Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time
Paper β’ 2203.05482 β’ Published β’ 8
How to use Bread-AI/Crumb-13B with Transformers:
# Use a pipeline as a high-level helper
from transformers import pipeline
pipe = pipeline("text-generation", model="Bread-AI/Crumb-13B") # Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("Bread-AI/Crumb-13B")
model = AutoModelForCausalLM.from_pretrained("Bread-AI/Crumb-13B")How to use Bread-AI/Crumb-13B with vLLM:
# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Bread-AI/Crumb-13B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "Bread-AI/Crumb-13B",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'docker model run hf.co/Bread-AI/Crumb-13B
How to use Bread-AI/Crumb-13B with SGLang:
# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
--model-path "Bread-AI/Crumb-13B" \
--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": "Bread-AI/Crumb-13B",
"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 "Bread-AI/Crumb-13B" \
--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": "Bread-AI/Crumb-13B",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'How to use Bread-AI/Crumb-13B with Docker Model Runner:
docker model run hf.co/Bread-AI/Crumb-13B
Merge of Noromaid, Thorns, and WizardLM for Bread AI
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: NeverSleep/Noromaid-13b-v0.1.1
parameters:
weight: 1.0
- model: WizardLM/WizardLM-13B-V1.2
parameters:
weight: 0.3
- model: CalderaAI/13B-Thorns-l2
parameters:
weight: 0.5
merge_method: linear
dtype: float16