A lightweight instruction-tuned version of Qwen3 Tiny trained on real-world terminal and tool-using coding conversations exported with DataClaw.
It is small, fast, and suitable for local deployment.
This model is a supervised fine-tune (SFT) of a Qwen3 Tiny base model on structured coding-agent conversations.
The training format preserves conversational context using a rolling window with anchored task instructions to improve task persistence.
This is a small model trained on narrow agentic data. It is not aligned for broad real-world deployment.
1from transformers import AutoTokenizer, AutoModelForCausalLM
2import torch
3
4model_id = "your-username/your-model-name"
5
6tokenizer = AutoTokenizer.from_pretrained(model_id)
7model = AutoModelForCausalLM.from_pretrained(
8 model_id,
9 torch_dtype=torch.bfloat16,
10 device_map="auto"
11)
12
13prompt = "List what directories and files are here. Just ls, no explanation needed."
14
15inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
16output = model.generate(**inputs, max_new_tokens=200, temperature=0.6)
17
18print(tokenizer.decode(output[0], skip_special_tokens=True))
