Instructions to use OrionZheng/openmoe-base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use OrionZheng/openmoe-base with Transformers:

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

pipe = pipeline("text-generation", model="OrionZheng/openmoe-base", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("OrionZheng/openmoe-base", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("OrionZheng/openmoe-base", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use OrionZheng/openmoe-base with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/OrionZheng/openmoe-base

SGLang

How to use OrionZheng/openmoe-base 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 "OrionZheng/openmoe-base" \
    --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": "OrionZheng/openmoe-base",
		"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 "OrionZheng/openmoe-base" \
        --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": "OrionZheng/openmoe-base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use OrionZheng/openmoe-base with Docker Model Runner:
```
docker model run hf.co/OrionZheng/openmoe-base
```

OrionZheng commited on Dec 30, 2023

Commit

e00ffd4

1 Parent(s): 1f9518c

Update config.json

Browse files

Files changed (1) hide show

config.json +5 -1

config.json CHANGED Viewed

@@ -19,6 +19,7 @@
   "hidden_size": 768,
   "initializer_range": 0.02,
   "intermediate_size": 2048,
   "layer_norm_epsilon": 1e-06,
   "load_balance_beam_width": 8,
   "load_balance_group_swap_factor": 0.4,
@@ -36,15 +37,18 @@
   "rms_norm_eps": 1e-06,
   "rope_scaling": null,
   "rope_theta": 10000.0,
   "router_capacity_factor_eval": 2.0,
   "router_capacity_factor_train": 1.25,
   "router_drop_tks": true,
   "router_min_capacity": 4,
   "router_noisy_policy": null,
   "router_topk": 2,
   "tie_word_embeddings": false,
   "torch_dtype": "float32",
   "transformers_version": "4.34.0",
   "use_cache": true,
-  "vocab_size": 256384
 }

   "hidden_size": 768,
   "initializer_range": 0.02,
   "intermediate_size": 2048,
+  "label_smoothing": 0.001,
   "layer_norm_epsilon": 1e-06,
   "load_balance_beam_width": 8,
   "load_balance_group_swap_factor": 0.4,
   "rms_norm_eps": 1e-06,
   "rope_scaling": null,
   "rope_theta": 10000.0,
+  "router_aux_loss_factor": 0.01,
   "router_capacity_factor_eval": 2.0,
   "router_capacity_factor_train": 1.25,
   "router_drop_tks": true,
   "router_min_capacity": 4,
   "router_noisy_policy": null,
   "router_topk": 2,
+  "router_z_loss_factor": 0.0001,
   "tie_word_embeddings": false,
   "torch_dtype": "float32",
   "transformers_version": "4.34.0",
   "use_cache": true,
+  "vocab_size": 256384,
+  "z_loss_factor": 0.01
 }