nhonhoccode/qwen3-0-6b-cybersecqa-sft-freeze2-20251031-2359

qwen3-0-6b — Cybersecurity QA (SFT)

Fine-tuned on Kaggle (2×T4) using SFT.

Base

• unsloth/Qwen3-0.6B

Validation (greedy)

Metric	Score
BLEU-4	1.07
ROUGE-L	14.62
F1	10.70
EM	0.00
Train Time (s)	20385.9

How to use

from transformers import AutoTokenizer, AutoModelForCausalLM
tok = AutoTokenizer.from_pretrained("nhonhoccode/qwen3-0-6b-cybersecqa-sft-freeze2-20251031-2359")
mdl = AutoModelForCausalLM.from_pretrained("nhonhoccode/qwen3-0-6b-cybersecqa-sft-freeze2-20251031-2359")
prompt = tok.apply_chat_template([{"role":"system","content":"You are a helpful assistant."},
                                  {"role":"user","content":"Explain SQL injection in one paragraph."}],
                                 tokenize=False, add_generation_prompt=True)
ids = tok(prompt, return_tensors="pt").input_ids
out = mdl.generate(ids, max_new_tokens=128)
print(tok.decode(out[0][ids.shape[-1]:], skip_special_tokens=True))

Trainable

• Trainable params: 187,044,352 / 596,049,920