Instructions to use macadeliccc/Laser-WestLake-2x7b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use macadeliccc/Laser-WestLake-2x7b with Transformers:

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

pipe = pipeline("text-generation", model="macadeliccc/Laser-WestLake-2x7b")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("macadeliccc/Laser-WestLake-2x7b")
model = AutoModelForCausalLM.from_pretrained("macadeliccc/Laser-WestLake-2x7b")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use macadeliccc/Laser-WestLake-2x7b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "macadeliccc/Laser-WestLake-2x7b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "macadeliccc/Laser-WestLake-2x7b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/macadeliccc/Laser-WestLake-2x7b

SGLang

How to use macadeliccc/Laser-WestLake-2x7b 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 "macadeliccc/Laser-WestLake-2x7b" \
    --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": "macadeliccc/Laser-WestLake-2x7b",
		"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 "macadeliccc/Laser-WestLake-2x7b" \
        --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": "macadeliccc/Laser-WestLake-2x7b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use macadeliccc/Laser-WestLake-2x7b with Docker Model Runner:
```
docker model run hf.co/macadeliccc/Laser-WestLake-2x7b
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Laser-Westlake-2x7B

This model is a moerge of cognitivecomputations/WestLake-7B-v2-laser and SanjiWatsuki/Kunoichi-DPO-v2-7B

Process

WestLake-7B-v2-laser is the base model and one of the experts rather than utilizing 3 different models.
Will attempt to laser the final product as well, but given that the base has already been lasered it may not work out.
I have another version using the original, non-lasered WestLake-7B that I will also attempt to laser and report the differences.

Usage

Usage is the same as the original WestLake-7B

GGUF

Available here

Evaluations

----Benchmark Complete----
2024-01-27 19:12:49
Time taken: 24.3 mins
Prompt Format: ChatML
Model: macadeliccc/Laser-WestLake-2x7b-GGUF
Score (v2): 75.42
Parseable: 171.0
---------------
Batch completed
Time taken: 24.4 mins
---------------

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	74.00
AI2 Reasoning Challenge (25-Shot)	72.27
HellaSwag (10-Shot)	88.44
MMLU (5-Shot)	64.71
TruthfulQA (0-shot)	69.25
Winogrande (5-shot)	85.79
GSM8k (5-shot)	63.53

Downloads last month: 70

Safetensors

Model size

13B params

Tensor type

BF16

Model tree for macadeliccc/Laser-WestLake-2x7b

Quantizations

2 models

Collection including macadeliccc/Laser-WestLake-2x7b

WestLake

Collection

5 items • Updated Mar 2 • 1

Evaluation results

normalized accuracy on AI2 Reasoning Challenge (25-Shot)
test set Open LLM Leaderboard

72.270
normalized accuracy on HellaSwag (10-Shot)
validation set Open LLM Leaderboard

88.440
accuracy on MMLU (5-Shot)
test set Open LLM Leaderboard

64.710
mc2 on TruthfulQA (0-shot)
validation set Open LLM Leaderboard

69.250
accuracy on Winogrande (5-shot)
validation set Open LLM Leaderboard

85.790
accuracy on GSM8k (5-shot)
test set Open LLM Leaderboard

63.530