49 lines
1.6 KiB
Markdown
49 lines
1.6 KiB
Markdown
---
|
|
license: apache-2.0
|
|
language:
|
|
- en
|
|
base_model: Shinzmann/naija-petro
|
|
pipeline_tag: text-generation
|
|
tags:
|
|
- petroleum-engineering
|
|
- oil-and-gas
|
|
- nigeria
|
|
- qwen3
|
|
- gguf
|
|
- llama-cpp
|
|
- ollama
|
|
- quantized
|
|
---
|
|
|
|
# Naija-Petro (32B) — GGUF
|
|
|
|
GGUF quantizations of [`Shinzmann/naija-petro`](https://huggingface.co/Shinzmann/naija-petro) (the 32B variant) for inference with [llama.cpp](https://github.com/ggerganov/llama.cpp), [Ollama](https://ollama.com), LM Studio, and compatible runtimes.
|
|
|
|
See the full [model card](https://huggingface.co/Shinzmann/naija-petro) for training details, intended use, and limitations. For a lighter footprint, consider the [8B GGUF](https://huggingface.co/Shinzmann/naija-petro-8b-GGUF). For Nigeria-specific accuracy, use these weights with the [Naija-Petro RAG system](https://github.com/Mystique1337/naija-petro).
|
|
|
|
## Available quantizations
|
|
|
|
| File suffix | Method | Notes |
|
|
|---|---|---|
|
|
| `Q4_K_M` | 4-bit (k-quant, medium) | Smallest; recommended default for 32B on limited RAM/VRAM |
|
|
| `Q5_K_M` | 5-bit (k-quant, medium) | Higher quality, larger |
|
|
| `Q8_0` | 8-bit | Near-lossless; largest and slowest |
|
|
|
|
> 32B GGUF files are large. Q4_K_M is the practical choice for most machines; ensure you have enough RAM/VRAM + disk for the chosen quant.
|
|
|
|
## Usage
|
|
|
|
**Ollama**
|
|
```bash
|
|
ollama run hf.co/Shinzmann/naija-petro-GGUF:Q4_K_M
|
|
```
|
|
|
|
**llama.cpp**
|
|
```bash
|
|
./llama-cli -hf Shinzmann/naija-petro-GGUF:Q4_K_M \
|
|
-p "What are the screening criteria for CO2 EOR?" -c 4096
|
|
```
|
|
|
|
## License
|
|
Apache-2.0 (inherited from Qwen3-32B). Validate outputs with qualified engineers before operational use.
|