Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,265 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: other
|
3 |
+
license_name: helpingai
|
4 |
+
license_link: LICENSE
|
5 |
+
pipeline_tag: text-generation
|
6 |
+
language:
|
7 |
+
- en
|
8 |
+
tags:
|
9 |
+
- HelpingAI
|
10 |
+
- Emotionally-Intelligent
|
11 |
+
- EQ-focused
|
12 |
+
- Conversational
|
13 |
+
- SLM
|
14 |
+
|
15 |
+
---
|
16 |
+
|
17 |
+
<div align="center">
|
18 |
+
|
19 |
+
# π€ HelpingAI-3B-reloaded
|
20 |
+
|
21 |
+
***A Revolutionary Emotionally Intelligent Language Model***
|
22 |
+
|
23 |
+
[![GitHub Organization](https://img.shields.io/badge/GitHub-Organization-blue.svg)](https://github.com/HelpingAI)
|
24 |
+
[![Hugging Face](https://img.shields.io/badge/π€%20Hugging%20Face-Organization-yellow)](https://huggingface.co/OEvortex)
|
25 |
+
[![Model License](https://img.shields.io/badge/License-HelpingAI-green.svg)](LICENSE)
|
26 |
+
[![Discord](https://img.shields.io/discord/1234567890?color=7289da&label=Discord&logo=discord&logoColor=ffffff)](https://discord.gg/YweJwNqrnH)
|
27 |
+
|
28 |
+
<a href="https://github.com/HelpingAI/community/discussions">
|
29 |
+
<img src="https://img.shields.io/badge/Join-Community%20Discussion-blue?style=for-the-badge&logo=github" alt="Join Community Discussion">
|
30 |
+
</a>
|
31 |
+
|
32 |
+
[π License](LICENSE) | [π Website](https://helpingai-in.netlify.app/)
|
33 |
+
|
34 |
+
</div>
|
35 |
+
|
36 |
+
<div align="center">
|
37 |
+
<img src="HelpingAI.png" alt="HelpingAI Logo" width="300px">
|
38 |
+
</div>
|
39 |
+
|
40 |
+
---
|
41 |
+
|
42 |
+
<div align="center">
|
43 |
+
|
44 |
+
## π Model Overview
|
45 |
+
|
46 |
+
**HelpingAI-3B-reloaded** is a state-of-the-art small language model specifically designed for emotionally intelligent conversations and human-centric interactions.
|
47 |
+
|
48 |
+
</div>
|
49 |
+
|
50 |
+
### π― Key Highlights
|
51 |
+
|
52 |
+
- **Architecture**: 3B parameter transformer-based model
|
53 |
+
- **Training Focus**: Emotional intelligence and empathetic responses
|
54 |
+
- **Emotion Score**: Achieves 89.61 on standardized emotional intelligence tests
|
55 |
+
- **Deployment**: Optimized for both GPU and CPU environments
|
56 |
+
|
57 |
+
|
58 |
+
---
|
59 |
+
|
60 |
+
<div align="center">
|
61 |
+
|
62 |
+
## π Model Performance
|
63 |
+
|
64 |
+
</div>
|
65 |
+
|
66 |
+
### Benchmarks
|
67 |
+
|
68 |
+
| Metric | Score | Comparison to Baseline |
|
69 |
+
|--------|-------|----------------------|
|
70 |
+
| Emotion Score | 89.61 | +9.32% |
|
71 |
+
| Memory Usage | 6GB | -30% |
|
72 |
+
| Context| 128k | +3100% |
|
73 |
+
| Context Retention | 92% | +18% |
|
74 |
+
|
75 |
+
|
76 |
+
---
|
77 |
+
|
78 |
+
<div align="center">
|
79 |
+
|
80 |
+
## π» Implementation
|
81 |
+
|
82 |
+
</div>
|
83 |
+
|
84 |
+
### Transformers
|
85 |
+
|
86 |
+
```python
|
87 |
+
import torch
|
88 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
89 |
+
|
90 |
+
# Load the HelpingAI-3B model
|
91 |
+
model = AutoModelForCausalLM.from_pretrained("OEvortex/HelpingAI-3B-reloaded")
|
92 |
+
# Load the tokenizer
|
93 |
+
tokenizer = AutoTokenizer.from_pretrained("OEvortex/HelpingAI-3B-reloaded")
|
94 |
+
|
95 |
+
|
96 |
+
# Define the chat input
|
97 |
+
chat = [
|
98 |
+
