Instructions to use Malkani6/Limbus_V1-GGUFs 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 Malkani6/Limbus_V1-GGUFs 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 Malkani6/Limbus_V1-GGUFs:Q4_K_M # Run inference directly in the terminal: llama cli -hf Malkani6/Limbus_V1-GGUFs:Q4_K_M
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf Malkani6/Limbus_V1-GGUFs:Q4_K_M # Run inference directly in the terminal: llama cli -hf Malkani6/Limbus_V1-GGUFs: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 Malkani6/Limbus_V1-GGUFs:Q4_K_M # Run inference directly in the terminal: ./llama-cli -hf Malkani6/Limbus_V1-GGUFs: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 Malkani6/Limbus_V1-GGUFs:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama-cli -hf Malkani6/Limbus_V1-GGUFs:Q4_K_M
Use Docker
docker model run hf.co/Malkani6/Limbus_V1-GGUFs:Q4_K_M
- LM Studio
- Jan
- Ollama
How to use Malkani6/Limbus_V1-GGUFs with Ollama:
ollama run hf.co/Malkani6/Limbus_V1-GGUFs:Q4_K_M
- Unsloth Studio
How to use Malkani6/Limbus_V1-GGUFs 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 Malkani6/Limbus_V1-GGUFs 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 Malkani6/Limbus_V1-GGUFs to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for Malkani6/Limbus_V1-GGUFs to start chatting
- Docker Model Runner
How to use Malkani6/Limbus_V1-GGUFs with Docker Model Runner:
docker model run hf.co/Malkani6/Limbus_V1-GGUFs:Q4_K_M
- Lemonade
How to use Malkani6/Limbus_V1-GGUFs with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull Malkani6/Limbus_V1-GGUFs:Q4_K_M
Run and chat with the model
lemonade run user.Limbus_V1-GGUFs-Q4_K_M
List all available models
lemonade list
- Atomic Chat
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
🧠 Limbus: The Honest Brain
"Limbus is the key to the 'honest brain.' It doesn't just know what people say; it knows why they flinch."
🧩 Model Overview
Limbus is a highly specialized large language model (LLM) fine-tuned on dense, academic, and practical texts across Cognitive, Social, and Behavioral Psychology.
It is designed to act as an expert reasoning engine capable of synthesizing complex, sometimes contradictory, human behavioral theories and psychological mechanisms.
Its primary strength is bridging the gap between abstract academic theory and real-world human behavior, making it invaluable for analysis in contexts involving persuasion, deception, emotional regulation, and decision-making.
🧠 Key Specializations
Limbus excels in the following areas, demonstrating deep mechanistic understanding, not just surface-level definitions:
| Domain | Core Capability | Signature Mechanism |
|---|---|---|
| Social-Cognitive Psychology | Analysis of social bias and attribution errors (FAE, Self-Serving Bias). | Modeling the interaction between beliefs, emotions, and environment (e.g., CAPS model: "If [Situation], Then [Behavior]"). |
| Behavioral Economics | Deconstructing decision-making biases in high-stakes scenarios. | Explaining the influence of Framing, Anchoring, Loss Aversion, and default effects (e.g., Nudge Theory). |
| Emotion & Deception | Reading emotional leakage and identifying manipulative intent. | Distinguishing between genuine emotional responses (limbic system) and deliberate cognitive masking (System 2 control). |
| Psychodynamic Theory | Understanding the role of unconscious motivation and developmental crises. | Linking core defense mechanisms (e.g., Repression) to later-life conflicts (e.g., Identity Confusion). |
🎯 Intended Use
Limbus is primarily intended for research, application development, and educational purposes where nuanced understanding of human decision-making is critical.
- Behavioral Modeling: Generating realistic simulations of human responses under psychological stress or persuasive influence.
- Consulting & Marketing: Analyzing the efficacy of persuasive strategies (e.g., Cialdini's principles) and predicting susceptibility to cognitive biases (e.g., Availability Heuristic).
- Education: Serving as an advanced tutor for graduate and undergraduate students in Psychology, Cognitive Science, and Behavioral Economics.
⚖️ Limitations and Ethical Considerations
Not a Diagnostic Tool:
Limbus is a theoretical model and must never be used for clinical diagnosis, therapeutic recommendations, or medical advice.Bias in Data:
The model is trained on human psychological findings, which contain documented academic biases and may reflect cultural norms present in the source material.
It may overemphasize behavioral factors in attribution.Deception Generation:
The model possesses high-fidelity knowledge of deception techniques (leakage, masking, framing).
Users must adhere to ethical guidelines when generating content related to manipulation.
⚙️ Technical Details
| Detail | Value |
|---|---|
| Base Model | Qwen2 (or relevant base model) |
| Architecture | Transformer (Causal LM) with full parameter merge |
| Quantization | Full FP16 (or BF16) precision (ready for GGUF quantization) |
| Training Method | Sequential Low-Rank Adaptation (LoRA) followed by full parameter merge |
| Training Data | Proprietary multi-domain dataset covering Cognitive Psychology, Social Psychology, Decision Theory, Embodied Cognition, and Attribution Theory |
| Merge Technique | Sequential Merge and Unload (successfully preventing Catastrophic Interference) |
“Limbus is the key to the 'honest brain.'
It doesn't just know what people say; it knows why they flinch.”
- Downloads last month
- -
4-bit
5-bit
8-bit
16-bit