| question_file=examples/multiturn/question.json | |
| answer_file=examples/multiturn/answer.json | |
| model_base=/data1/speech/anhnmt2/Speech2Speech/LLaMA-Omni/models/llm/Qwen2.5-3B-Instruct | |
| model_path=/data1/speech/anhnmt2/Speech2Speech/half-streaming-speech-nlp/checkpoints/omni_whisper-medium_Qwen2.5-3B_pretrained-sft-lora | |
| prompt_version=qwen | |
| python3 infer_llm_multiturn.py \ | |
| --model-path $model_path \ | |
| --model-base $model_base \ | |
| --question-file $question_file \ | |
| --answer-file $answer_file \ | |
| --num-chunks 1 \ | |
| --chunk-idx 0 \ | |
| --temperature 0 \ | |
| --conv-mode $prompt_version \ | |
| --input_type mel \ | |
| --mel_size 80 \ | |
| --is_lora |