Downloads · 30 days
2
5% of all-time downloads
psyphon1/Llama-2-7b-Therapist-Aura-Full
Llama-2-7b-Therapist-Aura-Full is a machine learning model from psyphon1. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Downloads · 30 days
2
5% of all-time downloads
All-time downloads
38
Public
Parameters
6.7B
13.5 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors13.5 GB · 100%
From the Hugging Face model README
Aura-Therapist is the specialized large language model powering the Aura Ai mental health platform. It is a fine-tuned iteration of Meta-Llama-2-7b-chat, specifically engineered to provide empathetic, non-judgmental, and psychologically grounded support interactions.
Unlike general-purpose assistants, this model has been aligned to prioritize active listening, emotional validation, and concise, human-like support.
Lead Developer: Chinmay Duse
The goal of the Aura Ai project is to bridge the gap between rigid chatbots and human-like emotional intelligence. Aura-Therapist is designed to:
Aura-Therapist is an AI tool, not a licensed medical professional.
Aura-Therapist is optimized for efficiency and can run on consumer hardware (minimum 6GB VRAM) using 4-bit quantization.
pip install torch transformers accelerate bitsandbytes
The model is hosted on Hugging Face. Use this script to integrate it into the Aura Ai backend or test environment.
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
# 📍 Official Model Identifier
model_id = "psyphon1/Llama-2-7b-Therapist-Aura-Full"
# ⚙️ Efficient 4-Bit Configuration
bnb_config = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_quant_type="nf4",
bnb_4bit_compute_dtype=torch.float16,
bnb_4bit_use_double_quant=True,
)
# 📥 Load Weights
print(f"⏳ Loading Aura-Therapist from: {model_id}")
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
quantization_config=bnb_config,
device_map="auto"
)
def get_aura_response(user_text):
# System Persona (Critical for maintaining character)
system_prompt = "You are an empathetic AI therapist. Be warm, concise, and helpful."
prompt = f"<s>[INST] <<SYS>>
{system_prompt}
<</SYS>>
{user_text} [/INST]"
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
with torch.no_grad():
output = model.generate(
**inputs,
max_new_tokens=256,
temperature=0.7, # Optimized for balanced empathy
top_p=0.9,
do_sample=True,
pad_token_id=tokenizer.eos_token_id
)
return tokenizer.decode(output[0], skip_special_tokens=True).split("[/INST]")[-1].strip()
# Example Interaction
user_query = "I feel really overwhelmed by my workload and I don't know where to start."
print(f"User: {user_query}")
print(f"Aura: {get_aura_response(user_query)}")
Rank (r): 32
Alpha: 16
Precision: Float16 (fp16)
Dataset: Aura-Curated-2k (Curated clinical instruction pairs, cleaned for repetition removal)
For rapid testing in a terminal environment:
print("🔮 Aura Ai Terminal Initialized...")
print("Type 'exit' to close session.
")
while True:
user_in = input("You: ")
if user_in.lower() in ['exit', 'quit']: break
response = get_aura_response(user_in)
print(f"🤖 Aura: {response}
")
Project: Aura Ai
Lead Developer: Chinmay Duse
License: This model is a derivative of Llama 2 and is subject to the Meta Llama 2 Community License Agreement. Users must comply with Meta's usage policy, particularly regarding responsible AI use.