Instructions to use lmsys/vicuna-13b-delta-v0 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use lmsys/vicuna-13b-delta-v0 with Transformers:

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

pipe = pipeline("text-generation", model="lmsys/vicuna-13b-delta-v0")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("lmsys/vicuna-13b-delta-v0")
model = AutoModelForCausalLM.from_pretrained("lmsys/vicuna-13b-delta-v0", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use lmsys/vicuna-13b-delta-v0 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "lmsys/vicuna-13b-delta-v0"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "lmsys/vicuna-13b-delta-v0",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/lmsys/vicuna-13b-delta-v0

SGLang

How to use lmsys/vicuna-13b-delta-v0 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 "lmsys/vicuna-13b-delta-v0" \
    --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": "lmsys/vicuna-13b-delta-v0",
		"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 "lmsys/vicuna-13b-delta-v0" \
        --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": "lmsys/vicuna-13b-delta-v0",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use lmsys/vicuna-13b-delta-v0 with Docker Model Runner:
```
docker model run hf.co/lmsys/vicuna-13b-delta-v0
```

RuntimeError: The size of tensor a (32000) must match the size of tensor b (32001) at non-singleton dimension 0

by zeroc - opened Apr 18, 2023

Discussion

zeroc

Apr 18, 2023

When trying to merge delta for 13B I am getting
RuntimeError: The size of tensor a (32000) must match the size of tensor b (32001) at non-singleton dimension 0

any help ?

zeroc changed discussion status to closed Apr 18, 2023

lucaardito

Apr 19, 2023

I have the same issue:

Loading base model
Loading checkpoint shards: 100%|██████████████████████████████████████████████████████████████████████| 3/3 [00:04<00:00, 1.65s/it]
Loading delta
Loading checkpoint shards: 100%|██████████████████████████████████████████████████████████████████████| 3/3 [00:14<00:00, 4.68s/it]
Applying delta
Applying delta: 0%| | 0/403 [00:00<?, ?it/s]
Traceback (most recent call last):
File "/opt/homebrew/Cellar/python@3.10/3.10.11/Frameworks/Python.framework/Versions/3.10/lib/python3.10/runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/opt/homebrew/Cellar/python@3.10/3.10.11/Frameworks/Python.framework/Versions/3.10/lib/python3.10/runpy.py", line 86, in _run_code
exec(code, run_globals)
File "/Users/hardo/git/FastChat/fastchat/model/apply_delta.py", line 49, in
apply_delta(args.base_model_path, args.target_model_path, args.delta_path)
File "/Users/hardo/git/FastChat/fastchat/model/apply_delta.py", line 34, in apply_delta
param.data += delta.state_dict()[name]
RuntimeError: The size of tensor a (32001) must match the size of tensor b (32000) at non-singleton dimension 0

wanghao-007

Apr 20, 2023

I have the same problem, and I don't know how to solve it. someone help me?

lmzheng

Large Model Systems Organization org Apr 20, 2023

It seems you were using a newer version of fschat with these old weights.
Please checkout the version compatibility here https://github.com/lm-sys/FastChat/blob/main/docs/weights_version.md
We suggest you use the newer v1.1 weights.

huxiaoke

Apr 21, 2023

I also get the similar problem when i deploy minigpt4 : RuntimeError: The size of tensor a (32001) must match the size of tensor b (32003) at non-singleton dimension 0
vicuna-13b-delta-v0
llama-13b-hf
fschat version is 0.1.10

lucaardito

Apr 21, 2023

after updating fastchat, and using v1.1, it worked.

Thanks!

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment