90 lines
2.5 KiB
Markdown
90 lines
2.5 KiB
Markdown
|
---
|
|||
|
|
license: apache-2.0
|
|||
|
|
tags:
|
|||
|
|
- gguf
|
|||
|
|
- llama.cpp
|
|||
|
|
- cannabis
|
|||
|
|
- science
|
|||
|
|
- text-generation
|
|||
|
|
- local-inference
|
|||
|
|
- domain-specific
|
|||
|
|
- chemistry
|
|||
|
|
- biology
|
|||
|
|
- medical
|
|||
|
|
language:
|
|||
|
|
- en
|
|||
|
|
base_model:
|
|||
|
|
- KellanF89/Newton-Insights-V1-cannabis-extraction-science
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## Note on source files
|
|||
|
|
|
|||
|
|
This GGUF was built from the `granite-science-merged` files included in the original repository.
|
|||
|
|
|
|||
|
|
# Newton Insights V1 – Cannabis Extraction Science (GGUF)
|
|||
|
|
|
|||
|
|
## Model Description
|
|||
|
|
|
|||
|
|
> GGUF quantized version for local inference (llama.cpp, LM Studio CLI)
|
|||
|
|
This is a GGUF quantized version of:
|
|||
|
|
KellanF89/Newton-Insights-V1-cannabis-extraction-science
|
|||
|
|
|
|||
|
|
The model is a domain-specific language model trained on cannabis science literature, focusing on:
|
|||
|
|
- extraction techniques
|
|||
|
|
- terpene and cannabinoid chemistry
|
|||
|
|
- pharmacology
|
|||
|
|
- cultivation
|
|||
|
|
|
|||
|
|
This version was converted and quantized for **local inference** using `llama.cpp`-compatible runtimes.
|
|||
|
|
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## Model Details
|
|||
|
|
|
|||
|
|
- **Developed by:** Kellan Finney / Eighth Revolution (original model)
|
|||
|
|
- **Model type:** Causal Language Model (fine-tuned)
|
|||
|
|
- **Language:** English
|
|||
|
|
- **License:** Apache 2.0
|
|||
|
|
- **Finetuned from:** IBM Granite 3.0 8B Instruct
|
|||
|
|
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## Available Quantizations
|
|||
|
|
|
|||
|
|
| File | Quant | Size | Notes |
|
|||
|
|
|------|------|------|------|
|
|||
|
|
| Newton-Insights-V1-cannabis-extraction-science-q4_k_m.gguf | Q4_K_M | ~4-5 GB | Good balance, lower VRAM |
|
|||
|
|
| Newton-Insights-V1-cannabis-extraction-science-q5_k_m.gguf | Q5_K_M | ~5-6 GB | Recommended |
|
|||
|
|
| Newton-Insights-V1-cannabis-extraction-science-q6_k.gguf | Q6_K | ~6-7 GB | Higher quality |
|
|||
|
|
| Newton-Insights-V1-cannabis-extraction-science-f16.gguf | F16 | ~16 GB | Full precision |
|
|||
|
|
|
|||
|
|
## Uses
|
|||
|
|
|
|||
|
|
### Direct Use
|
|||
|
|
|
|||
|
|
- Domain-specific Q&A about cannabis science
|
|||
|
|
- Educational and research purposes
|
|||
|
|
- Local inference setups (offline AI, homelab, experimentation)
|
|||
|
|
|
|||
|
|
### Out-of-Scope Use
|
|||
|
|
|
|||
|
|
- Medical advice or treatment recommendations
|
|||
|
|
- Legal interpretation
|
|||
|
|
- Safety-critical decisions
|
|||
|
|
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## Limitations
|
|||
|
|
|
|||
|
|
- The model may reflect **inconsistencies in cannabis terminology** (e.g., sativa vs. indica classification).
|
|||
|
|
- Not all responses are guaranteed to be scientifically accurate despite domain fine-tuning.
|
|||
|
|
- Performance may vary depending on prompt formatting and frontend compatibility (Granite-based models use non-standard chat templates).
|
|||
|
|
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## How to Use
|
|||
|
|
|
|||
|
|
### llama.cpp
|
|||
|
|
|
|||
|
|
```bash
|
|||
|
|
./main -m ewald1976/Newton-Insights-V1-cannabis-extraction-science-Q5_K_M.gguf -p "Explain terpene preservation in extraction"
|