Text Generation
Transformers
Safetensors
English
hrm_text_moe
hrm
hierarchical-reasoning
mixture-of-experts
Mixture of Experts
sparse-activation
prefix-lm
pre-alignment
non-chat
non-instruction-tuned
custom_code
Instructions to use Xiaoye08/HRM-MoE with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Xiaoye08/HRM-MoE with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="Xiaoye08/HRM-MoE", trust_remote_code=True)# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("Xiaoye08/HRM-MoE", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use Xiaoye08/HRM-MoE with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "Xiaoye08/HRM-MoE" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Xiaoye08/HRM-MoE", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/Xiaoye08/HRM-MoE
- SGLang
How to use Xiaoye08/HRM-MoE 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 "Xiaoye08/HRM-MoE" \ --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": "Xiaoye08/HRM-MoE", "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 "Xiaoye08/HRM-MoE" \ --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": "Xiaoye08/HRM-MoE", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use Xiaoye08/HRM-MoE with Docker Model Runner:
docker model run hf.co/Xiaoye08/HRM-MoE
| { | |
| "model_type": "hrm_text_moe", | |
| "architectures": [ | |
| "HrmTextMoEForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_hrm_text_moe.HrmTextMoEConfig", | |
| "AutoModelForCausalLM": "modeling_hrm_text_moe.HrmTextMoEForCausalLM" | |
| }, | |
| "vocab_size": 65536, | |
| "hidden_size": 1536, | |
| "intermediate_size": 512, | |
| "num_hidden_layers": 16, | |
| "num_attention_heads": 12, | |
| "num_key_value_heads": 12, | |
| "head_dim": 128, | |
| "H_cycles": 2, | |
| "L_cycles": 3, | |
| "L_bp_cycles": [ | |
| 0, | |
| 3 | |
| ], | |
| "max_position_embeddings": 4096, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 10000.0, | |
| "tie_word_embeddings": false, | |
| "initializer_range": 0.025515518153991442, | |
| "embedding_scale": 39.191835884530846, | |
| "prefix_lm": true, | |
| "moe_num_experts": 64, | |
| "moe_top_k": 8, | |
| "moe_intermediate_size": 512, | |
| "moe_implementation": "grouped_triton", | |
| "moe_norm_topk_prob": true, | |
| "moe_router_aux_loss_coef": 0.001, | |
| "pad_token_id": 5, | |
| "condition_mapping": { | |
| "cot": "<|object_ref_end|>", | |
| "direct": "<|object_ref_start|>", | |
| "noisy": "<|quad_start|>", | |
| "synth": "<|quad_end|>" | |
| }, | |
| "bos_token_id": 6, | |
| "eos_token_id": 11 | |
| } | |