File size: 1,433 Bytes
6ad90cc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
---
license: other
base_model: TheDrummer/Behemoth-R1-123B-v2
tags:
  - nvfp4
  - modelopt
  - quantized
  - blackwell
  - b200
library_name: transformers
---

# Behemoth-R1-V2 ModelOpt NVFP4

NVFP4 quantized version of [TheDrummer/Behemoth-R1-123B-v2](https://huggingface.co/TheDrummer/Behemoth-R1-123B-v2) using NVIDIA Model Optimizer.

## Quantization Details

| Property | Value |
|----------|-------|
| **Original Model** | TheDrummer/Behemoth-R1-123B-v2 |
| **Quantization** | NVFP4 (FP4 weights, FP16 activations) |
| **Method** | NVIDIA ModelOpt PTQ |
| **Calibration Samples** | 512 |
| **Max Sequence Length** | 4096 |

## Hardware Requirements

- **Optimal**: NVIDIA Blackwell GPUs (B100, B200, RTX PRO 6000 Blackwell)
- **Compatible**: Hopper/Ampere (will use weight-only mode)

## Usage with vLLM

```python
from vllm import LLM, SamplingParams

llm = LLM(
    model="TheHouseOfTheDude/Behemoth-R1-V2_ModelOpt-NVFP4",
    quantization="modelopt",
    trust_remote_code=True,
)

sampling_params = SamplingParams(temperature=0.8, top_p=0.95, max_tokens=512)
outputs = llm.generate(["Write a story about..."], sampling_params)
print(outputs[0].outputs[0].text)
```

## Chat Template

Uses Mistral v7 (Non-Tekken) format. See the original model card for usage details.

## Credits

- Original Model: [TheDrummer](https://huggingface.co/TheDrummer)
- Quantization: TheHouseOfTheDude
- Quantization Framework: NVIDIA ModelOpt