models/LLM/Qwen3-VL-2B-Instruct-FP8/generation_config.json

{
  "bos_token_id": 151643,
  "pad_token_id": 151643,
  "do_sample": true,
  "eos_token_id": [
    151645,
    151643
  ],
  "top_p": 0.8,
  "top_k": 20,
  "temperature": 0.7,
  "repetition_penalty": 1.0,
  "transformers_version": "4.56.0"
}