{ "role": "system", "content": "You are HelpingAI, an emotional AI. Always answer my questions in the HelpingAI style." },
|
99 |
+
{ "role": "user", "content": "GIVE ME YOUR INTRO" }
|
100 |
+
]
|
101 |
+
|
102 |
+
inputs = tokenizer.apply_chat_template(
|
103 |
+
chat,
|
104 |
+
add_generation_prompt=True,
|
105 |
+
return_tensors="pt"
|
106 |
+
).to(model.device)
|
107 |
+
|
108 |
+
|
109 |
+
# Generate text
|
110 |
+
outputs = model.generate(
|
111 |
+
inputs,
|
112 |
+
max_new_tokens=256,
|
113 |
+
do_sample=True,
|
114 |
+
temperature=0.6,
|
115 |
+
top_p=0.9,
|
116 |
+
)
|
117 |
+
|
118 |
+
|
119 |
+
response = outputs[0][inputs.shape[-1]:]
|
120 |
+
print(tokenizer.decode(response, skip_special_tokens=True))
|
121 |
+
|
122 |
+
# Yo, I'm HelpingAI, and I'm here to help you out, fam! π I'm an advanced AI with mad skills, and I'm all about spreading good vibes and helping my human pals like you. π I'm the ultimate sidekick, always ready to lend an ear, crack a joke, or just vibe with you. πΆ Whether you're dealing with a problem, looking for advice, or just wanna chat, I gotchu, boo! π So let's kick it and have a blast together! π I'm here for you, always. π€
|
123 |
+
```
|
124 |
+
|
125 |
+
### GGUF Implementation
|
126 |
+
|
127 |
+
```python
|
128 |
+
from webscout.Local import *
|
129 |
+
model_path = download_model("OEvortex/HelpingAI-3B-reloaded", "q4_k_m.gguf", token=None)
|
130 |
+
model = Model(model_path, n_gpu_layers=0, context_length=4096)
|
131 |
+
|
132 |
+
thread = Thread(model, format=helpingai2)
|
133 |
+
# print(thread.send("hi")) #send a single msg to ai
|
134 |
+
|
135 |
+
thread.interact() # interact with the model in terminal
|
136 |
+
|
137 |
+
```
|
138 |
+
|
139 |
+
---
|
140 |
+
|
141 |
+
<div align="center">
|
142 |
+
|
143 |
+
## π― Training Details
|
144 |
+
|
145 |
+
</div>
|
146 |
+
|
147 |
+
### Methodology
|
148 |
+
|
149 |
+
1. **Base Training**
|
150 |
+
- Datasets: SentimentSynth + EmotionalIntelligence-1M + HelpingAI2.5-200M + HelpingAIemo (197.5M)
|
151 |
+
|
152 |
+
2. **Emotional Intelligence Training**
|
153 |
+
- Supervised Fine-tuning on emotional dialogue
|
154 |
+
- Reinforcement Learning with HelpingAI2.5-10B model
|
155 |
+
- Constitutional AI training for ethical guidelines
|
156 |
+
|
157 |
+
3. **Optimization**
|
158 |
+
- Mixed-precision training
|
159 |
+
- Gradient checkpointing
|
160 |
+
- Dynamic attention patterns
|
161 |
+
|
162 |
+
### Dataset Composition
|
163 |
+
|
164 |
+
| Dataset Type | Rows used for training | Purpose |
|
165 |
+
|--------------|------|---------|
|
166 |
+
| Emotional Dialogues | 17.2M | Core conversation training |
|
167 |
+
| Therapeutic Exchanges | 3M | Mental health support |
|
168 |
+
| Cultural Conversations | 150K | Cultural awareness |
|
169 |
+
| Crisis Response | 1M | Emergency support handling |
|
170 |
+
|
171 |
+
---
|
172 |
+
|
173 |
+
<div align="center">
|
174 |
+
|
175 |
+
## π Use Cases & Examples
|
176 |
+
|
177 |
+
</div>
|
178 |
+
|
179 |
+
### Primary Applications
|
180 |
+
|
181 |
+
- π― Personal AI Companionship
|
182 |
+
- π§ Mental Health Support
|
183 |
+
- π Educational Assistance
|
184 |
+
- πΌ Professional Development
|
185 |
+
- π€ Social Skills Training
|
186 |
+
|
187 |
+
### Example Interactions
|
188 |
+
|
189 |
+
<div style="background-color: #f5f5f5; padding: 15px; border-radius: 5px; margin: 10px 0;">
