RWKV is a RNN with transformer-level LLM performance. It can be directly trained like a GPT (parallelizable). So it's combining the best of RNN and transformer - great performance, fast inference, saves VRAM, fast training, "infinite" ctx_len, and free sentence embedding.
Apache License 2.0
405
stars
41
forks
source link
【feature advice】Int8 mode to run original model #15
when using LoRA to fine-tune, the original model parameters are locked, which can be converted to INT8 mode to inference, and other trainable parameters with fp16/bf16/fp32/tf32 mode. It is just like what peft library do.
Uing INT8 to run original model will save GPU memory and accelerate training speed.
when using LoRA to fine-tune, the original model parameters are locked, which can be converted to INT8 mode to inference, and other trainable parameters with fp16/bf16/fp32/tf32 mode. It is just like what peft library do.
Uing INT8 to run original model will save GPU memory and accelerate training speed.