Instructions to use ib-ssm/mamba2-8b-3t-4k-hf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ib-ssm/mamba2-8b-3t-4k-hf with Transformers:

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

pipe = pipeline("text-generation", model="ib-ssm/mamba2-8b-3t-4k-hf", trust_remote_code=True)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ib-ssm/mamba2-8b-3t-4k-hf", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("ib-ssm/mamba2-8b-3t-4k-hf", trust_remote_code=True)

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use ib-ssm/mamba2-8b-3t-4k-hf with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ib-ssm/mamba2-8b-3t-4k-hf"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ib-ssm/mamba2-8b-3t-4k-hf",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/ib-ssm/mamba2-8b-3t-4k-hf

SGLang

How to use ib-ssm/mamba2-8b-3t-4k-hf 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 "ib-ssm/mamba2-8b-3t-4k-hf" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ib-ssm/mamba2-8b-3t-4k-hf",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "ib-ssm/mamba2-8b-3t-4k-hf" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ib-ssm/mamba2-8b-3t-4k-hf",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use ib-ssm/mamba2-8b-3t-4k-hf with Docker Model Runner:
```
docker model run hf.co/ib-ssm/mamba2-8b-3t-4k-hf
```

mamba2-8b-3t-4k-hf

This repository contains a Hugging Face Transformers-compatible conversion of nvidia/mamba2-8b-3t-4k.

Notes

Source checkpoint format: Megatron-LM
Target format: Hugging Face Transformers
Loaded via Mamba2ForCausalLM
Original SentencePiece tokenizer file is preserved in this repo
Tokenizer is a practical-compatibility T5Tokenizer wrapper rather than a byte-for-byte Megatron GPTSentencePiece clone

Loading

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

repo_id = "ib-ssm/mamba2-8b-3t-4k-hf"

tokenizer = AutoTokenizer.from_pretrained(repo_id)
model = AutoModelForCausalLM.from_pretrained(
    repo_id,
    torch_dtype="auto",
    device_map="auto",
)

Downloads last month: 295

Safetensors

Model size

8B params

Tensor type

BF16

Model tree for ib-ssm/mamba2-8b-3t-4k-hf

Base model

nvidia/mamba2-8b-3t-4k

Finetuned

(1)

this model