Instructions to use LongSafari/hyenadna-large-1m-seqlen-hf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LongSafari/hyenadna-large-1m-seqlen-hf with Transformers:

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

pipe = pipeline("text-generation", model="LongSafari/hyenadna-large-1m-seqlen-hf", trust_remote_code=True)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("LongSafari/hyenadna-large-1m-seqlen-hf", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use LongSafari/hyenadna-large-1m-seqlen-hf with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "LongSafari/hyenadna-large-1m-seqlen-hf"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LongSafari/hyenadna-large-1m-seqlen-hf",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/LongSafari/hyenadna-large-1m-seqlen-hf

SGLang

How to use LongSafari/hyenadna-large-1m-seqlen-hf 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 "LongSafari/hyenadna-large-1m-seqlen-hf" \
    --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": "LongSafari/hyenadna-large-1m-seqlen-hf",
		"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 "LongSafari/hyenadna-large-1m-seqlen-hf" \
        --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": "LongSafari/hyenadna-large-1m-seqlen-hf",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use LongSafari/hyenadna-large-1m-seqlen-hf with Docker Model Runner:
```
docker model run hf.co/LongSafari/hyenadna-large-1m-seqlen-hf
```

Matt commited on Jan 24, 2024

Commit

0a629ab

1 Parent(s): 8eb99a8

Mark z correctly as a buffer

Browse files

Files changed (1) hide show

modeling_hyena.py +2 -2

modeling_hyena.py CHANGED Viewed

@@ -62,8 +62,8 @@ class HyenaPositionalEmbedding(nn.Module):
         f = torch.linspace(1e-4, bands - 1, bands)[None, None]
         z = torch.cat([t, torch.cos(-f * w), torch.sin(-f * w)], dim=-1)
-        # The original code sets z's LR to lr_pos_emb, which is 1e-5 by default
-        self.z = nn.Parameter(z, requires_grad=True)
         self.register_buffer("t", t)
     def forward(self, L):

         f = torch.linspace(1e-4, bands - 1, bands)[None, None]
         z = torch.cat([t, torch.cos(-f * w), torch.sin(-f * w)], dim=-1)
+        self.register_buffer("z", z)
         self.register_buffer("t", t)
     def forward(self, L):