Instructions to use cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text")
model = AutoModelForCausalLM.from_pretrained("cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text

SGLang

How to use cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text with Docker Model Runner:
```
docker model run hf.co/cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text
```

Ministral-3 3B Instruct FP8 Llama Text

Text-only Llama-compatible conversion of mistralai/Ministral-3-3B-Instruct-2512.

What changed:

dropped vision and multimodal projector tensors
converted native consolidated Mistral text keys to Llama/HF keys
applied the required native-to-HF Q/K RoPE weight permutation
wrote a plain LlamaForCausalLM config
kept tokenizer assets and chat template with the corrected regex
kept chat formatting in chat_template.jinja
removed the strict user/assistant alternation assertion from the template
left tokenizer loading on the generic fast backend, not LlamaTokenizerFast

Verification:

reference: mistralai/Ministral-3-3B-Instruct-2512
candidate: this checkpoint
dataset: /home/alvion/valve/services/training/datasets/think2-2025-12-07_gpt-5.4_reasoning.jsonl
rows: 3
max length: 512
tokenizer IDs: identical
worst KL: 0
worst logit diff: 0
plain candidate AutoTokenizer matched the fixed reference tokenizer

FP8 forward passes require the Transformers fine-grained FP8 kernels package.

Conversion tools: https://github.com/cascade-tech-ai/mistral-convert

Downloads last month: 174

Safetensors

Model size

3B params

Tensor type

BF16

F8_E4M3

Model tree for cascade-tech/Ministral-3-3B-Instruct-2512-FP8-llama-text

Base model

mistralai/Ministral-3-3B-Base-2512

Quantized

mistralai/Ministral-3-3B-Instruct-2512

Quantized

(38)

this model