231 lines
7.5 KiB
Markdown
231 lines
7.5 KiB
Markdown
---
|
|
license: apache-2.0
|
|
base_model: Qwen/Qwen3-1.7B
|
|
tags:
|
|
- qwen3
|
|
- fine-tuned
|
|
- hito
|
|
- hitonet
|
|
- reasoning
|
|
- conversational
|
|
- thinking
|
|
- self-aware-ai
|
|
- anti-hallucination
|
|
- synthetic-data
|
|
- gguf
|
|
- llama-cpp
|
|
- ollama
|
|
pipeline_tag: text-generation
|
|
language:
|
|
- en
|
|
library_name: transformers
|
|
---
|
|
|
|
<div align="center">
|
|
|
|
<img src="https://hitonet.com/assets/img/hitonet-logo-dark.png" alt="Hitonet" height="60"/>
|
|
|
|
<img src="meet_hito.png" alt="Meet Hito" width="400"/>
|
|
|
|
# Hito 1.7B
|
|
|
|
### Brain, Heart, and a Really Good Memory
|
|
|
|
[](https://huggingface.co/hitonet/hito-1.7b-GGUF)
|
|
|
|
[](https://hitonet.com)
|
|
[](https://chat.hitonet.com)
|
|
[](https://platform.hitonet.com)
|
|
[](https://platform.hitonet.com/pricing)
|
|
|
|
---
|
|
|
|
<img src="https://img.shields.io/badge/Status-Production-22c55e?style=flat-square" alt="Status"/>
|
|
<img src="https://img.shields.io/badge/Parameters-1.7B-blue?style=flat-square" alt="Parameters"/>
|
|
<img src="https://img.shields.io/badge/Context-32K-green?style=flat-square" alt="Context"/>
|
|
<img src="https://img.shields.io/badge/License-Apache_2.0-red?style=flat-square" alt="License"/>
|
|
|
|
<img src="https://img.shields.io/badge/Model_Weights-Apache_2.0_(Open)-green?style=flat-square" alt="Model License"/>
|
|
<img src="https://img.shields.io/badge/Training_Method-Commercial_License_Required-red?style=flat-square" alt="Method License"/>
|
|
|
|
</div>
|
|
|
|
---
|
|
|
|
## Cognitive Bias Resistance
|
|
|
|
Hito is specifically trained to resist cognitive biases that trip up most AI models and humans alike.
|
|
|
|
### The Bat and Ball Test
|
|
|
|
> *"A bat and a ball cost $1.10 together. The bat costs $1.00 more than the ball. How much does the ball cost?"*
|
|
|
|
Most people (and AI models) instinctively say **10 cents**. That's wrong.
|
|
|
|
| Model | Parameters | Answer | Correct |
|
|
|-------|------------|--------|---------|
|
|
| **Hito 1.7B** | **1.7B** | **$0.05** | ✅ |
|
|
| llama3.1 | 8B | $0.10 | ❌ |
|
|
| deepseek-r1 | 7B | $0.10 | ❌ |
|
|
| deepseek-r1 | 32B | $0.10 | ❌ |
|
|
| mistral | 7B | $0.10 | ❌ |
|
|
| tinyllama | 1.1B | $0.10 | ❌ |
|
|
| llama3.2 | 1B | $0.10 | ❌ |
|
|
|
|
**Hito's reasoning:**
|
|
```
|
|
<think>
|
|
Ball + Bat = $1.10, Bat = Ball + $1.00
|
|
Intuition says 10 cents... but let me verify.
|
|
If ball = $0.10, bat = $1.10, total = $1.20. WRONG.
|
|
Let ball = x: x + (x + 1) = 1.10, 2x = 0.10, x = 0.05
|
|
Ball $0.05 + Bat $1.05 = $1.10 ✓
|
|
</think>
|
|
The ball costs five cents.
|
|
```
|
|
|
|
---
|
|
|
|
## Benchmark Results
|
|
|
|
Tested against public Ollama endpoints with identical prompts:
|
|
|
|
| Model | Params | Counting | Math | Reasoning | Cognitive Bias | Overall |
|
|
|-------|--------|----------|------|-----------|----------------|---------|
|
|
| **Hito 1.7B** | **1.7B** | **100%** | **100%** | **100%** | ✅ **Resistant** | **100%** |
|
|
| llama3.1 | 8B | 100% | 67% | 100% | ❌ Fails | 89% |
|
|
| deepseek-r1:7b | 7B | 100% | 67% | 100% | ❌ Fails | 89% |
|
|
| deepseek-r1:32b | 32B | 100% | 67% | 100% | ❌ Fails | 89% |
|
|
| mistral | 7B | 33% | 67% | 100% | ❌ Fails | 67% |
|
|
| llama3.2 | 1B | 0% | 67% | 67% | ❌ Fails | 44% |
|
|
| tinyllama | 1.1B | 0% | 33% | 33% | ❌ Fails | 33% |
|
|
|
|
> **Note:** Cognitive Bias test uses the bat-and-ball problem. Models marked "Fails" gave the intuitive wrong answer ($0.10) instead of the correct answer ($0.05).
