Instructions to use HelpingAI/HelpingAI-3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use HelpingAI/HelpingAI-3 with Transformers:

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

pipe = pipeline("text-generation", model="HelpingAI/HelpingAI-3")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("HelpingAI/HelpingAI-3")
model = AutoModelForCausalLM.from_pretrained("HelpingAI/HelpingAI-3", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use HelpingAI/HelpingAI-3 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "HelpingAI/HelpingAI-3"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HelpingAI/HelpingAI-3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/HelpingAI/HelpingAI-3

SGLang

How to use HelpingAI/HelpingAI-3 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 "HelpingAI/HelpingAI-3" \
    --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": "HelpingAI/HelpingAI-3",
		"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 "HelpingAI/HelpingAI-3" \
        --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": "HelpingAI/HelpingAI-3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use HelpingAI/HelpingAI-3 with Docker Model Runner:
```
docker model run hf.co/HelpingAI/HelpingAI-3
```

HelpingAI-3 / README.md

Abhaykoul

Update README.md

0b33dd5 verified over 1 year ago

preview code

Raw

History Blame Contribute Delete

3.21 kB

	---
	license: other
	license_name: helpingai
	license_link: https://helpingai.co/license
	pipeline_tag: text-generation
	language:
	- en
	tags:
	- HelpingAI
	- Emotionally-Intelligent
	- EQ-focused
	- Conversational
	- SLM
	library_name: transformers
	---

	# HelpingAI3

	## Model Description

	HelpingAI3 is an advanced language model developed to excel in emotionally intelligent conversations. Building upon the foundations of HelpingAI2.5, this model offers enhanced emotional understanding and contextual awareness.

	## Model Details

	- Developed by: HelpingAI
	- Model type: Decoder-only large language model
	- Language: English
	- License: [HelpingAI License](https://helpingai.co/license)

	## Training Data

	HelpingAI3 was trained on a diverse dataset comprising:

	- Emotional Dialogues: 15 million rows to enhance conversational intelligence.
	- Therapeutic Exchanges: 3 million rows aimed at providing advanced emotional support.
	- Cultural Conversations: 250,000 rows to improve global awareness.
	- Crisis Response Scenarios: 1 million rows to better handle emergency situations.

	## Training Procedure

	The model underwent the following training processes:

	- Base Model: Initiated from HelpingAI2.5.
	- Emotional Intelligence Training: Employed Reinforcement Learning for Emotion Understanding (RLEU) and context-aware conversational fine-tuning.
	- Optimization: Utilized mixed-precision training and advanced token efficiency techniques.

	## Intended Use

	HelpingAI3 is designed for:

	- AI Companionship & Emotional Support: Offering empathetic interactions.
	- Therapy & Wellbeing Guidance: Assisting in mental health support.
	- Personalized Learning: Tailoring educational content to individual needs.
	- Professional AI Assistance: Enhancing productivity in professional settings.

	## Limitations

	While HelpingAI3 strives for high emotional intelligence, users should be aware of potential limitations:

	- Biases: The model may inadvertently reflect biases present in the training data.
	- Understanding Complex Emotions: There might be challenges in accurately interpreting nuanced human emotions.
	- Not a Substitute for Professional Help: For serious emotional or psychological issues, consulting a qualified professional is recommended.

	## How to Use

	### Using Transformers

	```python
	import torch
	from transformers import AutoModelForCausalLM, AutoTokenizer

	# Load the HelpingAI3 model
	model = AutoModelForCausalLM.from_pretrained("HelpingAI/HelpingAI-3")
	# Load the tokenizer
	tokenizer = AutoTokenizer.from_pretrained("HelpingAI/HelpingAI-3")

	# Define the chat input
	chat = [
	{"role": "system", "content": "You are HelpingAI, an emotional AI. Always answer my questions in the HelpingAI style."},
	{"role": "user", "content": "Introduce yourself."}
	]

	inputs = tokenizer.apply_chat_template(
	chat,
	add_generation_prompt=True,
	return_tensors="pt"
	).to(model.device)

	# Generate text
	outputs = model.generate(
	inputs,
	max_new_tokens=256,
	do_sample=True,
	temperature=0.6,
	top_p=0.9,
	)

	response = outputs[0][inputs.shape[-1]:]
	print(tokenizer.decode(response, skip_special_tokens=True))
	```