Instructions to use harmtech/SthenoWriter-L2-13B-GPTQ with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use harmtech/SthenoWriter-L2-13B-GPTQ with Transformers:

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

pipe = pipeline("text-generation", model="harmtech/SthenoWriter-L2-13B-GPTQ")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("harmtech/SthenoWriter-L2-13B-GPTQ")
model = AutoModelForCausalLM.from_pretrained("harmtech/SthenoWriter-L2-13B-GPTQ")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use harmtech/SthenoWriter-L2-13B-GPTQ with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "harmtech/SthenoWriter-L2-13B-GPTQ"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "harmtech/SthenoWriter-L2-13B-GPTQ",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/harmtech/SthenoWriter-L2-13B-GPTQ

SGLang

How to use harmtech/SthenoWriter-L2-13B-GPTQ 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 "harmtech/SthenoWriter-L2-13B-GPTQ" \
    --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": "harmtech/SthenoWriter-L2-13B-GPTQ",
		"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 "harmtech/SthenoWriter-L2-13B-GPTQ" \
        --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": "harmtech/SthenoWriter-L2-13B-GPTQ",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use harmtech/SthenoWriter-L2-13B-GPTQ with Docker Model Runner:
```
docker model run hf.co/harmtech/SthenoWriter-L2-13B-GPTQ
```

Daddy Dave's stamp of approval 👍

4-bit GPTQ quants of the writer version of Sao10K's fantastic SthenoWriter model (Stheno model collection link)

The main branch contains 4-bit groupsize of 128 and no act_order.

The other branches contain groupsizes of 128, 64, and 32 all with act_order.

⬇︎ Original card ⬇︎

A Stheno-1.8 Variant focused on writing.

Stheno-1.8 + Storywriter, mixed with Holodeck + Spring Dragon qLoRA. End Result is mixed with One More Experimental Literature-based LoRA.

Re-Reviewed... it's not bad, honestly.

Support me here :)

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	48.35
ARC (25-shot)	62.29
HellaSwag (10-shot)	83.28
MMLU (5-shot)	56.14
TruthfulQA (0-shot)	44.72
Winogrande (5-shot)	74.35
GSM8K (5-shot)	11.22
DROP (3-shot)	6.48

Downloads last month: 6

Safetensors

Model size

2B params

Tensor type

I32

F16

Collection including harmtech/SthenoWriter-L2-13B-GPTQ

Stheno GPTQ

Collection

GPTQ quants of Sao10K's (https://huggingface.co/Sao10K) Stheno 13B model • 6 items • Updated Dec 24, 2023