Instructions to use laughatsky/DARE_TIES_InstructMath with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use laughatsky/DARE_TIES_InstructMath with Transformers:

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

pipe = pipeline("text-generation", model="laughatsky/DARE_TIES_InstructMath")

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

tokenizer = AutoTokenizer.from_pretrained("laughatsky/DARE_TIES_InstructMath")
model = AutoModelForCausalLM.from_pretrained("laughatsky/DARE_TIES_InstructMath", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use laughatsky/DARE_TIES_InstructMath with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/laughatsky/DARE_TIES_InstructMath

SGLang

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

Docker Model Runner
How to use laughatsky/DARE_TIES_InstructMath with Docker Model Runner:
```
docker model run hf.co/laughatsky/DARE_TIES_InstructMath
```

DARE_TIES_InstructMath / mergekit_config.yml

laughatsky

Upload folder using huggingface_hub

088fd45 verified 10 months ago

Raw History Blame Contribute Delete

437 Bytes

	base_model: unsloth/llama-2-13b
	chat_template: auto
	dtype: bfloat16
	merge_method: dare_ties
	modules:
	default:
	slices:
	- sources:
	- layer_range: [0, 40]
	model: WizardLMTeam/WizardLM-13B-V1.2
	parameters:
	weight: 1.0
	- layer_range: [0, 40]
	model: layoric/llama-2-13b-code-alpaca
	parameters:
	weight: 1.0
	- layer_range: [0, 40]
	model: unsloth/llama-2-13b