|
|
|
|
<details open>
|
|
<summary><b>Visual Benchmarks</b></summary>
|
|
|
|
<img src="benchmarks/size_vs_performance.png" alt="Size vs Performance" width="600"/>
|
|
|
|
<img src="benchmarks/counting_comparison.png" alt="Counting Comparison" width="600"/>
|
|
|
|
<img src="benchmarks/strawberry_example.png" alt="Strawberry Example" width="600"/>
|
|
|
|
</details>
|
|
|
|
---
|
|
|
|
## What Makes Hito Different
|
|
|
|
### 1. Cognitive Bias Resistance
|
|
While larger models fall for intuitive traps, Hito is trained to **stop and verify** before answering.
|
|
|
|
### 2. Structured Thinking
|
|
Uses `<think>` tags for transparent, traceable reasoning.
|
|
|
|
### 3. Self-Aware Identity
|
|
Hito knows who it is, who made it, and its purpose. No generic "I'm an AI assistant" responses.
|
|
|
|
### 4. Humble by Design
|
|
Admits uncertainty rather than hallucinating answers.
|
|
|
|
---
|
|
|
|
## Quick Start
|
|
|
|
### Python (Transformers)
|
|
|
|
```python
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
|
|
model = AutoModelForCausalLM.from_pretrained("hitonet/hito-1.7b", torch_dtype="auto", device_map="auto")
|
|
tokenizer = AutoTokenizer.from_pretrained("hitonet/hito-1.7b")
|
|
|
|
messages = [
|
|
{"role": "system", "content": "You are Hito by Hitonet.com."},
|
|
{"role": "user", "content": "A bat and ball cost $1.10. The bat costs $1 more than the ball. How much is the ball?"}
|
|
]
|
|
|
|
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt", add_generation_prompt=True).to(model.device)
|
|
outputs = model.generate(inputs, max_new_tokens=512, temperature=0.7, do_sample=True)
|
|
print(tokenizer.decode(outputs[0], skip_special_tokens=False))
|
|
```
|
|
|
|
### Ollama
|
|
|
|
```bash
|
|
# Download GGUF from hitonet/hito-1.7b-GGUF
|
|
ollama create hito -f Modelfile
|
|
ollama run hito
|
|
```
|
|
|
|
### API
|
|
|
|
```bash
|
|
curl https://api.hitonet.com/v1/chat/completions \
|
|
-H "Authorization: Bearer YOUR_API_KEY" \
|
|
-H "Content-Type: application/json" \
|
|
-d '{"model": "hito", "messages": [{"role": "user", "content": "Hello!"}]}'
|
|
```
|
|
|
|
Try the full API at [platform.hitonet.com](https://platform.hitonet.com) - $1 free credit included.
|
|
|
|
---
|
|
|
|
## Model Variants
|
|
|
|
| Repository | Format | Use Case |
|
|
|------------|--------|----------|
|
|
| [hitonet/hito-1.7b](https://huggingface.co/hitonet/hito-1.7b) | Safetensors | Python/Transformers |
|
|
| [hitonet/hito-1.7b-GGUF](https://huggingface.co/hitonet/hito-1.7b-GGUF) | GGUF | Ollama/llama.cpp/LM Studio |
|
|
|
|
### Recommended GGUF Quantizations
|
|
|
|
| Quantization | Size | Quality | Use Case |
|
|
|--------------|------|---------|----------|
|
|
| Q4_K_M | 1.1 GB | Best Balance | Most users |
|
|
| Q5_K_M | 1.2 GB | Excellent | Quality-focused |
|
|
| Q8_0 | 1.8 GB | Highest | Maximum quality |
|
|
|
|
---
|
|
|
|
## Licensing
|
|
|
|
| Component | License | Commercial Use |
|
|
|-----------|---------|----------------|
|
|
| **Model Weights** | Apache 2.0 | ✅ Free to use |
|
|
| **Training Methodology** | Proprietary | ⚠️ **Commercial License Required** |
|
|
|
|
### Model Weights (Apache 2.0)
|
|
The model weights are open source under Apache 2.0. You may use, modify, and distribute them freely.
|
|
|
|
### Training Methodology (Commercial License Required)
|
|
The training methodology, cognitive framework, and structured thinking approach used to create this model are proprietary to Hitonet.
|
|
|
|
**Commercial use of the training methodology requires a license.**
|
|
|
|
This includes but is not limited to:
|
|
- Replicating the training approach to create similar models
|
|
- Using the methodology in commercial products or services
|
|
- Derivative works based on the training techniques
|
|
|
|
**Attribution is mandatory** when using this model or discussing its capabilities.
|
|
|
|
For commercial licensing inquiries: **legal@hitonet.com**
|
|
|
|
---
|
|
|
|
## Links
|
|
|
|
- **Website:** [hitonet.com](https://hitonet.com)
|
|
- **Chat:** [chat.hitonet.com](https://chat.hitonet.com)
|
|
- **API:** [platform.hitonet.com](https://platform.hitonet.com)
|
|
- **Blog:** [hitonet.com/blog](https://hitonet.com/blog)
|
|
|
|
---
|
|
|
|
<div align="center">
|
|
<b>Made with genuine curiosity by Hitonet</b><br>
|
|
<i>Teaching AI to think, doubt, and learn.</i>
|
|
</div>
|