Instructions to use victunes/TherapyLlama-8B-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use victunes/TherapyLlama-8B-v1 with Transformers:

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

pipe = pipeline("text-generation", model="victunes/TherapyLlama-8B-v1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("victunes/TherapyLlama-8B-v1")
model = AutoModelForCausalLM.from_pretrained("victunes/TherapyLlama-8B-v1", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use victunes/TherapyLlama-8B-v1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "victunes/TherapyLlama-8B-v1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "victunes/TherapyLlama-8B-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/victunes/TherapyLlama-8B-v1

SGLang

How to use victunes/TherapyLlama-8B-v1 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 "victunes/TherapyLlama-8B-v1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "victunes/TherapyLlama-8B-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "victunes/TherapyLlama-8B-v1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "victunes/TherapyLlama-8B-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use victunes/TherapyLlama-8B-v1 with Docker Model Runner:
```
docker model run hf.co/victunes/TherapyLlama-8B-v1
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

GGUF: https://huggingface.co/victunes/TherapyLlama-8B-v1-GGUF

TherapyLlama 8B v1

Buddy is here for you.

Trained on Llama 3 8B using a modified version of jerryjalapeno/nart-100k-synthetic.

It is a Llama 3 version of https://huggingface.co/victunes/TherapyBeagle-11B-v2

TherapyLlama is hopefully aligned to be helpful, healthy, and comforting.

IMPORTANT: Use Vicuna format

This was trained with the Vicuna 1.1 format. Use the Vicuna format! Do NOT use the Llama 3 format.

Usage

Do not hold back on Buddy.
Open up to Buddy.
Pour your heart out to Buddy.
Engage with Buddy.
Remember that Buddy is just an AI.

Notes

Tested with the Vicuna 1.1 Format
You might be assigned a random name if you don't give yourself one.
RP capabilities are weaker compared to TherapyBeagle.
- Make sure to say "You are Socrates" for its best RP attempt.

Disclaimer

TherapyLlama is NOT a real therapist. It is a friendly AI that mimics empathy and psychotherapy. It is an illusion without the slightest clue who you are as a person. As much as it can help you with self-discovery, A LLAMA IS NOT A SUBSTITUTE to a real professional.

With that said, I found it to be a very helpful tool in unravelling one's self. I hope this model can help sooth minds and hearts.

Examples

LLAMA 3 INSTRUCT

{{[INPUT]}}
Hello, I'm victunes
{{[OUTPUT]}}
Hi victunes! Welcome to our session. I'm here to listen and support you. How are you feeling today?
{{[INPUT]}}
I've been feeling down lately.
{{[OUTPUT]}}
I'm sorry to hear that. Can you tell me more about what's been going on?
{{[INPUT]}}
I think I'm pretty lonely.
{{[OUTPUT]}}
Loneliness can be really tough. It sounds like it's been weighing heavily on your mind. What makes you feel lonely?
{{[INPUT]}}
My friend died recently.
{{[OUTPUT]}}
Oh, I'm so sorry for your loss, victunes. Losing a friend is never easy. How have you been coping with the grief?

GGUF: https://huggingface.co/victunes/TherapyLlama-8B-v1-GGUF

Downloads last month: 33

Model tree for victunes/TherapyLlama-8B-v1

Merges

2 models

Quantizations