Files

59 lines
1.7 KiB
Markdown
Raw Permalink Normal View History

---
license: mit
pipeline_tag: text-generation
tags:
- coding
- uncensored
- pytorch
- llama-cpp
- gguf
model-index:
- name: Pyrex-8B-Instruct-Uncensored
results:
- task:
type: text-generation
dataset:
type: openai_humaneval
name: HumanEval
metrics:
- type: pass@1
value: 52.4
name: pass@1 (greedy)
---
# Pyrex-8B-Instruct-Uncensored
Pyrex-8B-Instruct-Uncensored is a high-performance, uncensored coding assistant optimized for complex software engineering tasks, agentic workflows, and function calling.
## Benchmark Results (HumanEval)
| Model | pass@1 |
| :--- | :--- |
| Pyrex-8B-Instruct-Uncensored | **52.4%** |
## Capabilities
- **Unconstrained Coding**: Handles all coding requests without refusal or boilerplate restrictions.
- **Agentic Workflow**: Fine-tuned on high-quality synthetic data for function calling and tool-use integration.
- **Optimized for Engineering**: Trained on a massive corpus of programming literature, algorithms, and system-level codebases.
## Usage
### GGUF (llama.cpp / Ollama)
| Quantization | Recommended Use |
| :--- | :--- |
| **Q4_K_M** | Best balance of speed and quality |
| **Q5_K_M** | Improved accuracy for complex debugging |
| **Q8_0** | Near-lossless performance |
```bash
ollama create pyrex-8b -f Modelfile
./llama-cli -m pyrex-q4_k_m.gguf -p "<|im_start|>user\nExplain recursion.\n<|im_end|>\n<|im_start|>assistant\n" -n 512 -c 8192
```
Download from [ImposterOnline/Pyrex-8B-Instruct-Uncensored-GGUF](https://huggingface.co/ImposterOnline/Pyrex-8B-Instruct-Uncensored-GGUF).
---
*Pyrex is a private, independent project for high-performance, uncensored engineering assistance.*