synth-lora-npu-test / generation_config.json
dylanneve1's picture
Synthetic LLM with LoRA adapters (12L, int4, rank=8, GQA 8/2) for NPUW NPU testing
420aeb8 verified
{
"bos_token_id": 1,
"eos_token_id": 2,
"pad_token_id": 0,
"max_length": 4096
}