Instructions to use hf-internal-testing/MiniMax-tiny with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hf-internal-testing/MiniMax-tiny with Transformers:

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

pipe = pipeline("text-generation", model="hf-internal-testing/MiniMax-tiny")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/MiniMax-tiny")
model = AutoModelForCausalLM.from_pretrained("hf-internal-testing/MiniMax-tiny")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use hf-internal-testing/MiniMax-tiny with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hf-internal-testing/MiniMax-tiny"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hf-internal-testing/MiniMax-tiny",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/hf-internal-testing/MiniMax-tiny

SGLang

How to use hf-internal-testing/MiniMax-tiny 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 "hf-internal-testing/MiniMax-tiny" \
    --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": "hf-internal-testing/MiniMax-tiny",
		"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 "hf-internal-testing/MiniMax-tiny" \
        --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": "hf-internal-testing/MiniMax-tiny",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use hf-internal-testing/MiniMax-tiny with Docker Model Runner:
```
docker model run hf.co/hf-internal-testing/MiniMax-tiny
```

MiniMax-tiny

File size: 1,104 Bytes

54b768b

{
  "architectures": [
    "MiniMaxForCausalLM"
  ],
  "attention_dropout": 0.0,
  "attn_type_list": [
    "linear_attention",
    "full_attention"
  ],
  "block_size": 256,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "full_attn_alpha_factor": 1,
  "full_attn_beta_factor": 1,
  "head_dim": null,
  "hidden_act": "silu",
  "hidden_size": 1024,
  "initializer_range": 0.02,
  "intermediate_size": 3584,
  "layer_types": [
    "full_attention",
    "linear_attention"
  ],
  "linear_attn_alpha_factor": 1,
  "linear_attn_beta_factor": 1,
  "max_position_embeddings": 32768,
  "mlp_alpha_factor": 1,
  "mlp_beta_factor": 1,
  "model_type": "minimax",
  "num_attention_heads": 32,
  "num_experts_per_tok": 2,
  "num_hidden_layers": 2,
  "num_key_value_heads": 8,
  "num_local_experts": 8,
  "output_router_logits": false,
  "rms_norm_eps": 1e-05,
  "rope_theta": 1000000.0,
  "router_aux_loss_coef": 0.001,
  "router_jitter_noise": 0.0,
  "sliding_window": null,
  "tie_word_embeddings": false,
  "torch_dtype": "float32",
  "transformers_version": "4.53.0.dev0",
  "use_cache": true,
  "vocab_size": 3200
}