File size: 1,466 Bytes
5f347ba
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: apache-2.0
base_model:
  - mistralai/Mistral-Small-3.1-24B-Instruct-2503
  - RedHatAI/Mistral-Small-3.1-24B-Instruct-2503-FP8-dynamic
tags:
  - flux2
  - text-encoder
  - fp8
  - mistral
---

# FLUX.2 Text Encoder (FP8)

Combined repo for FLUX.2 text encoding with FP8 quantization (~24GB VRAM instead of ~48GB).

## Components

| Component | Source |
|-----------|--------|
| FP8 Model Weights | [RedHatAI/Mistral-Small-3.1-24B-Instruct-2503-FP8-dynamic](https://huggingface.co/RedHatAI/Mistral-Small-3.1-24B-Instruct-2503-FP8-dynamic) |
| Tokenizer/Processor | [mistralai/Mistral-Small-3.1-24B-Instruct-2503](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503) |

## Usage

```python
from transformers import AutoProcessor, Mistral3ForConditionalGeneration

model = Mistral3ForConditionalGeneration.from_pretrained(
    "TensorTemplar/flux2-text-encoder-fp8",
    local_files_only=True,  # or False for download
)
processor = AutoProcessor.from_pretrained(
    "TensorTemplar/flux2-text-encoder-fp8",
    use_fast=False,
)
```

## Purpose

This repo exists to simplify FLUX.2 deployment by combining all necessary text encoder
components into a single download. Used for extracting intermediate hidden states
(layers 10/20/30) for image generation conditioning.

## Attribution

- FP8 quantization by [RedHatAI](https://huggingface.co/RedHatAI) using llm-compressor
- Original model by [Mistral AI](https://huggingface.co/mistralai)