qwen30b-codeclash-sft-6turns-s3-merged

Merged full checkpoint for CodeClash SFT.

• Base model: Qwen/Qwen3-Coder-30B-A3B-Instruct
• Fine-tuning: LoRA SFT on CodeClash trajectories (6-turns s3 dataset)
• This repo is a merged model (base + LoRA), ready for inference/deployment.

Usage (Transformers)

from transformers import AutoTokenizer, AutoModelForCausalLM

model_id = "z050209/qwen30b-codeclash-sft-6turns-s3-merged"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype="auto",
    device_map="auto",
)

Notes

• Chat template/tokenizer files are included in this repository.
• For Hugging Face Inference Endpoints, prefer vLLM container when available.