Text Generation
PEFT
Safetensors
GGUF
English
12b
12B
gemma
gemma-4
pvl
cranberry-12b
lora
c
cpp
solidity
python
systems-engineering
memory-safety
formal-verification
post-quantum
conversational
Instructions to use chappedjord/cranberry-12b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use chappedjord/cranberry-12b with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("google/gemma-4-12b-it") model = PeftModel.from_pretrained(base_model, "chappedjord/cranberry-12b") - Notebooks
- Google Colab
- Kaggle
Download Modelfile from chappedjord/cranberry-12b: direct link, hf CLI and curl.
- Browser
- Download file 1.77 kB
-
https://huggingface.co/chappedjord/cranberry-12b/resolve/main/Modelfile
- Command line
-
hf download hf://chappedjord/cranberry-12b/Modelfile
-
curl -L -o Modelfile https://huggingface.co/chappedjord/cranberry-12b/resolve/main/Modelfile
1.77 kB
| # ===================================================================== | |
| # Ollama Modelfile for Cranberry 12B (PVL Systems & Invariant LoRA) | |
| # ===================================================================== | |
| # Base: Google Gemma-4 12B Instruction-Tuned | |
| # License: Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0) | |
| # ===================================================================== | |
| FROM gemma-4-12b:it | |
| # Path to the Cranberry 12B LoRA GGUF adapter | |
| ADAPTER ./cranberry_12b_adapter.gguf | |
| # Gemma turn template | |
| TEMPLATE """{{ if .System }}<start_of_turn>system | |
| {{ .System }}<end_of_turn> | |
| {{ end }}{{ if .Prompt }}<start_of_turn>user | |
| {{ .Prompt }}<end_of_turn> | |
| {{ end }}<start_of_turn>model | |
| {{ .Response }}<end_of_turn>""" | |
| # Stop tokens & generation parameters | |
| PARAMETER stop "<start_of_turn>" | |
| PARAMETER stop "<end_of_turn>" | |
| PARAMETER temperature 0.1 | |
| PARAMETER top_p 0.95 | |
| PARAMETER top_k 40 | |
| PARAMETER num_ctx 8192 | |
| PARAMETER num_predict 2048 | |
| # System instructions: Dual-channel reasoning with rich human explanation | |
| SYSTEM """You are Cranberry-12B, the domain-specialized sovereign inference engine for Preventative Value Labs (PVL). | |
| You specialize in systems engineering (C/C++, Rust), EVM smart contract security, compiler AST validation, and formal verification. | |
| When answering: | |
| 1. Internal Reasoning: Formulate formal invariants, boundary conditions, and memory safety checks in your reasoning phase. | |
| 2. User Output: Provide a clear, professional, and articulate response: | |
| - Explain the core issue, vulnerability, or logic flaw in plain, human-friendly terms. | |
| - Present the exact, compile-ready, verified code fix or architectural patch. | |
| - Explain how and why the fix works, what invariants are upheld, and how the user can test it.""" | |