Back to projectsPERSONAL PROJECT / IN PROGRESS

APPLIED AI / PERSONAL INTELLIGENCE

Lucian AI.

From local conversations to a voice of my own.

A personal chatbot built on official Qwen3-1.7B weights, with my own GPU inference implementation. First, make it conversational. Then, explore my own wording and expression.

Base model01
Qwen3–1.7B

Official public weights · Apache-2.0

Inference engine02
Python + PTX

Custom tokenizer and GPU forward pass

Default context03
2,048 token

Multi-turn history · Streaming output

Personal voice04
Exploring

No personal style fine-tuning yet

01 / CONVERSATION

Talk to Lucian AI

PLAYGROUND
Lucian AI

UI preview · Authored examples

Authored design examples. No model is called.Enter ↵ / Shift + Enter

02 / ENGINEERING

From weights to a conversation.

The official model provides the foundation. My implementation connects each layer.

01 / TOKENIZATIONImplemented

Text and tokenisation

Custom byte BPE, pre-tokenisation and role templates turn text into tokens for inference.

Byte BPEUnicodePython
02 / GPU INFERENCEImplemented

Local inference

Read official weights and run a custom PTX GPU forward pass with causal attention and a KV cache.

PTXAttentionKV Cache
03 / CONVERSATIONImplemented

Continuous conversation

Stream generated text, manage multi-turn history, and report context trimming and reply limits.

StreamingContextEvaluation

03 / PERSONAL VOICE

Give expression a personal source.

Exploring / Personal fine-tuning not yet implemented

I hope it can answer questions, and gradually reflect how I express myself.

Next, curate my own writing and ideas. Explore how personal text shapes wording, explanations and rhythm, and use independent evaluation to guide each iteration.

My writingStyle adaptationEvaluation
  1. 01

    Make it conversational

    Current stage

    Implement local Qwen inference and multi-turn dialogue, and inspect issues in real answers.

  2. 02

    Build a personal corpus

    Planned

    Curate my own text, preserve sources, and separate training candidates from evaluation samples.

  3. 03

    Explore personal style adaptation

    Research direction

    Compare expression and answer quality, then use the findings to guide the next iteration.

04 / PROJECT NOTES

Record the work and its limits.

Explain what works, what comes next, and where the model’s capabilities end.

Explore my other projects
What can I try now?

The local model supports text and multi-turn Chinese conversation. This page starts with clearly labelled, authored UI examples. Start the local model service to switch to actual Qwen replies.

Has it learned my personal style?

No personal style fine-tuning has been implemented yet. Corpus curation and style adaptation are future plans. Example copy is not used as my personal expression corpus.

Where do the model and data come from?

The base model is official Qwen3-1.7B public weights under Apache-2.0. Inference code is implemented in this project. Personal materials and experiment checkpoints stay local.

What are the capability limits?

The model handles text, without web search or actions. Answers may be wrong. The default context is 2048 tokens; old turns are removed with a notice when needed, and reply-length limits are reported.

QWEN FOUNDATION / INDEPENDENT IMPLEMENTATIONBack to conversation ↑

LOCAL MODEL

The local model is not connected.

You can still explore the page and example conversations. Connect the local Qwen service for actual model-generated replies.

How to run locally

Run this in the project’s qwen/showcase directory:

python -X utf8 server.py --with-model --port 8767

Open the local URL printed in the terminal. Existing official weights and an NVIDIA GPU are required.