Text Generation
Transformers
Safetensors
English
olmo_hils
long-context
sparse-attention
efficient-attention
pretraining
olmo3
hils-attention
Instructions to use tencent/HiLS-Attention-7B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use tencent/HiLS-Attention-7B with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="tencent/HiLS-Attention-7B")# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("tencent/HiLS-Attention-7B", dtype="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use tencent/HiLS-Attention-7B with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "tencent/HiLS-Attention-7B" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "tencent/HiLS-Attention-7B", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/tencent/HiLS-Attention-7B
- SGLang
How to use tencent/HiLS-Attention-7B 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 "tencent/HiLS-Attention-7B" \ --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": "tencent/HiLS-Attention-7B", "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 "tencent/HiLS-Attention-7B" \ --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": "tencent/HiLS-Attention-7B", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use tencent/HiLS-Attention-7B with Docker Model Runner:
docker model run hf.co/tencent/HiLS-Attention-7B
| { | |
| "architectures": [ | |
| "HiLSForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 100257, | |
| "eos_token_id": 100257, | |
| "hidden_act": "silu", | |
| "hidden_size": 4096, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 11008, | |
| "max_position_embeddings": 131072, | |
| "max_window_layers": 32, | |
| "model_type": "olmo_hils", | |
| "num_attention_heads": 32, | |
| "num_key_value_heads": 32, | |
| "num_hidden_layers": 32, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 500000.0, | |
| "sliding_window": 512, | |
| "hils_sliding_window": 512, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.40.1", | |
| "use_cache": true, | |
| "use_sliding_window": true, | |
| "vocab_size": 100278, | |
| "full_attn_interleave": 4, | |
| "chunk_size": 64, | |
| "hils_topk": 32, | |
| "_attn_implementation": "flash_attention_3", | |
| "adjust_lmk_pos": true, | |
| "enable_lmk_q_proj": true, | |
| "layerwise_qk_norm": true, | |
| "apply_hils_rope": true, | |
| "enable_prior_query": true, | |
| "mask_lmk_token": true, | |
| "lmk_q_lora_dim": 256, | |
| "enable_external_lmk_embed": true, | |
| "layerwise_lmkq_norm": true, | |
| "enable_softmax1": false, | |
| "use_hope": true, | |
| "enable_inrange_rope": true, | |
| "rope_context_length": 8192, | |
| "rope_period_multiplier": 2.0 | |
| } | |