Instructions to use hf-internal-testing/tiny-random-step3p7 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hf-internal-testing/tiny-random-step3p7 with Transformers:

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

pipe = pipeline("text-generation", model="hf-internal-testing/tiny-random-step3p7")

# Load model directly
from transformers import AutoModelForSeq2SeqLM
model = AutoModelForSeq2SeqLM.from_pretrained("hf-internal-testing/tiny-random-step3p7", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use hf-internal-testing/tiny-random-step3p7 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hf-internal-testing/tiny-random-step3p7"
# 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/tiny-random-step3p7",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/hf-internal-testing/tiny-random-step3p7

SGLang

How to use hf-internal-testing/tiny-random-step3p7 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/tiny-random-step3p7" \
    --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/tiny-random-step3p7",
		"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/tiny-random-step3p7" \
        --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/tiny-random-step3p7",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use hf-internal-testing/tiny-random-step3p7 with Docker Model Runner:
```
docker model run hf.co/hf-internal-testing/tiny-random-step3p7
```

tiny-random-step3p7

File size: 2,813 Bytes

{
  "architectures": [
    "Step3p7ForConditionalGeneration"
  ],
  "dtype": "float32",
  "image_token_id": 63,
  "model_type": "step3p7",
  "projector_bias": false,
  "text_config": {
    "attention_bias": false,
    "attention_dropout": 0.0,
    "bos_token_id": 0,
    "eos_token_id": 2,
    "gating": true,
    "head_dim": 8,
    "hidden_act": "silu",
    "hidden_size": 32,
    "initializer_range": 0.02,
    "intermediate_size": 88,
    "layer_types": [
      "full_attention",
      "sliding_attention",
      "sliding_attention",
      "sliding_attention",
      "full_attention",
      "sliding_attention",
      "sliding_attention",
      "sliding_attention"
    ],
    "max_position_embeddings": 262144,
    "mlp_bias": false,
    "mlp_layer_types": [
      "dense",
      "dense",
      "sparse",
      "sparse",
      "sparse",
      "sparse",
      "sparse",
      "sparse"
    ],
    "model_type": "step3p5",
    "moe_intermediate_size": 10,
    "moe_router_scaling_factor": 3.0,
    "n_routed_experts": 8,
    "num_attention_heads": 8,
    "num_attention_heads_per_layer": [
      8,
      12,
      12,
      12,
      8,
      12,
      12,
      12
    ],
    "num_experts_per_tok": 2,
    "num_hidden_layers": 8,
    "num_key_value_heads": 2,
    "num_nextn_predict_layers": 0,
    "num_sliding_attention_heads": 12,
    "pad_token_id": 1,
    "query_pre_attn_scalar": 8,
    "rms_norm_eps": 1e-05,
    "rope_parameters": {
      "full_attention": {
        "factor": 2.0,
        "high_freq_factor": 32.0,
        "low_freq_factor": 1.0,
        "original_max_position_embeddings": 131072,
        "partial_rotary_factor": 0.5,
        "rope_theta": 5000000.0,
        "rope_type": "llama3"
      },
      "sliding_attention": {
        "partial_rotary_factor": 1.0,
        "rope_theta": 10000.0,
        "rope_type": "default"
      }
    },
    "share_expert_dim": 10,
    "sliding_window": 6,
    "swiglu_limits": [
      null,
      null,
      null,
      null,
      null,
      null,
      7,
      7
    ],
    "swiglu_limits_shared": [
      null,
      null,
      null,
      null,
      null,
      null,
      16,
      16
    ],
    "tie_word_embeddings": false,
    "use_cache": true,
    "vocab_size": 64
  },
  "transformers_version": "5.15.0.dev0",
  "vision_config": {
    "attention_dropout": 0.0,
    "hidden_act": "quick_gelu",
    "hidden_size": 16,
    "image_size": 16,
    "intermediate_size": 93,
    "layer_norm_eps": 1e-05,
    "layer_scale_init_value": 0.1,
    "max_position_embeddings": 16,
    "mlp_ratio": 5.833333333333333,
    "model_type": "step3p5_vision",
    "num_attention_heads": 4,
    "num_channels": 3,
    "num_hidden_layers": 2,
    "patch_size": 4,
    "rope_parameters": {
      "rope_theta": 10000.0,
      "rope_type": "default"
    }
  }
}