Text Generation
Transformers
Safetensors
English
dami
linear-attention
long-context
initialization
ruler
Instructions to use SpeedAI-Lab/DAMI with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use SpeedAI-Lab/DAMI with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="SpeedAI-Lab/DAMI")# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("SpeedAI-Lab/DAMI", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use SpeedAI-Lab/DAMI with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "SpeedAI-Lab/DAMI" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "SpeedAI-Lab/DAMI", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/SpeedAI-Lab/DAMI
- SGLang
How to use SpeedAI-Lab/DAMI 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 "SpeedAI-Lab/DAMI" \ --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": "SpeedAI-Lab/DAMI", "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 "SpeedAI-Lab/DAMI" \ --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": "SpeedAI-Lab/DAMI", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use SpeedAI-Lab/DAMI with Docker Model Runner:
docker model run hf.co/SpeedAI-Lab/DAMI
Download checkpoints/main_gdn2_random/config.json from SpeedAI-Lab/DAMI: direct link, hf CLI and curl.
- Browser
- Download file 1.24 kB
-
https://huggingface.co/SpeedAI-Lab/DAMI/resolve/refs%2Fpr%2F1/checkpoints/main_gdn2_random/config.json
- Command line
-
hf download hf://SpeedAI-Lab/DAMI@refs/pr/1/checkpoints/main_gdn2_random/config.json
-
curl -L -o config.json https://huggingface.co/SpeedAI-Lab/DAMI/resolve/refs%2Fpr%2F1/checkpoints/main_gdn2_random/config.json
1.24 kB
| { | |
| "allow_neg_eigval": false, | |
| "architectures": [ | |
| "GDN2ForCausalLM" | |
| ], | |
| "attn": null, | |
| "attn_mode": "chunk", | |
| "attnres_block_size": null, | |
| "bos_token_id": 1, | |
| "conv_size": 4, | |
| "dtype": "float32", | |
| "eos_token_id": 2, | |
| "expand_v": 2, | |
| "fuse_cross_entropy": true, | |
| "fuse_linear_cross_entropy": false, | |
| "fuse_norm": true, | |
| "fuse_swiglu": true, | |
| "head_dim": 64, | |
| "head_init_design": null, | |
| "hidden_act": "swish", | |
| "hidden_ratio": 4, | |
| "hidden_size": 512, | |
| "initialization_manifest": { | |
| "A_log_rule": "GDN v3_random_a source A_log", | |
| "dt_bias_rule": "GDN-aligned head means with deterministic small channel noise; channel_std=0.03", | |
| "experiment": "v5_random_a", | |
| "family": "GDN2", | |
| "gdn_source_csv": "/root/checkpoints/gdn/training/gdn100m-v3-random-5b-v3_random_a/initial_head_params.csv" | |
| }, | |
| "initializer_range": 0.02, | |
| "intermediate_size": null, | |
| "max_position_embeddings": 4096, | |
| "model_type": "gdn2", | |
| "norm_eps": 1e-06, | |
| "num_heads": 6, | |
| "num_hidden_layers": 18, | |
| "num_v_heads": 6, | |
| "pad_token_id": 0, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "5.9.0", | |
| "use_cache": true, | |
| "use_gate": true, | |
| "use_l2warp": false, | |
| "use_short_conv": true, | |
| "vocab_size": 32000 | |
| } | |