs

Model Card Contact

For more information, please visit our website https://viss.ai/ or contact us at hello@viss.ai.

Downloads last month
22
Safetensors
Model size
70.6B params
Tensor type
BF16
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for viss-ai/LLama3-70B-SWE-LLM

Quantizations
1 model