Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time
Paper • 2203.05482 • Published • 8
How to use umisetokikaze/vt1 with Transformers:
# Use a pipeline as a high-level helper
from transformers import pipeline
pipe = pipeline("text-generation", model="umisetokikaze/vt1") # Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM
tokenizer = AutoTokenizer.from_pretrained("umisetokikaze/vt1")
model = AutoModelForMultimodalLM.from_pretrained("umisetokikaze/vt1")How to use umisetokikaze/vt1 with vLLM:
# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "umisetokikaze/vt1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "umisetokikaze/vt1",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'docker model run hf.co/umisetokikaze/vt1
How to use umisetokikaze/vt1 with SGLang:
# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
--model-path "umisetokikaze/vt1" \
--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": "umisetokikaze/vt1",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'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 "umisetokikaze/vt1" \
--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": "umisetokikaze/vt1",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'How to use umisetokikaze/vt1 with Docker Model Runner:
docker model run hf.co/umisetokikaze/vt1
This is a merge of pre-trained language models created using mergekit.
This model was merged using the linear merge method.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
models:
- model: models/swallow_cv
parameters:
weight: 1
- model: models/wizard_cv
parameters:
weight: 1
- model: Aratako/Japanese-Starling-ChatV-7B-RP
parameters:
weight: 1
- model: Exveria/merge003
parameters:
weight: 1
merge_method: linear
dtype: float16