96 lines
2.0 KiB
Markdown
96 lines
2.0 KiB
Markdown
|
---
|
||
|
|
license: apache-2.0
|
||
|
|
language:
|
||
|
|
- en
|
||
|
|
library_name: transformers
|
||
|
|
pipeline_tag: text-generation
|
||
|
|
tags:
|
||
|
|
- chat
|
||
|
|
- instruct
|
||
|
|
- conversational
|
||
|
|
- fine-tuned
|
||
|
|
- leechanrx
|
||
|
|
- assistant
|
||
|
|
---
|
||
|
|
|
||
|
|
# 🧠 LeeChan-3B-Instruct
|
||
|
|
|
||
|
|
LeeChan-3B-Instruct is a conversational AI assistant model created and fine-tuned by LeeChanRX.
|
||
|
|
|
||
|
|
Built on top of Qwen2.5-3B-Instruct, this model is designed to provide natural conversations, helpful responses, coding assistance, and instruction-following behavior with a friendly and stable personality.
|
||
|
|
|
||
|
|
The model has been customized to act as “LeeChan”, an intelligent and conversational AI assistant focused on clarity, reliability, and user-friendly interaction.
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
# ✨ Features
|
||
|
|
|
||
|
|
- Conversational AI assistant
|
||
|
|
- Instruction-following optimized
|
||
|
|
- Coding and programming support
|
||
|
|
- Friendly and natural responses
|
||
|
|
- Stable chat behavior
|
||
|
|
- Fine-tuned personality alignment
|
||
|
|
- Lightweight 3B parameter architecture
|
||
|
|
- Transformers compatible
|
||
|
|
- Standalone merged model
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
# 🏗️ Base Model
|
||
|
|
|
||
|
|
This model is fine-tuned from:
|
||
|
|
|
||
|
|
Qwen/Qwen2.5-3B-Instruct
|
||
|
|
|
||
|
|
Credits and appreciation go to the original Qwen team for providing the open-source foundation model.
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
# 🚀 Usage
|
||
|
|
|
||
|
|
## Transformers
|
||
|
|
|
||
|
|
```python
|
||
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
||
|
|
|
||
|
|
model_name = "LeeChanRX/LeeChan-3B-Instruct"
|
||
|
|
|
||
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
||
|
|
|
||
|
|
model = AutoModelForCausalLM.from_pretrained(
|
||
|
|
model_name,
|
||
|
|
torch_dtype="auto",
|
||
|
|
device_map="auto"
|
||
|
|
)
|
||
|
|
|
||
|
|
messages = [
|
||
|
|
{
|
||
|
|
"role": "system",
|
||
|
|
"content": "You are LeeChan, a helpful AI assistant."
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"role": "user",
|
||
|
|
"content": "Hello"
|
||
|
|
}
|
||
|
|
]
|
||
|
|
|
||
|
|
text = tokenizer.apply_chat_template(
|
||
|
|
messages,
|
||
|
|
tokenize=False,
|
||
|
|
add_generation_prompt=True
|
||
|
|
)
|
||
|
|
|
||
|
|
inputs = tokenizer(
|
||
|
|
text,
|
||
|
|
return_tensors="pt"
|
||
|
|
).to(model.device)
|
||
|
|
|
||
|
|
outputs = model.generate(
|
||
|
|
**inputs,
|
||
|
|
max_new_tokens=128,
|
||
|
|
temperature=0.7,
|
||
|
|
repetition_penalty=1.1
|
||
|
|
)
|
||
|
|
|
||
|
|
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
|