Instructions to use openbmb/MiniCPM-SALA with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use openbmb/MiniCPM-SALA with Transformers:

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

pipe = pipeline("text-generation", model="openbmb/MiniCPM-SALA", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("openbmb/MiniCPM-SALA", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use openbmb/MiniCPM-SALA with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/openbmb/MiniCPM-SALA

SGLang

How to use openbmb/MiniCPM-SALA 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 "openbmb/MiniCPM-SALA" \
    --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": "openbmb/MiniCPM-SALA",
		"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 "openbmb/MiniCPM-SALA" \
        --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": "openbmb/MiniCPM-SALA",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use openbmb/MiniCPM-SALA with Docker Model Runner:
```
docker model run hf.co/openbmb/MiniCPM-SALA
```

MiniCPM-SALA

File size: 2,397 Bytes

{
  "_name_or_path": "openbmb/MiniCPM-SALA",
  "architectures": [
    "MiniCPMSALAForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "attn_use_rope": false,
  "auto_map": {
    "AutoConfig": "configuration_minicpm_sala.MiniCPMSALAConfig",
    "AutoModel": "modeling_minicpm_sala.MiniCPMSALAModel",
    "AutoModelForCausalLM": "modeling_minicpm_sala.MiniCPMSALAForCausalLM",
    "AutoModelForSeq2SeqLM": "modeling_minicpm_sala.MiniCPMSALAForCausalLM",
    "AutoModelForSequenceClassification": "modeling_minicpm_sala.MiniCPMSALAForSequenceClassification"
  },
  "bos_token_id": 1,
  "eos_token_id": [
    2,
    73440
  ],
  "pad_token_id": 2,
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_size": 4096,
  "initializer_range": 0.1,
  "intermediate_size": 16384,
  "lightning_head_dim": 128,
  "lightning_nh": 32,
  "lightning_nkv": 32,
  "lightning_scale": "1/sqrt(d)",
  "lightning_use_rope": true,
  "max_position_embeddings": 524288,
  "model_type": "minicpm_sala",
  "mixer_types": [
    "minicpm4",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "minicpm4",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "minicpm4",
    "minicpm4",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "minicpm4",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "lightning-attn",
    "minicpm4",
    "minicpm4",
    "minicpm4"
  ],
  "sparse_config": {
    "kernel_size": 32,
    "kernel_stride": 16,
    "init_blocks": 1,
    "block_size": 64,
    "window_size": 2048,
    "topk": 64,
    "use_nope": false,
    "dense_len": 8192
  },
  "num_attention_heads": 32,
  "num_hidden_layers": 32,
  "num_key_value_heads": 2,
  "qk_norm": true,
  "rand_init": false,
  "rms_norm_eps": 1e-06,
  "torch_dtype": "bfloat16",
  "dtype": "bfloat16",
  "transformers_version": "4.56.0",
  "use_cache": true,
  "vocab_size": 73448,
  "rope_theta": 10000.0,
  "scale_emb": 12,
  "scale_depth": 1.4,
  "mup_denominator": 32,
  "dim_model_base": 256,
  "tie_word_embeddings": false,
  "use_output_gate": true,
  "use_output_norm": true,
  "attn_use_output_gate": true
}