Instructions to use QatrixInfotech/Cella-Omin1.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use QatrixInfotech/Cella-Omin1.1 with llama.cpp:
Install (macOS, Linux)
curl -LsSf https://llama.app/install.sh | sh # Start a local OpenAI-compatible server with a web UI: llama serve -hf QatrixInfotech/Cella-Omin1.1:Q4_K_M # Run inference directly in the terminal: llama cli -hf QatrixInfotech/Cella-Omin1.1:Q4_K_M
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf QatrixInfotech/Cella-Omin1.1:Q4_K_M # Run inference directly in the terminal: llama cli -hf QatrixInfotech/Cella-Omin1.1:Q4_K_M
Use pre-built binary
# Download pre-built binary from: # https://github.com/ggerganov/llama.cpp/releases # Start a local OpenAI-compatible server with a web UI: ./llama-server -hf QatrixInfotech/Cella-Omin1.1:Q4_K_M # Run inference directly in the terminal: ./llama-cli -hf QatrixInfotech/Cella-Omin1.1:Q4_K_M
Build from source code
git clone https://github.com/ggerganov/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-server llama-cli # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama-server -hf QatrixInfotech/Cella-Omin1.1:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama-cli -hf QatrixInfotech/Cella-Omin1.1:Q4_K_M
Use Docker
docker model run hf.co/QatrixInfotech/Cella-Omin1.1:Q4_K_M
- LM Studio
- Jan
- Ollama
How to use QatrixInfotech/Cella-Omin1.1 with Ollama:
ollama run hf.co/QatrixInfotech/Cella-Omin1.1:Q4_K_M
- Unsloth Studio
How to use QatrixInfotech/Cella-Omin1.1 with Unsloth Studio:
Install Unsloth Studio (macOS, Linux, WSL)
curl -fsSL https://unsloth.ai/install.sh | sh # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for QatrixInfotech/Cella-Omin1.1 to start chatting
Install Unsloth Studio (Windows)
irm https://unsloth.ai/install.ps1 | iex # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for QatrixInfotech/Cella-Omin1.1 to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for QatrixInfotech/Cella-Omin1.1 to start chatting
- Docker Model Runner
How to use QatrixInfotech/Cella-Omin1.1 with Docker Model Runner:
docker model run hf.co/QatrixInfotech/Cella-Omin1.1:Q4_K_M
- Lemonade
How to use QatrixInfotech/Cella-Omin1.1 with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull QatrixInfotech/Cella-Omin1.1:Q4_K_M
Run and chat with the model
lemonade run user.Cella-Omin1.1-Q4_K_M
List all available models
lemonade list
- Atomic Chat
- Cella Omin 1.1
Cella Omin 1.1
Cella Omin 1.1 is a multilingual AI assistant developed and fine-tuned by Qatrix Infotech.
The model is designed for general conversation, programming assistance, logical reasoning, writing, text summarization, design discussions, and multilingual interaction.
Model Details
| Specification | Details |
|---|---|
| Model Name | Cella Omin 1.1 |
| Developer | Qatrix Infotech |
| Parameters | Approximately 4 billion |
| Format | GGUF |
| Quantization | Q4_K_M |
| Primary File | Cella-Omin-1.1-Q4_K_M.gguf |
| Task | Text generation |
| License | Apache 2.0 |
Core Identity
Cella Omin 1.1 includes a baked-in identity established through supervised fine-tuning.
The identity adapter was merged directly into the model weights before GGUF conversion. A separate system prompt is therefore not normally required for basic identity questions.
When asked about its identity, the model is designed to respond:
I am Cella Omin 1.1, a multilingual AI assistant developed by Qatrix Infotech.
Identity consistency may vary depending on the prompt, chat template, inference software, and generation settings.
Capabilities
Cella Omin 1.1 is designed to assist with:
- General conversation
- Multilingual conversation
- Translation assistance
- Programming assistance
- Basic code debugging
- Writing and rewriting
- Text summarization
- Logical reasoning
- Mathematical reasoning
- UI and UX discussions
- Design-related discussions
- Educational tasks
- Local AI experiments
Supported Languages
The model has been tested for basic identity responses and general interaction in the following languages:
- English
- Malayalam
- Tamil
- Kannada
- Telugu
- Urdu
- Punjabi
- Bengali
- Arabic
Performance may vary depending on the language, dialect, prompt structure, subject, and task complexity.
The model may understand additional languages, but those languages have not been tested to the same extent.
Available Files
Cella-Omin-1.1-Q4_K_M.gguf
The primary quantized model file.
Use this file for:
- Chat
- Writing
- Coding assistance
- Reasoning
- Summarization
- Translation
- Multilingual text generation
Recommended Generation Parameters
| Parameter | Recommended Value |
|---|---|
| Temperature | 1.0 |
| Top K | 64 |
| Top P | 0.95 |
| Context Size | 8192 |
| Repetition Penalty | 1.0 |
These values are starting recommendations. Different applications may require different settings.
Lower temperatures may produce more predictable responses, while higher temperatures may produce more varied responses.
Running with llama.cpp
Install or build a recent version of llama.cpp.
