GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
sarath-om
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
sarath-om
Model Tree
sarath-om/sarath_om-1.0
Input Modalities
Output Modalities
Supported Functionality
Sarath OM 1.1 is a model project within the Sarath OM AI initiative, focused on developing useful, accessible, and extensible AI-powered experiences.
This model card documents the project's identity, intended use, limitations, and technical information. Details such as the underlying architecture, model weights, training datasets, and evaluation results should be confirmed from the repository files before being reported as facts.
| Field | Details |
|---|---|
| Model name | Sarath OM 1.1 |
| Repository | sarath-om/sarath_om-1.1 |
| Project | Sarath OM |
| Version | 1.1 |
| Developer and owner | Rishabh Srivastava |
| License | MIT declared in repository metadata; verify applicable rights and files |
| Architecture | Qwen2 decoder-only Transformer (Qwen2ForCausalLM), 28 layers, hidden size 1536 |
| Parameter count | 1.54B (F16) |
| Training datasets | Not yet documented |
| Benchmark results | Not yet documented |
Depending on the verified implementation, potential use cases to evaluate may include:
These are potential use cases, not claims that the current repository implements every capability listed.
The following information should be completed once verified from the repository:
Qwen2ForCausalLM), 28 layers, hidden size 1536Qwen/Qwen2.5-1.5B-Instruct (verified from adapter_config.json); fine-tuned with LoRA and merged into full weights.model.safetensors (F16, safetensors)max_position_embeddings in config.json)AutoModelForCausalLM).If Sarath OM provides an application or interface that calls an external model rather than distributing its own trained weights, describe it accurately as an application or integration instead of claiming it is a standalone trained model.
See the repository files for the implementation-specific setup and inference instructions:
Open Sarath OM 1.1 on Hugging Face
Before publishing installation commands or code examples, verify the available files, dependencies, model format, and supported inference interface. A generic Transformers example may not work for a model that uses another framework or runtime.
Training details have not been established in this model card.
When available, document:
Do not imply that a model was trained from scratch or fine-tuned on a particular dataset unless the repository or accompanying documentation supports that statement.
No verified benchmark results are included in this card.
Before publishing evaluation claims, document the datasets, metrics, testing methodology, software environment, and comparison baselines. Include both successful and unsuccessful results where relevant, and distinguish measured performance from expected performance.
Users should evaluate the model for their intended application, protect sensitive information, respect applicable law and third-party rights, and establish suitable human oversight where needed.
Do not submit confidential or personal data unless the deployment's data-handling and retention practices have been reviewed.
The repository metadata declares the MIT License. Before redistribution or commercial use, verify that the repository contains the intended license file and that the developer has the rights to distribute all included weights, code, datasets, and other assets under the stated terms. Third-party components may have separate license requirements.
Current documented version: Sarath OM 1.1
For future releases, document changes to the implementation, dependencies, supported capabilities, evaluation results, and known limitations.
Rishabh Srivastava
Developer and owner of the Sarath OM project.
If you reference this project, you may use the following provisional citation:
bibtex
@misc{sarath_om_1_1, title = {Sarath OM 1.1}, author = {Rishabh Srivastava}, year = {2026}, howpublished = {\url{https://huggingface.co/sarath-om/sarath_om-1.1}}, note = {Sarath OM AI project}}Sarath OM is an AI initiative focused on useful technology, practical innovation, and continued development.
Intelligence. Innovation. Possibility.
@misc{sarath_om_1_1, title = {Sarath OM 1.1}, author = {Rishabh Srivastava}, year = {2026}, howpublished = {\url{https://huggingface.co/sarath-om/sarath_om-1.1}}, note = {Sarath OM AI project}}