Instructions to use MarsupialAI/KitchenSink_103b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MarsupialAI/KitchenSink_103b with Transformers:

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

pipe = pipeline("text-generation", model="MarsupialAI/KitchenSink_103b")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MarsupialAI/KitchenSink_103b")
model = AutoModelForCausalLM.from_pretrained("MarsupialAI/KitchenSink_103b")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use MarsupialAI/KitchenSink_103b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MarsupialAI/KitchenSink_103b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MarsupialAI/KitchenSink_103b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/MarsupialAI/KitchenSink_103b

SGLang

How to use MarsupialAI/KitchenSink_103b 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 "MarsupialAI/KitchenSink_103b" \
    --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": "MarsupialAI/KitchenSink_103b",
		"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 "MarsupialAI/KitchenSink_103b" \
        --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": "MarsupialAI/KitchenSink_103b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use MarsupialAI/KitchenSink_103b with Docker Model Runner:
```
docker model run hf.co/MarsupialAI/KitchenSink_103b
```

MarsupialAI commited on Jan 15, 2024

Commit

cafad1a

verified ·

1 Parent(s): 77234f9

Update README.md

Browse files

Files changed (1) hide show

README.md +9 -4

README.md CHANGED Viewed

@@ -3,8 +3,8 @@ license: llama2
 language:
 - en
 tags:
-- RP
-- ERP
 - chat
 - storywriting
 ---
@@ -14,7 +14,7 @@ tags:
 This model is a rotating-stack merge of three 70b models in a 103b (120 layer) configuration inspired by Venus 103b.  The result of
 this "frankenmerge" is a large model that contains a little bit of everything - including the kitchen sink.  RP, chat, storywriting,
 and instruct are all well supported.  It may or may not code well - I lack the expertise to test it in that capacity, but considering
-the source models, it is unlikely.
 Component models for the rotating stack are
 - royallab/Aetheria-L2-70B
@@ -24,8 +24,13 @@ Component models for the rotating stack are
 Components of those models are purported to include: Nous-Hermes-Llama2-70b, Xwin-LM-7B-V0.1, Mythospice-70b, Euryale-1.3-L2-70B,
 tulu-2-dpo-70b, GOAT-70B-Storytelling, Platypus2-70B-instruct, Lila-70B, SunsetBoulevard, and some private LoRAs.
 This model is uncensored and perfectly capable of generating objectionable material.  However, it is not an explicitely-NSFW model,
-and it has never "gone rogue" and tried to insert NSFW content into SFW prompts in my experience.

 language:
 - en
 tags:
+- rp
+- erp
 - chat
 - storywriting
 ---
 This model is a rotating-stack merge of three 70b models in a 103b (120 layer) configuration inspired by Venus 103b.  The result of
 this "frankenmerge" is a large model that contains a little bit of everything - including the kitchen sink.  RP, chat, storywriting,
 and instruct are all well supported.  It may or may not code well - I lack the expertise to test it in that capacity, but considering
+the source models, I suspect it is unlikely.
 Component models for the rotating stack are
 - royallab/Aetheria-L2-70B
 Components of those models are purported to include: Nous-Hermes-Llama2-70b, Xwin-LM-7B-V0.1, Mythospice-70b, Euryale-1.3-L2-70B,
 tulu-2-dpo-70b, GOAT-70B-Storytelling, Platypus2-70B-instruct, Lila-70B, SunsetBoulevard, and some private LoRAs.
+As all components are based on Llama2 70b, native context length is 4k tokens.  Coherency out to 8k is extremely good with rope scaling,
+but starts to decline beyond that.
 This model is uncensored and perfectly capable of generating objectionable material.  However, it is not an explicitely-NSFW model,
+and it has never "gone rogue" and tried to insert NSFW content into SFW prompts in my experience.  As with any LLM, no factual claims
+made by the model should be taken at face value.  You know that boilerplate safety disclaimer that most professional models have?
+Assume this has it too.  This model is for entertainment purposes only.