Terms of Service
Last updated: March 2026
BeaverYard provides AI/LLM fine-tuning as a service. By creating an account, submitting a job, uploading data, purchasing a training run, downloading artifacts, or otherwise using BeaverYard, you agree to these Terms of Service.
1. User responsibilities
- You represent and warrant that you have all necessary rights, licenses, permissions, and legally required consents to upload, use, process, and fine-tune on any data you submit to BeaverYard.
- You must not upload or use unlawful, infringing, harmful, deceptive, abusive, privacy-violating, or otherwise prohibited content.
- You are solely responsible for your datasets, prompts, configurations, outputs, downstream uses, and compliance with applicable laws and regulations.
2. Acceptable use and third-party model terms
- Certain BeaverYard services may use third-party models, including Gemma 1.1 7B Instruct.
- Your use of any BeaverYard service powered by such models is subject to these Terms and to the applicable third-party model terms and use restrictions referenced by BeaverYard.
- For services using Gemma 1.1 7B Instruct, you must not use BeaverYard or any resulting output or artifact in any manner prohibited by the Gemma Terms of Use or the Gemma Prohibited Use Policy.
- You must not use BeaverYard to create, support, or distribute unlawful content, infringing content, scams, phishing, spam, malware, deceptive impersonation, harassment, hateful content, violent extremist content, or other prohibited material.
- BeaverYard may suspend, reject, remove, disable, or terminate any dataset, job, artifact, or account that it reasonably believes violates these Terms, third-party model restrictions, or applicable law.
3. Payment terms
- Jobs are billed as one-time purchases based on the automatically assigned plan and configuration.
- Payments are non-refundable once training or processing has begun.
4. Data handling and retention
- Uploaded datasets are stored in BeaverYard-managed infrastructure for processing.
- Training artifacts and related outputs are stored for customer download after completion.
- Unless otherwise stated at purchase, retention is 7 days across all plans.
- After the applicable retention period, BeaverYard may delete datasets, artifacts, and related files from active systems.
5. Outputs and service limitations
- Training results, model behavior, and generated outputs depend on dataset quality, formatting, configuration, and other variables. Results are not guaranteed.
- Outputs may be inaccurate, incomplete, biased, or unsuitable for a particular purpose. You are responsible for reviewing and validating outputs before relying on them.
- Service availability is provided on a best-effort basis. Unless expressly stated otherwise, BeaverYard provides no service level agreement.
6. Downloaded artifacts
- Where BeaverYard makes downloadable model artifacts, adapters, checkpoints, or derivatives available, those artifacts may remain subject to applicable third-party model terms and restrictions.
- You are responsible for complying with any terms, notices, or downstream restrictions that apply to downloaded artifacts or their further use, distribution, hosting, or commercialization.
7. Liability disclaimer
- To the maximum extent permitted by law, BeaverYard disclaims all implied warranties.
- BeaverYard is not liable for indirect, incidental, special, consequential, or lost-profit damages arising from use of the service.
Contact
Questions about these terms? Contact us.