Text Generation
Transformers
Safetensors
English
deepseek-v3
multi-head-latent-attention
mixture-of-experts
Mixture of Experts
tinystories
tiny-model
validation
debug-model
Instructions to use shibatch/tinydeepseekv3-3m with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use shibatch/tinydeepseekv3-3m with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="shibatch/tinydeepseekv3-3m")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("shibatch/tinydeepseekv3-3m", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use shibatch/tinydeepseekv3-3m with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "shibatch/tinydeepseekv3-3m" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "shibatch/tinydeepseekv3-3m", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/shibatch/tinydeepseekv3-3m
- SGLang
How to use shibatch/tinydeepseekv3-3m 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 "shibatch/tinydeepseekv3-3m" \ --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": "shibatch/tinydeepseekv3-3m", "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 "shibatch/tinydeepseekv3-3m" \ --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": "shibatch/tinydeepseekv3-3m", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use shibatch/tinydeepseekv3-3m with Docker Model Runner:
docker model run hf.co/shibatch/tinydeepseekv3-3m
File size: 1,190 Bytes
6036082 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 | {
"architectures": [
"DeepseekV3ForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"aux_loss_alpha": 0.0001,
"bos_token_id": 1000,
"dtype": "float32",
"eos_token_id": 1001,
"first_k_dense_replace": 1,
"head_dim": 16,
"hidden_act": "silu",
"hidden_size": 216,
"initializer_range": 0.02,
"intermediate_size": 432,
"kv_lora_rank": 64,
"max_position_embeddings": 2048,
"model_type": "deepseek_v3",
"moe_intermediate_size": 128,
"n_group": 1,
"n_routed_experts": 4,
"n_shared_experts": 1,
"norm_topk_prob": true,
"num_attention_heads": 8,
"num_experts_per_tok": 1,
"num_hidden_layers": 5,
"num_key_value_heads": 8,
"num_nextn_predict_layers": 0,
"pad_token_id": 1000,
"pretraining_tp": 1,
"q_lora_rank": 64,
"qk_head_dim": 32,
"qk_nope_head_dim": 16,
"qk_rope_head_dim": 16,
"rms_norm_eps": 1e-06,
"rope_interleave": true,
"rope_parameters": {
"rope_theta": 10000,
"rope_type": "default"
},
"routed_scaling_factor": 2.5,
"seq_aux": true,
"tie_word_embeddings": true,
"topk_group": 1,
"transformers_version": "5.14.1",
"use_cache": false,
"v_head_dim": 32,
"vocab_size": 1024
}
|