52 lines
1.4 KiB
Markdown
52 lines
1.4 KiB
Markdown
---
|
|
language:
|
|
- en
|
|
license: mit
|
|
base_model: Qwen/Qwen2.5-7B-Instruct
|
|
tags:
|
|
- math
|
|
- reasoning
|
|
- fine-tuned
|
|
- specialized-intelligence
|
|
library_name: transformers
|
|
pipeline_tag: text-generation
|
|
---
|
|
|
|
# math-llm-sit-7b
|
|
|
|
Fine-tuned math reasoning model based on [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct).
|
|
|
|
Trained using the **Specialized Intelligence Theory** 4-phase pipeline:
|
|
SFT -> Feedback -> Posterior -> Framework.
|
|
|
|
## Training Details
|
|
|
|
| Property | Value |
|
|
|----------|-------|
|
|
| Base model | `Qwen/Qwen2.5-7B-Instruct` |
|
|
| Training method | LoRA + 4-phase SIT pipeline |
|
|
| Upload date | 2026-04-02 |
|
|
|
|
## Usage
|
|
|
|
```python
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
|
|
model = AutoModelForCausalLM.from_pretrained("usersina/math-llm-sit-7b")
|
|
tokenizer = AutoTokenizer.from_pretrained("usersina/math-llm-sit-7b")
|
|
|
|
prompt = "Solve: What is the integral of x^2 from 0 to 1?"
|
|
inputs = tokenizer(prompt, return_tensors="pt")
|
|
outputs = model.generate(**inputs, max_new_tokens=512)
|
|
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
|
|
```
|
|
|
|
## Training Pipeline
|
|
|
|
This model was trained using the 4-phase Specialized Intelligence Theory pipeline:
|
|
|
|
1. **SFT (Supervised Fine-Tuning)**: Foundation math reasoning
|
|
2. **Feedback**: Weighted feedback integration
|
|
3. **Posterior**: Internal posterior calibration
|
|
4. **Framework**: Full framework integration with resource allocation
|