Run the model using:
llama-cli \
--model Cella-Omin-1.1-Q4_K_M.gguf \
--ctx-size 8192 \
--temp 1.0 \
--top-k 64 \
--top-p 0.95 \
--repeat-penalty 1.0 \
--conversation
Depending on your version of llama.cpp, command names and supported options may differ.
Running with Ollama
Create a file named Modelfile in the same directory as the GGUF file.
Add the following content:
FROM ./Cella-Omin-1.1-Q4_K_M.gguf
PARAMETER temperature 1.0
PARAMETER top_k 64
PARAMETER top_p 0.95
PARAMETER num_ctx 8192
PARAMETER repeat_penalty 1.0
Create the Ollama model:
ollama create cella-omin-1.1 -f Modelfile
Run the model:
ollama run cella-omin-1.1
Example Prompts
Identity
Who are you?
Writing
Rewrite this paragraph in a more professional tone.
Programming
Explain this Python error and suggest a fix.
Summarization
Summarize the following text in five bullet points.
Translation
Translate the following English paragraph into Malayalam.
Reasoning
Explain your answer clearly and solve the problem step by step.
Intended Uses
Cella Omin 1.1 is intended for local and offline AI workflows, including:
- Personal AI assistants
- Multilingual chat applications
- Educational demonstrations
- Writing assistance
- Programming assistance
- Translation experiments
- Local text-generation applications
- Research prototypes
- AI interface testing
- Prompt-engineering experiments
Out-of-Scope Uses
The model should not be used as the sole source of information for:
- Medical diagnosis or treatment
- Legal decisions
- Financial decisions
- Emergency response
- Safety-critical systems
- High-risk automated decision-making
- Production code without human review
The model should not be used to facilitate unlawful activity, abuse, deception, privacy violations, harassment, or harm.
Training Summary
Cella Omin 1.1 is an early release in the development of the Cella model family.
| Training Detail | Description |
|---|---|
| Framework | Unsloth |
| Training Method | Supervised Fine-Tuning |
| Adapter Method | LoRA |
| Training Precision | 4-bit |
| Hardware | NVIDIA Tesla T4 GPU |
| Training Focus | Multilingual identity, capability boundaries, and identity-consistency repair |
| Export Format | GGUF |
| Quantization | Q4_K_M |
The trained LoRA adapter was merged into the model weights before conversion to GGUF.
Dataset Summary
The fine-tuning dataset included examples related to:
- Model identity
- Developer attribution
- Multilingual identity responses
- General assistant behavior
- Capability boundaries
- Identity consistency
- Targeted repair prompts
The dataset focused primarily on improving identity consistency and basic assistant behavior rather than creating a large-scale general-knowledge model.
Limitations
Cella Omin 1.1 is an experimental early release.
Hallucinations
The model may generate incorrect, outdated, misleading, biased, or fabricated information.
Important claims should be independently verified.
Limited Knowledge
The model does not have live internet access unless it is connected to an external search, retrieval, or tool system.
Its internal knowledge may be incomplete or outdated.
Language Variance
Quality may vary across languages.
Performance may decrease with:
- Uncommon dialects
- Mixed-language prompts
- Transliteration
- Complex technical content
- Long conversations
- Ambiguous instructions
Prompt Sensitivity
Responses may change depending on:
- The chat template
- The system prompt
- Prompt wording
- Context length
- Inference software
- Sampling settings
Identity behavior may become inconsistent when the model receives aggressive system instructions, conflicting identity instructions, unusual prompts, or incompatible chat templates.
Code Reliability
Generated code may contain:
- Syntax errors
- Security issues
- Incorrect assumptions
- Outdated APIs
- Missing dependencies
- Logical errors
All generated code should be reviewed and tested before use.
Reasoning Reliability
The model may produce confident explanations containing incorrect reasoning.
Step-by-step answers should not automatically be treated as correct.
Context Limitations
Actual usable context length may depend on:
- Available memory
- Inference backend
- Quantization
- Prompt size
- Conversation history
- Runtime configuration
Safety
Users and application developers are responsible for implementing appropriate safety measures.
Recommended safeguards include:
- Human review
- Output filtering
- Input validation
- Rate limiting
- Privacy protection
- Abuse monitoring
- Clear user disclosures
- Independent fact-checking
- Restricted access for high-risk workflows
Applications using this model should clearly inform users that AI-generated responses may contain errors.
Responsible Use
Do not use Cella Omin 1.1 to facilitate:
- Illegal activity
- Violence or physical harm
- Harassment or abuse
- Fraud or deception
- Impersonation
- Privacy violations
- Unauthorized surveillance
- Malware development
- Exploitation
- Discrimination
- Dangerous or safety-critical decisions
Users are responsible for ensuring that their use of the model complies with applicable laws, regulations, licenses, and platform policies.
Disclaimer
Cella Omin 1.1 is provided as an experimental model.
The model is provided without guarantees of accuracy, reliability, fitness for a particular purpose, or suitability for high-risk applications.
Model outputs do not necessarily represent the views of Qatrix Infotech.
Users are responsible for evaluating the model and its outputs before deployment or use.
Developer
Developed by Qatrix Infotech.
License
This model is released under the Apache License 2.0.
Review the repository license and any applicable dependency licenses before redistribution or commercial use.
- Downloads last month
- 103
4-bit
Model tree for QatrixInfotech/Cella-Omin1.1
Base model
QatrixInfotech/Cella-Nexa