---
license: apache-2.0
base_model: ProCreations/grug-27b
tags:
- grug
- nvfp4
- fp4
- quantized
- reasoning
language:
- en
---

# grug-27b-nvfp4

grug brain in NVIDIA four-bit float rock. Blackwell GPU eat this format raw =
big speed, small memory, quality mostly keep.

- source: [grug-27b](https://huggingface.co/ProCreations/grug-27b) (v2.1)
- scheme: **NVFP4 (W4A4)** via llm-compressor one-shot
- calibration: 512 samples of grug's own training distribution (agent
  sessions + grug reasoning) so scales match real usage
- kept high precision: lm_head, vision tower, router gates

## how run

```bash
vllm serve ProCreations/grug-27b-nvfp4 --max-model-len 32768 \
  --reasoning-parser deepseek_r1
```

best on Blackwell (native FP4). works on Hopper too (weight-only benefit).
grug think dense inside `<think>` (arrives in `message.reasoning`), answer
normal english. sampling temp 0.6-1.0, top_p 0.95, top_k 20.

family: [grug-27b](https://huggingface.co/ProCreations/grug-27b) |
[grug-27b-mtp](https://huggingface.co/ProCreations/grug-27b-mtp) |
[gguf](https://huggingface.co/ProCreations/grug-27b-gguf) |
[qat-q4](https://huggingface.co/ProCreations/grug-27b-qat-q4-gguf)

grug made by ProCreations.