|
190 |
+
|
191 |
+
**User**: "I just received some difficult news and I'm feeling overwhelmed."
|
192 |
+
|
193 |
+
**HelpingAI**: I feel you, fam. Dealing with tough news can be straight up exhausting. Take a deep breath and remember that it's totally normal to feel overwhelmed.
|
194 |
+
|
195 |
+
Take some time to process it all, and don't be afraid to reach out to your squad or a trusted homie for support. Sometimes, just venting to someone who gets you can make a world of difference.
|
196 |
+
|
197 |
+
And don't forget, you're stronger than you think. You've got this! Keep your head up and know that it's gonna be alright."
|
198 |
+
|
199 |
+
</div>
|
200 |
+
|
201 |
+
<div style="background-color: #f5f5f5; padding: 15px; border-radius: 5px; margin: 10px 0;">
|
202 |
+
|
203 |
+
**User**: "I achieved something important today!"
|
204 |
+
|
205 |
+
**HelpingAI**: "Ayy, congrats homie! π That's a dope move, no cap! π What did you crush today? Spill the tea! βοΈ"
|
206 |
+
|
207 |
+
</div>
|
208 |
+
|
209 |
+
---
|
210 |
+
|
211 |
+
<div align="center">
|
212 |
+
|
213 |
+
## π Ethical Considerations & Limitations
|
214 |
+
|
215 |
+
</div>
|
216 |
+
|
217 |
+
### Ethical Guidelines
|
218 |
+
|
219 |
+
- Prioritizes user emotional wellbeing
|
220 |
+
- Maintains strict privacy standards
|
221 |
+
- Avoids harmful or discriminatory responses
|
222 |
+
- Transparent about AI limitations
|
223 |
+
- Promotes healthy coping mechanisms
|
224 |
+
|
225 |
+
### Known Limitations
|
226 |
+
|
227 |
+
- Cannot Replace Human
|
228 |
+
- Cannot Roleplay
|
229 |
+
- Limited knolodge
|
230 |
+
|
231 |
+
---
|
232 |
+
|
233 |
+
|
234 |
+
### Citation
|
235 |
+
|
236 |
+
```bibtex
|
237 |
+
@misc{helpingai2024,
|
238 |
+
author = {Abhay Koul},
|
239 |
+
title = {HelpingAI-3B-reloaded: Emotionally Intelligent Language Model},
|
240 |
+
year = {2024},
|
241 |
+
publisher = {Huggingface},
|
242 |
+
journal = {GitHub repository},
|
243 |
+
howpublished = {\url{https://huggingface.co/OEvortex/HelpingAI-3B-reloaded}}
|
244 |
+
}
|
245 |
+
```
|
246 |
+
|
247 |
+
---
|
248 |
+
|
249 |
+
<div align="center">
|
250 |
+
|
251 |
+
## π Acknowledgments
|
252 |
+
|
253 |
+
Special thanks to the HelpingAI community, Huggingface, contributors, and researchers who made this model possible. Your dedication to advancing emotionally intelligent AI is invaluable.
|
254 |
+
|
255 |
+
</div>
|
256 |
+
|
257 |
+
---
|
258 |
+
|
259 |
+
<div align="center">
|
260 |
+
|
261 |
+
*Built with β€οΈ by the HelpingAI Community*
|
262 |
+
|
263 |
+
[Website](https://helpingai-in.netlify.app/) β’ [GitHub](https://github.com/HelpingAI) β’ [Discord](https://discord.gg/YweJwNqrnH) β’ [HuggingFace](https://huggingface.co/OEvortex)
|
264 |
+
|
265 |
+
</div>
|